FIFTH EDITION

Richard A. Schmidt, PhD, Professor Emeritus

Psychology Department
University of California, Los Angeles
and
Human Performance Research
Marina del Rey, California

Timothy D. Lee, PhD

Department of Kinesiology
McMaster University, Ontario, Canada

HUMAN KINETICS
 Library of Congress Cataloging-in-Publication Data

Schmidt,RichardA., 1941-
Motor control and learning : a behavioral emphasis I Richard A. Schmidt and Timothy D. Lee. -- 5th ed.
p. ;cm.
Includes bibliographical references and indexes.
ISBN-13: 978-0-7360-7961-7 (hard cover)
ISBN-10: 0-7360-7961-0 (hard cover)
1. Movement, Psychology of. 2. Motor learning. I. Lee, Timothy Donald, 1955- II. Title.
[DNLM: 1. Motor Activity. 2. Motor Skills. 3. Learning. WE 103]
BF295.S248 2011
152.3--dc22
2010044051

ISBN-10: 0-7360-7961-0 (print)

ISBN-13: 978-0-7360-7961-7 (print)

Copyright© 2011, 2005, 1999 by Richard A. Schmidt and Timothy D. Lee

Copyright © 1988, 1982 by Richard A. Schmidt

All rights reserved. Except for use in a review, the reproduction or utilization of this work in any form or by any electronic, mechani-
cal, or other means, now known or hereafter invented, including xerography, photocopying, and recording, and in any information
storage and retrieval system, is forbidden without the written permission of the publisher.

The Web addresses cited in this text were current as of August 2010, unless otherwise noted.

Acquisitions Editor: Myles Schrag; Managing Editor: Melissa J. Zavala; Assistant Editors: Antoinette Pomata and Steve Calde-
rwood; Copyeditor: Joyce Sexton; Indexer: Andrea J. Hepner; Permission Manager: Dalene Reeder; Graphic Designer: Bob
Reuther; Graphic Artist: Denise Lowry; Cover Designer: Keith Blomberg; Photographer (cover): ©Yuri Kadobnov/AFP/Getty
Images; Photographer (interior): ©Human Kinetics, unless otherwise noted; Photo Production Manager: Jason Allen; Art
Manager: Kelly Hendren; Associate Art Manager: Alan L. Wilborn; Art Style Development: Joanne Brummett; Illustrations:
© Human Kinetics; Printer: Sheridan Books

Printed in the United States of America 10 9 8 7 6 5 4 3 2

The paper in this book is certified under a sustainable forestry program.

Human Kinetics
Web site: www.HumanKinetics.com

United States: Human Kinetics, P.O. Box 5076, Champaign, IL 61825-5076

800-747-4457
e-mail: humank@hkusa.com

Canada: Human Kinetics, 475 Devonshire Road Unit 100, Windsor, ON N8Y 2L5
800-465-7301 (in Canada only)
e-mail: info@hkcanada.com

Europe: Human Kinetics, 107 Bradford Road, Stanningley, Leeds LS28 6AT, United Kingdom
+44 (0) 113 255 5665
e-mail: hk@hkeurope.com

Australia: Human Kinetics, 57 A Price Avenue, Lower Mitcham, South Australia 5062
08 8372 0999
e-mail: info@hkaustralia.com

New Zealand: Human Kinetics, P.O. Box 80, Torrens Park, South Australia 5062
0800 222 062
e-mail: info@hknewzealand.com

E4689
 <+j!~';i~c-'.>':/ /::<;;~;;;->-- '<'~-,.,,- _,---, -'
'>>:.:;::''s/:>;;-,:,;-/;{;:C:-.c'>":-"'->~ . ---' ,
-"" :;:,o_; Y'i",. \&:i·C:>"-"-, CONTENTS

Preface vii Acknowledgments ix

PART I INTRODUCTION TO MOTOR BEHAVIOR 1

CHAPTER l Evolution of a Field of Study . .......................... .3

Understanding Movement . . 4
Origins of the Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

CHAPTER 2 Methodology for Studying Motor Performance ......... .21

Classification of Motor Skills . . . . . . . 21
Basic Considerations in Measurement. . .24
Measuring Motor Behavior. . . . . . . . .26
Measuring and Evaluating Relationships .48
Summary . . . . . . . . . . . . . . . . . .54

CHAPTER 3 Human Information Processing ....................... .57

The Information-Processing Model . . . .57
Three Stages of Information Processing. .58
Anticipation . . . . . . . . 78
Signal-Detection Theory . . 85
Memo~ . . . . . . . . . .88
Summary . . . . . . . . . .93

CHAPTER 4 Attention and Performance ......................... .97

Types of Attention . . . . . .97
Theories of Attention . . . . 100
Competition for Attention. . 103
Attention During Movement 116
Focus of Attention . . 124
Attention and Anxiety 127
Summary . . . . . . . 130

iii
iv Contents

PART II MOTOR CONTROL 133

CHAPTER 5 Sensory Contributions to Motor Control. . . . . . . . . . . . . . . . . . 135

Closed-Loop Control Systems . 135
Vision . . . . . . 136
Audition . . . . . . . . . . . . 153
Proprioceptors . . . . . . . . . 153
Proprioception and Motor Control . 158
Feedforward Influences on Motor Control . 170
Summary . . . . . . . . . . . . . . . . . . 174

CHAPTER 6 Central Contributions to Motor Control . ................. 177

Open-Loop Processes . . . . . . . . 177
Central Control Mechanisms. . . . . 182
Central Control of Rapid Movements 188
Generalized Motor Programs . . . . 208
Summary . . . . . . . . . . . . . . . 221

CHAPTER 7 Principles of Speed and Accuracy ..................... 223

Fitts' Law: The Logarithmic Speed-Accuracy Trade-Off. . . . . 224
The Linear Speed-Accuracy Trade-Off . . . . . . . . . . . . . 231
The Temporal Speed-Accuracy Trade-Off. . . . . . . . . . . . 234
Central Contributions to the Spatial Speed-Accuracy Trade-Off 241
Correction Models of the Speed-Accuracy Trade-Off. . . . . . 257
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260

CHAPTER 8 Coordination . ....................................... 263

Discrete Tasks . . . 264
Continuous Tasks . . . . . . . . . . . . . . . 277
Summary . . . . . . . . . . . . . . . . . . . . 294

CHAPTER 9 Individual Differences and Capabilities .... _............. 297

Experimental Versus Differential Approaches . . 297
Abilities . . . 301
Taxonomies. . . . . . . . . . . 310
Prediction . . . . . . . . . . . 312
Individual-Difference Variables 315
Summary . . . . . . . . . . . . . 324

PART III MOTOR LEARNING 325

CHAPTER l 0 Motor Learning Concepts and Research Methods . ...... 327

Motor Learning Defined . . 327
Measuring Motor Learning . . . . . . . . . . . . . . . . . . . . . . . 329
 Contents v

Designing Experiments on Learning . . . . . 335

Alternative Methods for Measuring Learning 338
Issues About the "Amount" of Learning . . . 342
Understanding Learning and Performance Variables . 345
Summary . . . . . . . . . . . . . . . . . . . . . . . 345

CHAPTER 11 Conditions of Practice ............................... 347

The Law of Practice . . 347
Motivation for Learning. 351
Verbal Information . . . 353
Perceptual Learning .. 355
Observational Learning 356
Mental Practice. . . . . 359
Distribution of Practice. 361
Variability of Practice .. 367
Contextual Interference 371
Part Versus Whole Practice 382
Guidance . . . . . . . . . . 386
Principles of Practice Specificity. 388
Summary . . . . . . . . . . . . . 390

CHAPTER 12 Augmented Feedback .............................. 393

Classifications and Definitions . . . . . . . . . . 393
Research on Augmented Feedback. . . . . . . . 395
Evaluating the Effects of Augmented Feedback. . 398
Knowledge of Performance . . . . . . . . . . . . 400
Knowledge of Results . . . . . . . . . . . . . . 405
Theoretical Issues: How Does Augmented Feedback "Work"? . 424
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 427

CHAPTER 13 The Learning Process ................................ 429

Characteristics of the Learning Process. 429
Closed-Loop Theory . . . . . . . . . . . . . . . . . . 438
Schema Theory . . . . . . . . . . . . . . . . . . . 441
Differing Theoretical Perspectives of Motor Learning . 446
Summary . . . . . . . . . . . . . . . . . . . . . . . . 456

CHAPTER 14 Retention and Transfer............................... 461

Fundamental Distinctions and Definitions . 461
Measuring Retention and Transfer . 463
Retention and Motor Memory 467
Retention Loss . . . 471
Transfer of Learning 481
Summary . . . . . . 488

Appendix 491 Glossary 493 References 503 Author Index 565

Subject Index 573 About the Authors 581
 PREFACE

ost of us have marveled at one time detection of errors, spinal cord processes that are
or another about how the most highly capable of generating patterns of skilled activi-
skilled performers in industry, sport, music, ties in locomotion, and various biomechanical
or dance seem to make their actions appear so factors that act to determine the nature of our
simple and easy-performed with incredible effi- movement behaviors are all important if we are
ciency, smoothness, style, and grace. Like the first to understand highly skilled performance. This
four editions of this text (Schmidt, 1982, 1988; blending of behavioral, neurophysiological, and
Schmidt & Lee, 1999, 2005), the fifth edition of biomechanical analyses reflects the fact that the
Motor Control and Learning: A Behavioral Emphasis fields of motor behavior and motor learning,
was written for those who would like to under- movement neurophysiology (or motor control),
stand how it is that these performers can achieve and biomechanics are rapidly moving together
such artistry while we, as beginners in a similar toward the shared understanding of complex
task, are clumsy, inept, and unskilled. This book movement behaviors.
was written particularly as a textbook for uni- This edition of the text retains the same goal
versity or college undergraduate and graduate of presenting an up-to-date review of the state of
students taking courses in human performance knowledge in movement control and learning,
or motor learning, primarily in fields such as and it does so with a format similar to that of
kinesiology or psychology. However, students in the previous editions. We have directed consid-
other fields, such as the neurosciences, physical erable effort toward including the most recent
therapy, occupational therapy, speech therapy, knowledge from a number of rapidly develop-
biomedical or industrial engineering, human ing subfields, and each chapter has been revised
factors or ergonomics, and sport, will also find extensively in light of these newer concepts. In
many of the concepts contained here to be of addition to including more than 280 new refer-
interest, as movement behavior is a part of all of ences to work published since the last edition (in
them. And for those who are, or who are becom- 2005), we have also endeavored to pay homage
ing, practitioners in these fields, the principles of to some of the importan~ early research develop-
motor behavior outlined here should provide a ments in the various areas. One of the features
solid basis for tasks such as designing human- introduced in the previous two editions was
machine systems, developing training programs sidebars highlighting specific research issues
in sport or industry, or teaching progressions in throughout the book. We have created more of
dance or music. these sidebars in this fifth
The emphasis of the text is behavioral. That edition with the idea that
is, the primary focus is on movement behavior certain material requires
that can be observed directly, as well as on the a more detailed treatment
many factors that affect the quality of these per- to ensure understanding
formances and the ease with which they can be than is typical of passages HumanKinetics.com
learned. In this sense, the book has strong ties in the text. Some of these
to the methods and traditions of experimental sidebars highlight quotations from selected his-
psychology. Yet, at the same time, we focus on torical papers; other sidebars highlight careers
the neurological and biomechanical processes of key researchers in motor control and learning;
out of which these complex movement behaviors and still others deal with various applications of
are crafted. Brain mechanisms that allow the specific concepts in motor control and learning.

vii
viii mi Preface

Some chapters from the previous edition have Part II deals with motor control. Chapter 5
been reduced in order to lessen complexities in views motor control from a closed-loop per-
the text without sacrificing the in-depth cover- spective, in which the sensory contributions to
age or richness of the concepts. And we have movement control are examined, with particular
expanded other chapters and sections to present emphasis on new research regarding the role of
new, exciting areas of research that have emerged vision. In chapter 6, the focus shifts to contribu-
since the previous edition. Many new figures tions of the central nervous system to movement
have also been included to help illustrate and control, with emphasis on preorganized actions
emphasize concepts and data that are difficult that involve motor programs and generalized
to communicate effectively in words. Practical motor programs. Some principles related to
examples from areas such as human factors, sport, speed and accuracy are presented in chapter 7,
therapy, and music are provided to illustrate together with a discussion of theoretical concepts
motor control and learning concepts and provide that integrate the central and sensory contribu-
suggestions for application. As before, the fifth tions to movement control. Chapter 8 presents a
edition reflects a logical progression so that later discussion of the factors involved in movement
chapters build upon concepts presented in earlier control that make coordination both easy and
chapters, with the final result being a consistent, difficult to achieve. The final chapter in this part
defensible framework of ideas about motor skills. contains a discussion of factors that determine
Having such a framework, or point of view, is skill differences between people and among
important for those who wish to use the informa- groups of people, with emphasis on important
tion presented here, both so that contributions to themes about abilities and the prediction of skills.
new applications may be made and so that the Part III deals with the acquisition of skill, or
design of new skills research is facilitated. motor learning. Chapter 10 concentrates on some
The book is divided into three parts. Part I pro- special methodological problems for studying
vides an introduction to research and fundamen- learning. The effects of practice, the structure
tal concepts that are important to understanding of the practice session, and the many variables
motor behavior. The first chapter, a brief history under the control of a teacher, coach, or therapist
of the field, is followed by a presentation of meth- are discussed in chapter 11, while feedback con-
ods in movement research in chapter 2, focusing tributions to learning are included in chapter 12.
on various paradigms and statistical techniques In both of these chapters, much new information
used in the study of movement behavior. In is covered that demands important changes in
chapter 3 the human is regarded as a processor our understanding of the processes involved in
of information, and we focus on the many ways practice and the ways in which these influence
that information is dealt with in motor behavior. learning. Chapter 13 presents various theoretical
The concept of attention is the focus of chapter 4, treatments of motor learning. And finally, chap-
with particular emphasis on the role of attention ter 14 deals with the factors associated with the
in motor behavior. retention and transfer of skills.
 hroughout the long process of this revision, Thanks also go to our editors, Melissa Zavala
a number of people provided very highly and Antoinette Pomata, for their excellent work
valued input. Judy Wright of Human Kinetics on the project. We called upon a number of col-
provided encouragement and many valuable leagues to read specific sections of the book. The
suggestions for not only this revision, but on many suggestions made by Ramesh Balasubra-
previous editions of the book as well. Her recent maniam, Ian Franks, Waclaw Petrynski, Gwen
retirement closes an important chapter not only Gordon, and Laurie Wishart were invaluable in
for the publisher, but also for the expression making these revisions. The final manuscript is
of ideas presented in this book. Judy, we wish much better as a result, and we are grateful for
you an energetic and successful next dance. their time and effort.

ix
CHAPTER 1 Evolution of a Field of Study
CHAPTER 2 Methodology for Studying Motor Performance
CHAPTER 3 Human Information Processing
CHAPTER 4 Attention and Performance

his first part introduces the field of motor control and learning. In chapter 1 the area
is described, and the important distinctions separating motor control and learning
from other, related fields of study are made. Then, a brief history of the field is given,
showing how knowledge about movements from psychology and kinesiology, as well
as from the neurosciences, has been combined. The second chapter deals with the vari-
ous scientific methods used for studying motor skills. Here, we explain the tools of
motor behavior research, focusing on the various ways in which motor behavior and
skill can be measured. Chapter 3 presents the information-processing approach, which
is fundamental to understanding how humans think and act. The last chapter in this
section describes attention and its role in motor behavior.

1
 ovement is a critical aspect of life. Without ing a typewriter, or performing a triple twist-
movement, we could not feed ourselves, ing somersault from the diving board. These
we could not reproduce, and we would not learned movements are often termed skills.
survive. Our capacity to move is more than just They are not inherited, and mastering them
a convenience that enables us to walk, play, or requires long periods of practice and experi-
manipulate objects; it is a critical aspect of our ence. Guthrie (1952) perhaps provided the best
evolutionary development, no less important definition of skills: "Skill consists in the ability
than the evolution of our intellectual and emo- to bring about some end result with maximum
tional capacities. Some assert that our highly certainty and minimum outlay of energy, or of
developed cognitive capacities evolved so that time and energy" (p. 136). Skills are especially
we could make the movements essential to critical to the study of human behavior, as they
survival-those involved in the construction of are involved in operating machines in industry,
shelter, the making of tools, and communication. controlling vehicles, preparing meals, playing
Surely the study of movement needs no further games, and so on. Skills .and genetically defined
justification than its significance in terms of the movements can range from very simple (e.g.,
evolution of humankind. snapping fingers or blinking eyes) to very com-
Movement takes many forms. Some forms plex (e.g., pole-vaulting).
can be regarded as genetically defined, such as This book is about all these kinds of move-
the way in which people control their limbs or ments, whether primarily genetically defined or
the ability of centipedes to coordinate their legs. learned through practice. In particular, we will
Other examples include the "scratch reflex" of be concerned with how these various movements
dogs or the rapid blink of the eye in response are controlled-how the central nervous system is
to an unexpected puff of air. Here, the patterns organized so that the many individual muscles
of action appear to be determined by genetic and joints become coordinated. We will also be
makeup, through growth and development, concerned with how sensory information from
or in both ways; and these actions appear to the environment, the body, or both is used in the
be quite stereotypical for members of the same control of movement. The scientific field of study
species. A second class of movements can be that addresses these issues is known as motor
thought of as "learned"-for example, those control-the study of the control of movements
involved in controlling an automobile, operat- in humans and animals.

ii 3
4 mi Motor Control and Learning

In this book, we add one important aspect to the historical and sociological effects of illnesses
motor control that is sometimes not included- on entire societies. Because these various ways
the study of how movements are learned, that is, of considering a single problem are so diverse, a
how movements are produced differently as a given scientist usually focuses on one, or at most
result of practice or experience. Indeed, much two, of these levels of analysis.
evidence suggests that many of the movements
already mentioned consist of a complex combina- A Behavioral Level of Analysis
tion of genetic determinants coupled with modi-
The focus of this text is primarily at the behavioral
fications made through practice or experience.
level of analysis, centering on cognitive, informa-
Understanding how movements are learned is
tion-processing concepts. The major goals will be
the major concern of a field of study called motor
to understand the variables that determine motor
learning. We see no good justification, however,
performance proficiency and to understand the
for separating the study of motor learning from
variables that are most important for the learn-
the study of movement or of motor control in
ing of movement behaviors. We also want to
general, as this artificial separation inhibits the
understand how such information can be used in
understanding of both issues. For these reasons,
the solution of certain practical problems such as
as the title reveals, the subject matter of the book
those involved in the design of equipment that
is motor control and learning.
humans must operate, in the selection of indi-
viduals for occupations, in the teaching of skills
Understanding Movement in sport and industry, and in the rehabilitation of
skills after injury or stroke.
This behavioral level of analysis, however, is
How can knowledge and information about
more interesting and complete when combined
movement be acquired? A logical way to proceed
with two other fields of study, each representing
would be to study some relevant aspect of the
a deeper level of analysis. The field of biomechan-
movement-control process using scientific meth-
ods. But which processes should be examined?
ics concerns the mechanical and physical bases of
biological systems. Certainly in order to under-
One possibility would be to focus on the nature of
stand movement we must understand something
biochemical interactions that occur within cells as
of the body itself, with all its joints, levers, and
individuals move. Or we could focus on the cell
associated mechanical characteristics. The field
itself, asking how cells interact with each other
of neurophysiology concerns the functioning of the
in the control of movement. In a similar way, we
brain and central nervous system and the ways
could consider groups of cells, such as a whole
in which they control the contractions of muscles
muscle, the spinal cord, or the nerves, and ask
that move the limbs. The study of movement will
how these relatively more complex structures are
be addressed at various levels of analysis-but as
involved in movement control. Another possibil-
the subtitle of the book suggests, the emphasis is
ity would be to focus on the movements of the
at the behavioral level.
freely moving animal or human, concentrating on
the factors that determine movement accuracy,
Emphasizing Movements
the choice of movement, or the patterns of action.
Along the same lines, we could study movement In considering movements, especially skills, it
in an even more global context, asking questions is often difficult to isolate a movement from its
about the role of movement in society, the choice environment. In driving a car, for example, there
of certain skilled occupations or sports, move- are the coordinated actions involved in changing
ment in groups or teams, and so on. gears (clutch, accelerator, shift lever, etc.) as well
Clearly, there are various ways to consider the as the movements involved in steering. These
same phenomenon. They involve the study of a parts of the skill are the means through which
phenomenon on different levels of analysis, and the driver affects his environment. But skills are
analogous levels are present in any area of scien- also affected by the environment. For example,
tific concern. Illnesses, for example, can be con- whether or not there are turns in the road or
sidered at levels that range from the biochemical whether snow is present influences the driver's
and neurological determinants of disease through interactions with the vehicle controls. Such recip-
 Evolution of a Field of Study 111 5

rocal relations between the environment and the such as using a lathe, typing, woodcarving, and
individual make it very difficult to pinpoint the handwriting, are of critical importance to this
various determinants of motor behavior, because field of study. Artistic performances, such as
the interaction of the many motor control and the playing of musical instruments, the creation
environmental factors is extremely complex and of a painting, or the production of a dance, are
difficult to study with experimental procedures. certainly under the heading of motor behavior
The approach taken in this text is to focus on as treated here. The use of voice, whether by the
the mutual interactions between the environment vocalist in an opera or by the student learning a
and the motor system. A large portion of this new language,1 is also a motor task, as the sounds
approach deals with the behavior and capabili- are controlled by muscular activity of the vocal
ties of the motor system to produce movements, apparatus in ways analogous to the control of
studied more or less independently of the role the hands and fingers of the skilled typist. The
of sensory or environmental information. But at potential applications for the principles discov-
the same time, the role of environmental infor- ered in the field of motor control are present in
mation such as vision, and the ways in which it nearly every aspect of our lives.
is processed and used to guide movements, is
important. In any case, we are deliberately not
concerned with skills in which the quality of the Origins of the Field
movement components per se is almost irrelevant
to the outcome (as in playing chess). In an examination of the early research on move-
In deciding which skills to include in our field ment and learning, it will be evident that the field,
of study, it is helpful to consider the probable as we know it today, emerged from two isolated
limiting factors in the performance. In the chess bodies of knowledge. These two areas are (a) the
example, intellectual decision making seems branch of neurophysiology primarily concerned
to be the important factor and should not be with the neural processes that are associated with
included in this treatment. In a marathon, or in (or are causes of) movements, with only slight
weightlifting, the factors seem to be more closely reference to the movements themselves; and (b)
related to cardiovascular fitness and strength, the branch of psychology and related fields pri-
respectively-also not within the confines of the marily concerned with high-level skills with very
present field of study. We will emphasize skills little reference to the neurological mechanisms
in which the focus is on the capabilities to use involved. For nearly a century, these two fields
environmental information in the complex con- developed knowledge at different levels of analy-
trol of the limbs. sis but with little mutual influence. Only toward
the end of the 1970s did the two fields begin to
Potential Applications come together. For the reader interested in more
detail on these historical.developments, see Irion
Given an understanding of some of the processes
(1966), Adams (1987), and Summers (1992, 2004).
underlying the control of movements, where can
these principles be applied? High-level sports,
Early Research
games, and athletic events come to mind as areas
for application, as these activities often involve A fascinating account of some of the early insights
the same kinds of processes that are studied in the regarding actions and movement appears in a
area of motor control and learning. But potential review by Meijer (2001). In this historical paper,
generalizations should not be limited to these Meijer traces the origins of a number of ideas
kinds of activities. Many apparently genetically within current thinking to philosophers such as
defined actions such as walking and maintain- Plato, Aristotle, and Galen. Some of the earliest
ing posture are under consideration here. How empirical investigations of motor skills were
these movement capabilities, when disrupted by performed around 1820 by the astronomer Bessel
injuries or disease, can be improved by treatments (cited by Welford, 1968), who tried to understand
emphasizing the learning of new movement pat- the differences among his colleagues in recording
terns-the subject matter of physical therapy-is the transit times of the movements of stars. This
also an application area. Many industrial skills, skill involved estimating the time required for the
6 Ill! Motor Control and Learning

image of a star to move through the crosshairs of in German and French publications during the
a telescope. Bessel was interested in the processes middle and late 1800s. Some of this research is
underlying this complex skill, as well as in the summarized by Worringham (1992).
reasons some of his colleagues estimated accu- A major influence of the time was Thorndike
rately and others could not. Considerably later, (1914), who was concerned with processes under-
studies addressed the visual contributions to lying the learning of skills and other behaviors.
hand movements in localizing targets (Bowditch His Law of Effect, which continues to have its
& Southard, 1882). Leuba and Chamberlain (1909) influences in psychology, states that responses
studied the accuracy of limb-positioning move- that are rewarded tend to be repeated. Responses
ments; Fullerton and Cattell (1892) examined that are not followed by a reward tend not to be
force reproducibility; Stevens (1886) studied repeated. This idea formed the cornerstone for
timing; and Judd (1908) studied transfer oflearn- much of the theorizing about learning that was to
ing with dart-throwing tasks. Some researchers follow in the 20th century (Adams, 1978). Thorn-
used experimental methods to study expertise dike was also a pioneer in the area of individual
in sport performance (Scripture, 1894; see also differences, in which the focus is on the differ-
Fuchs, 1998). An important trend was established ences among individuals surrounding practice
by Bryan and Harter's (1897, 1899) work on (see chapter 9).
receiving and sending Morse code; periods of Most of the work mentioned here originated
no improvement (plateaus) between segments of from the field of psychology, and much of the
improvement were identified, and considerable field of motor behavior today is the legacy of
debate about the existence and interpretation this early thinking and research. But the early
of these plateaus continued for some decades research, which is similar in method to at least
(e.g., Book, 1908/1925; Keller, 1958). Galton (see some of today's work, marked a severe break in
Boring, 1950) studied the relationships among tradition from the pre-1900 views of behavior.
strength, steadiness, and body configuration The pre-1900 research often involved introspec-
in over 9,000 British males and females; Book tion, including subjective self-reports of feelings
(1908I1925) examined typing skills for very large that were unobservable. Skills were studied only
samples of subjects ranging widely in ability and because they were thought to provide "access
age. Retention of skills over long intervals of no to the mind." As the 19th century ended, there
practice was an important theme, and typing was was a shift to more systematic and objective
a convenient way to study it (e.g., Bean, 1912; approaches to the study of skills. And, of equal
Swift & Schuyler, 1907). A remarkable series of importance, skills were beginning to be studied
studies on the retention of typing skill, initiated because investigators wanted to know about the
by Hill, Rejall, and Thorndike (1913), showed skills themselves.
"savings," in terms of practice time or the amount Toward the end of this period, the number of
of practice, involved in the relearning of typing studies involving skills increased slightly. Some
skill after two consecutive 25-year periods of no of these concerned handwriting proficiency, ways
practice (Hill, 1934, 1957). in which practice sessions could be structured to
One of the earliest systematic approaches to maximize motor learning, and whether or not
the understanding of motor skills was used by skills should be "broken down" into their com-
Woodworth (1899), who sought to identify some ponents for practice. Skills research placed greater
of the fundamental principles of rapid arm and emphasis on industrial applications (Gilbreth,
hand movements. This work, together with that 1909; Stimpel, 1933). So-called time-and-motion
of Hollingworth (1909), uncovered principles studies analyzed production-line assembly move-
about visual-motor performance that remain a ments; such research became the target of criti-
topic of current debate (e.g., Elliott, Helsen, & cism by workers because of the strict standards
Chua, 2001; Newell & Vaillancourt, 2001b). Some of performance it imposed on them. There was
other research, published in German and French, rising interest in the most efficient ways to per-
went unnoticed in the English literature for many form tasks such as carrying mortar and shoveling
years. Work on such topics as memory for move- coal and in methods of improving the conduct of
ments, speed-accuracy trade-offs, and phase work in extremely hot environments; these stud-
transitions in bimanual movements appeared ies became the early contributions to the emerg-
 Evolution of a Field of Study 7

ing fields of human factors and ergonomics. Some of commands delivered to the muscles. Indeed,
early theories of learning were published (e.g., Sherrington's early writings (e.g., Sherrington,
Snoddy, 1935), and work by physical educators 1906) remain relevant today (see tributes to his
interested in sports and athletic performances work in Gallistel, 1980; Stuart, Pierce, Callister,
emerged (e.g., McCloy, 1934, 1937). An interest Brichta, & McDonagh, 2001).
in factors associated with growth, maturation, Sherrington was one of those involved in
and motor performance began to surface; and research on the perception of movement. Various
studies by Bayley (1935), Espenschade (1940), sensory receptors were identified, such as the
McGraw (1935, 1939), and Shirley (1931) led the Golgi tendon organ, thought to signal changes in
way to the formation of the subarea that we now muscle tension, and the muscle spindle, thought
call motor development (see Thomas, 1997, for a to be involved in the perception of muscle length
historical review). and hence joint position. Sherrington coined the
The evolution of the study of the physiological now-common term proprioception, which refers
or neural bases of movement paralleled work in to the sense of body position and orientation
the motor behavior area during this period, but thought to be signaled by the various muscle and
without much formal contact between the fields. joint receptors together with receptors located in
The characteristics and contraction properties the inner ear.
of muscle tissue were a topic of early study by Somewhat later, scientists conducted research
Blix (1892-1895) and Weber (1846; see Partridge, on various brain structures. Herrick (1924) pro-
1983), who identified "spring-like" properties of posed numerous hypotheses about the functions
muscle that were later "rediscovered." Jackson of the cerebellum, many of which seem at least
conducted early investigations of the neural reasonable today. Also, patients with accidental
control of movement in the 1870s, well before the cerebellar damage were studied (e.g., by Holmes
advent of electrophysiological techniques that [1939]) in an attempt to pinpoint some of the
were to revolutionize the field. But what led to movement-control deficits associated with this
the development of various electrophysiological structure. Other brain structures, studied in
methods was the discovery by Fritsch and Hitzig patients with various kinds of brain damage,
(1870) that the brain is electrically excitable. These became subjects of interest (Adrian & Buytendijk,
methods gave rise to studies by Ferrier (1888) on 1931).
the responses in the brain's cortex to movements, Early neural control research mainly involved
as well as to the work by Beevor and Horsely very simple movements. Indeed, experimenters
(1887, 1890) on sensory and motor areas of the sometimes isolated nerve-muscle preparations
brain. or used animals with various degrees of experi-
One of the more important influences in the mentally induced spinal cord damage; here the
neural control area was the work on reflexes at concern about movement was usually secondary
about the end of the 19th century by Sherrington to interest in the neurological processes. When
and his coworkers. Sherrington studied and clas- movements were studied, the movement was
sified the major responses to stimuli presented often not considered in much detail; and mea-
to the extremities, and he believed that most of sures of the speed, accuracy, or patterns of move-
our voluntary movements resulted from these ment were usually missing from these reports.
fundamental reflexes. Sherrington is credited The motor behavior work, on the other hand,
with the creation of a number of classical con- typically involved very complex actions (e.g.,
cepts of motor control, most of which influence typing, telegraphy) but with very little empha-
thinking today. For example, he first talked of sis on the underlying neural or biomechanical
reciprocal innervation, the idea that when the mechanisms that controlled these actions.
flexors of a joint are activated, the extensors tend We can see an exception to this general sepa-
to be automatically deactivated, and vice versa. ration of the neural control and motor behavior
Also, Sherrington coined the term final common areas in the research of two important physiolo-
path, which referred to the notion that influences gists in the 1930s and 1940s. During this period,
from reflexes and sensory sources, as well as Nikolai Bernstein and Erich von Holst published
from" command" sources in the brain, eventually a number of seminal papers that have had a
converge at spinal levels to produce the final set significant impact on motor control theorizing
8 111 Motor Control and Learning

today (for more on Bernstein see "Nikolai Bern- until translations of their work appeared in Eng-
stein"). Unfortunately, many scientists involved lish-Bernstein's work had been published in
in the study of movement, from both behavioral Russian, von Holst's in German. Their early papers
and neural control areas, were unaware of the reappeared in English in the late 1960s and early
contributions made by Bernstein and von Holst 1970s (see Bernstein, 1967, 1996; Whiting, 1984;

Nikolai A. Bernstein

B!!Pri~ted; ~y permission, .from Fizkultura Sports

\~ai~i, .• · · · PUblishers. ·

* ~.In'; Befl'lsteil1 used the degrees of

e.~~ly•stages Of skilf acquisition, tearn-
.t~e;nufober o1 degrees of freedom so
·· · · nee' of movement is seen when
ttledegrees of .freedom,
i;lrrent motor control theorizing is
j<~~~~li\<~\Qglish t~anstations of Bernstein's
· · .i~cy got:)tim1esJo this day.

tJ~forcJ: Perg.a.mon Press.

;~~T.Ilj~.e}f(Eds.), Dexterity

,<_;'-,;:':,-!'

· ~V.M. Zatsiorsky (Eds.),

tics.

.,l(\)sciMce:JnM.L. Latash &

.,Bh~ll'tpaign,.tL: Murnan .Kinetics.
 Evolution of a Field of Study m 9

vonHolst, 1937/1973; Gallistel, 1980). Thus, while In addition to the formal laboratories that were
the two areas were being blended in Russia and supported by defense funds, research relevant to
Germany, these trends were not seen in the United the military was given increased federal funding.
States or England, where most of the work on This funding, in the form of contracts, grants,
movement was being conducted. Ironically, it was and training programs, was responsible for a
the translation of this work many years later, and shift of attention among psychologists toward
the attention that it received (e.g., Turvey, 1977), motor behavior research. The directions imposed
that served as a significant catalyst to the merging by federal funding agencies had, and continue
of the neural control and motor behavior areas. to have, a profound influence on the behaviors
studied and the research questions asked. The
Postwar Research area of motor behavior was important at the time,
World War II had profound effects on the world, and a great deal of funding was directed toward
and it is not surprising that it also had major it, convincing a large number of psychologists to
effects on movement research. One of the earliest become interested in research in this area.
and most direct effects can be traced to the need to A second major influence in the creation of the
select the most suitable people for pilot training, boom in motor behavior research in the postwar
which resulted in the creation of the U.S. Army period was the emergence of various theories
Air Force's Psycho-Motor Testing Program, initi- of learning, most notably that of Hull (1943).
ated by Arthur Melton in the early stages of the In scientific inquiry, theories generally provide
war (see Melton, 1947, for a description of some an organization of the conceptual issues and
of this work). Important studies were conducted findings as well as strong suggestions for future
on underlying motor, perceptual, and intellectual research. Theories stimulate and provide focus
abilities as they related to the selection of pilots for the research of others, and Hull's theory was
and other military personnel (see chapter 9). no exception. His was a general learning theory,
Similar studies were conducted in England. In applying to animals and humans and to verbal
addition, scientists studied gunnery, physical and motor behavior, and it was often tested with
training in the heat and cold, vehicle control, and motor tasks. A major emphasis of the theory was
many other issues related to combat performance. the fatigue-like process associated with long
When the war ended in 1945, the prevailing practice periods. The theory attempted to explain
attitude in the United States was that the efforts how fatigue and recovery processes combined to
related to selection and training of military per- determine the learning of motor skills, and many
sonnel should not be abandoned. Consequently, scientists worked with motor tasks to test Hull's
this research continued for many years (e.g., predictions. Most of this work has relevance to the
Druckman & Bjork, 1991, 1994). The military distribution of practice (see chapter 11) or to the
research effort was sustained when Arthur effects of fatigue on performance and learning.
Melton created (in 1949) the U.S. Air Force Hull's theory later proyed to be an inadequate
Human Resources Research Center, which car- account of the processes and variables that deter-
ried on many of the wartime programs but also mine motor learning and performance. However,
expanded to include studies of more general theories like Hull's provide strong directions for
interest. A major contribution of this program was research and contribute experimental data for use
Fleishman's work on individual differences and by future generations, even though the original
abilities (e.g., Fleishman, 1965, 2004). The war- theory may be shown to be inadequate.
time programs, devoted to personnel selection As the complexity of machines increased in
and motor abilities, had not resulted in the suc- this period and industrial methods became more
cess in pilot selection that had been anticipated. complicated, it became obvious that the capabili-
Researchers began to realize that training-not ties of humans to operate machinery effectively
selection-was perhaps more important to the were being exceeded. For example, a number
development of proficient pilots. Hence, much of serious airplane accidents that were initially
attention was directed toward procedures for attributed to "pilot error" were eventually traced
teaching motor skills, the transfer of motor skills to the way in which the instruments and controls
from one activity to another, and the retention of in the cockpit were arranged (Chapanis, 1965;
skills (chapter 14). Fitts & Jones, 1947; Schlager, 1994). Thus, shortly
1O Iii! Motor Control and Learning

after the war, a study of man-machine interac- movements-their movement time, their move-
tions, variously termed human factors, ergonomics, ment extent, and their accuracy (see chapter 7 and
or engineering psychology (a subarea of industrial "Paul Fitts"). The discovery of these two laws of
psychology), emerged. The guiding concepts behavior was an important advance in research,
were that humans were an important component tied together by information theory (Seow, 2005).
in most of the machinery involved in industry, In the middle of this postwar period, a great
and that such machinery must be designed with deal of motor behavior research was being con-
humans in mind. Although this thinking began in ducted-enough that Robert and Carol Ammons,
the military, it is now seen in automobile design themselves researchers in this area, created a
(Lee, 2008), the organization of assembly lines journal in 1949 titled Perceptual and Motor Skills
and work spaces, the design of home appliances Research Exchange. 2 The journal now publishes
and computer workstations, and many other both motor and nonmotor research, but during its
areas (Chapanis, 1999; Jagacinski & Flach, 2003; early years it served as a major outlet for motor
Karwowski, 2001; Wickens & Hollands, 2000). behavior work. In addition, Research Quarterly,3 a
Professional societies and journals were founded physical education research journal, and the Jour-
in the mid-20th century and continue to flourish nal of Experimental Psychology4 published a great
today (Cooke, 2008; Waterson & Sell, 2006). deal of motor behavior research during this period.
This period also saw a great deal of experimen- Toward the end of the postwar period, the
tal effort in England. One of the most important number of psychologists interested in motor
contributions was by Craik (1948), who proposed behavior research gradually declined, while
that we consider the brain as a kind of computer the number of physical educators interested in
in which information is received, processed, and the study of motor skills greatly increased. The
then output to the environment in the form of psychologists' lack of interest may be attributed
overt actions of the limbs. An important part of to decreased federal support for motor behavior
this general idea is the notion of central intermit- research, disillusionment with Hull's theory, and
tency, by which the human movement is seen as a increasing interest in other types of human behav-
series of discrete bursts rather than as continuous ior such as verbal learning and memory. This
(as it might appear). Craik's idea paved the way trend away from motor behavior research reached
for other English psychologists such as Welford, its peak in the mid-1960s when an "academic
who in 1952 proposed the still-relevant single- funeral" sponsored by Ina and Edward Bilodeau
channel hypothesis of attention (see chapter 4). was held at Tulane University. Renowned motor
Also, a great deal of work was done in ergonom- behavior psychologists gathered to hear the
ics, on training and conditions of practice, and on "last rites" and to bid each other farewell as each
hand movement control, particularly with respect moved on to different research topics in psychol-
to anticipation and timing (Poulton, 1950). ogy. The eulogies were recorded in a volume titled
The ideas about central intermittency and Acquisition of Skill (~ilodeau, 1966), which well
the analogies of the brain to the computer were describes the attitude of the times.
accompanied by similar new directions in psy- Motor behavior research was dead, or so the
chology and related fields. One of the new ideas psychologists thought; but they did not consider
was represented by Wiener's (1948) book Cyber- a man named Franklin Henry, trained in psy-
netics, which outlined an information-processing chology and working in the Physical Education
basis for human behavior. Also, Shannon and Department at Berkeley, who had a continuing
Weaver's (1949) The Mathematical Theory of Com- interest in motor behavior research (see "Frank-
munication established important principles of lin Henry" on p. 12). Together with A.T. Slater-
information processing that later led to system- Hammel and other leaders in physical education,
atic attempts to study the motor system in terms these new motor behavior scientists organized
of its capabilities and limitations in processing the North American Society for the Psychology
information (see Hick's law in chapter 3). In of Sport and Physical Activity (NASPSPA) and
keeping with the information-processing basis the Canadian Society for Psychomotor Learning
for behavior suggested by Craik and others, and Sport Psychology. These groups flourished in
Fitts (1954) established some now-famous funda- the 1970s. During this period, two books devoted
mental relations among characteristics of aiming strictly to motor behavior and motor learning
 Evolution of a Field of Study 11

Paul Fitts
Probably no empirical discovery in the area of motor control is better
known than Fitts' law, which states that the time observed to complete
an aimed movement depends on a simple mathematical relationship
between the distance to move and the size of the intended target (see
chapter 7). Called "Fitts' law" out of respect for its originator, this for-
mulation was an early attempt to apply mathematical and information-
processing principles to the understanding of human movements, and
it suggested that more complex limb control could be understood by
future application of such methods and thinking. The mathematical
equation described in Fitts' law characterizes the trade-off between
speed and error during simple aiming movements in a way that has
remarkable generalizability.
But Fitts' law was just one of many legacies of a psychologist whose
research had many implications for activities of daily living, especially
FIGURE 1.2 Paul M. Fitts (1912- for equipment design. Fitts' early research on the effects of the spatial
1965). compatibility between work-space displays and the controls used in
responding to these displays had a profound influence on the then-
emerging field of ergonomics and human factors (see chapter 3). In later years, Fitts also wrote about
perceptual-motor learning, suggesting that learning involves a progression through various stages,
each with distinctive characteristics regarding the capabilities of the human to process information
(see chapter 13). Paul Fitts was widely regarded as a leader in his area of research when he died
unexpectedly in the mid-1960s at the age of 53, well before his full potential could be realized.

Selected Bibliography
Fitts, P.M. (1954). The information capacity of the human motor system in controlling the amplitude of
movement. Journal of Experimental Psychology, 47, 381-391.
Fitts, P.M. (1964). Perceptual-motor skills learning. In A.W. Melton (Ed.), Categories of human learning
(pp. 243-285). New York: Academic Press.
Fitts, P.M., & Seeger, C.M. (1953). S-R compatibility: Spatial characteristics of stimulus and response
codes. Journal of Experimental Psychology, 46, 199-210.

Further Reading
Pew, R.W. (1994). Paul Morris Fitts, 1912-1965. In H.L. Taylor (Ed.), Division 21 members who made
distinguished contributions to engineering psychology (pp. 23-44). Washington, DC: APA.

were published, one in England (Knapp, 1963) ing; and Keele (1968) wrote an often quoted
and one in the United States (Cratty, 1964). Many review of motor control (see "Steve Keele" on p.
more followed (including the first edition of this 13). But these were the exceptions. As the 1970s
book in 1982). approached, the cluster of scientists in physical
Not all psychologists of the period were bored education and (to a limited extent) psychology
with motor behavior research. Fitts and Peterson began to evolve in new directions. Posner and
(1964) presented influential experiments on limb Konick (1966) and Adams and Dijkstra (1966) pre-
movement accuracy; Bilodeau and Bilodeau sented seminal articles dealing with short-term
(1961), Adams (1964), and Noble (1968) wrote memory for movements; Henry and his students
needed reviews of motor behavior research; (e.g., Henry & Rogers, 1960) were interested in
Adams (1968) wrote a theoretical treatment of motor programs; Posner (1969) studied attention
the role of sensory feedback in movement learn- and movement control; Pew (1966) examined
12 111 Motor Control and Learning

6~P~\~edormattceson two or more tasks, even

:i1as~~rn~l'Jiiu~derlying ability (e.g., dynamic vs. static
~ 1 :~~#0Qtbtal 1 StLJclents who subscribed to his general
· p€>$iUo~§lff Physical.education .departments during
;~f th~se j!>cholars created Ph.D programs and trained
:'~.resulfthc:ll Henry's influence became pervasive by

1
:n~~i~lea~hi~grnotdrskilL In R.C.. Brown·& G.S. Kenyon
i~~p.3lJ.1~340);.!Englewood Cliffs, NJ: Prentice Hall.

'<~s~~1afet)by•fot complicated movements and a

~S:'e~t~h <f71ul1tterty, $1,448-458.

~~~~~~riffien~'6rioPeraonal perapectives
tp~1Q'rd~ye!()prh~nt, and sport psychology. Champaign,

practice and automaticity; and Adams (1971) initi- were being made in neural control that would
ated a return to theorizing about motor learning. later be influential in joining the two areas. One
These emphases provided strong leadership for of the more important contributions was the
the motor behavior area in the 1970s. work on muscle spindle mechanisms by Merton
As in the early period, the neural control and (1953; Marsden, Merton, & Morton, 1972), to
motor behavior scientists were nearly oblivi- be discussed in chapter 5. While the specific
ous to each other; but important contributions mechanisms proposed by Merton now appear
 Evolution of a Field of Study fill 13

Steve Keele
The 1960s marked a critical time in the history of motor qehavior research,
as many psychologists shifted their focus to issues of cognition\ such as
memory and attention. Psychology appeared to have abandoned the
study of motor behavior, with a few notable exceptions. The appearance
of Keele's motor control literature review in 1968 created renewed inter~
est in important issues such as the motor program and the role. of vision
in movement regulation. His paper (with Michael Posner) on the period
during a movement that is required to process and act on visual information
remains one of our favorites for its simple, elegant methods of answering
a difficult question. Keele's book, Attention and Human Performance,
published several years later, situated movement prominently within a
late-filter theory of attention, reminding psychologists that movement is
not just a simple "output" in the human-computer metaphor (Rosenbaum,
FIGURE 1.4 Steve Keele
(1940-2005). 2005). Updated and expanded reviews of motor control ih later years
Courtesy of Betty Jean Keele. were considered mandatory reading for students in this area. Keele's later
research interests included important contributions to our understanding
of individual differences, timing, and sequence learning.

Selected Bibliography
Keele, S. W. (1968). Movement control in skilled motor performance. Psychological Bulletin, 70, 387-403.
Keele, S.W. (1973). Attention and human performance. Pacific Palisades, CA: Goodyear.
Keele, S.W. (1986). Motor control. In K.R. Boff, L. Kaufman, & J.P. Thomas (Eds.), Handbook of per~
ception and performance (pp. 30.1-30.60). New York: Wiley.
Keele, S.W., Jennings, P., Jones, S., Caulton, D., & Cohen, A. (1995). On the modularity of sequence
representation. Journal of Motor Behavior, 27, 17-30.
Keele, S.W., & Posner, M.I. (1968). Processing of visual feedback in rapid movements. Journal of
Experimental Psychology, 77, 155-158.

Further Reading
lvry, R.B., Mayr, U., Corcos, D.M., & Posner, M.I. (2006). Psychological processes and neural mecha-
nisms for action: The legacy of Steven W. Keele. Journal of Motor Behavior, 38, 3-6.
Keele, S.W. (1983). This week's citation classic. Current Contents, 37 (Sept. 12), 20.

to be incorrect (Houk, 1979; Smith, 1977), Mer- joint angles, suggesting that these receptors have
ton's original ideas about automatic regulation a large role in the perception of joint position.
of movement are reasonable in very general Numerous studies on the nature of muscle and
terms. Merton was one of the first to measure its contractile and mechanical (e.g., spring-like)
movements and neurophysiological processes in properties were also completed during these
the same investigation, creating a beginning for postwar years, and these studies attracted the
a blend of behavior and neurological emphases attention of contemporary researchers in motor
that was to follow. At about the same time, a great behavior and motor control (Rack & Westbury,
deal of research was devoted to the sensory recep- 1969). These mechanical characteristics of muscle
tors associated with movement perception and and of the motor apparatus were utilized by scien-
kinesthesis. Skoglund (1956) published a classic tists in the Moscow laboratories who were follow-
paper showing that the various receptors in a joint ing the earlier tradition of Bernstein. The exten-
capsule appear to be activated at certain specific sive work on movement control by this group,
14 Motor Control and Learning

originally published in Russian and thus generally Led by such researchers as Adams and Dijkstra
unknown to American and British researchers, (1966) and Posner and Konick (1966), the process
attracted a great deal of attention through vari- orientation helped to create the area of short-
ous translations (e.g., Gelfand, Gurfinkel, Tomin, term motor memory-the study of the processes
& Tsetlin, 1971; Kots, 1977). This research has underlying memory loss in simple movements
special relevance for the control of locomotion over short periods of time. Many studies were
and provides important links between the neural conducted in this area during the late 1960s and
control mechanisms and behavioral principles. early 1970s (see chapter 14). Studies were also
But despite these efforts, by 1970 almost no asso- completed on information-processing activities
ciation existed between the behavioral scientists during the learning of simple motor tasks (see
interested in more global and complex skills and chapter 12).
the neurophysiological scientists interested in More importantly, theorizing returned to motor
simple movements and neural control. behavior and learning, a style of inquiry that
had been relatively dormant since the failure of
The End of the Century Hull's (1943) theory. Adams sparked the interest
in theory when he presented a feedback-based
The 1970s brought massive changes in the field theory of verbal learning (Adams & Bray, 1970),
of movement control and learning. The strict followed the next year by a similar theory devoted
stimulus-response (S-R) orientation that had had to motor learning (Adams, 1971) (see "Jack
such a strong foothold during most of the century Adams"). Pew (1974a) returned to the old idea of
was overshadowed by the cognitive, information- a movement schema (Bartlett, 1932)-the abstract
processing approach. The publication of two hypothetical structures responsible for movement
books during the 1960s-Miller, Galanter, and control and evaluation, to be discussed in chapter
Pribram's (1960) Plans and the Structure of Behavior 13. And, one year later, the schema theory for the
and Neisser's (1967) Cognitive Psychology-had a learning of simple motor skills was presented
large impact on the field of experimental psychol- (Schmidt, 1975b). Together, these theoretical ideas
ogy in general and, later, on motor behavior too. generated a great deal of interest in motor skills,
The move toward cognitive psychology was a as this text makes evident later.
reaction to S-R theories of behavior. Ideas about The motor behavior field not only changed its
mental and motor processes, together with many direction, but also grew rapidly. Formal courses of
methods and paradigms for understanding them, study in universities flourished, and new journals
took the place of S-R theories. Perhaps more than appeared. In 1969, Schmidt founded the Journal
anything else, the books by Miller and colleagues of Motor Behavior, which was closely followed in
and Neisser popularized the study of mental pro- 1975 by the Journal of Human Movement Studies,
cesses such as response selection and movement created by the English motor behavior scientist
programming, whose existence must be inferred John Whiting (see "H.T.A. (John) Whiting" on
from the behaving individual rather than directly p. 16). A review journal titled Exercise and Sport
observed (see also Baars, 1986; Miller, 2003). Sciences Reviews was created in this period, and
Influenced by cognitive psychology, the motor it devoted a major portion of its space to motor
behavior field seemed to undergo a transition behavior research. Two more journals devoted to
from a task orientation, which focuses primarily the study of motor control also appeared before
on the effects of variables on the performance or the turn of the century; Human Movement Science
learning of certain motor tasks (or both), to a pro- appeared in 1982 and Motor Control in 1997. And
cess orientation, which focuses on the underlying throughout this time, the psychological journals
mental or neural events that support or produce (e.g., Journal of Experimental Psychology: Human
movements (Pew, 1970, 1974b; Schmidt, 1975b, Perception and Performance, Psychological Research,
1989a). Humans were considered processors of Psychological Bulletin, British Journal of Psychology,
information, and this approach was an attempt Psychological Review, Human Factors, and Ergonom-
to understand how movement information is ics) continued to publish motor behavior research.
coded and stored, how actions are represented As the field grew, motor behavior textbooks
in memory, and how information about errors is proliferated. More than 30 textbooks written
processed so that learning can occur. subsequent to Knapp's (1963) and Cratty's (1964)
 Evolution of a Field of Study 15

Jack Adams
One of the landmark advances in the developing field of motor
learning was the publication of '7\ Closed-Loop Theory of Motor
Learning" in the Journal of Motor Behavior in 1971. This paper
described one of the very first theories directed specifically at
motor learning and presented numerous testable hypotheses that
became the focus of a considerable number of research studies in
the 1970s. Although many of the tenets of closed-loop theory were
later found to need revision (see chapter 13), which is the case
for most theories that are scrutinized carefully, Adams' theory was
clearly a catalyst that moved the research "bar" in motor learning
much higher. But the 1971 theory was only one of many contribu-
tions to motor learning research that Adams made in a long and
fruitful career. His early work on psychological warm-up effects
as a source of memory decrement in practice (e.g., Adams, 1961)
led to one of the very first of many investigations on short-term
FIGURE 1.5 Jack A. Adams (1922-
2010). memory for movement information (Adams & Dijkstra, 1966).
Courtesy of Jack Adams. Throughout his research career, Adams maintained an interest in
the application of motor learning research in the area of human
factors and ergonomics, beginning with his work at Lackland Air Force Base in Texas. Later he turned
his attention to the effects of modeling and observational learning (e.g., Adams, 1986), again spear-
heading a rise in research activity in that area. Perhaps his most lasting legacy, however, will be as a
research historian-Jack Adams' ability to accurately and concisely synthesize volumes of research
history into manageable chunks of understandable knowledge will serve students of motor control
and learning for many years to come. His 1964 paper in the Annual Review of Psychology and the
1987 paper in the Psychological Bulletin remain as landmark reference works in motor behavior.

Selected Bibliography
Adams, J.A. (1961). The second facet of forgetting: A review of warm-up decrement. Psychological
Bulletin, 58, 257-273.
Adams, J.A. (1964). Motor skills. Annual Review of Psychology, 15, 181-202.
Adams, J.A. (1971). A closed-loop theory of motor learning. Journal of Motor Behavior, 3, 111-150.
Adams, J.A. (1987). Historical review and appraisal of research on the learning, retention, and transfer
of human motor skills. Psychological Bulletin, 101, 41-74.

further Reading
Schmidt, R.A. (1990). Distinguished scholar award to Jack Ashton Adams. NASPSPA Newsletter, 15, 4-5.
Schmidt, R.A. (2011 ). Jack Adams, a giant of motor behavior, has died. Journal of Motor Behavior,
43, 83-84.

work were published, as were a large number of about movement control and made increased
edited volumes on more specific topics. use of various electrophysiological and biome-
The 1970s were the beginning of a long-needed chanical techniques to understand the functions
merger between the neural control and the motor of the central nervous system in movement. The
behavior scientists. Many people were trained for- neural control scientists were shifting from stud-
mally in both motor behavior and neural control, ies that examined only the neural mechanisms to
and these people completed the bridge between studies investigating these mechanisms during
the two levels of analysis. More and more behav- complex movements. Much of this latter work
ior-oriented scientists began to ask questions was done with animals, principally monkeys
16 Motor Control and Learning

H.T.A. (John) Whiting

Another pioneer who had tremendous influence on motor
behavior research was H.T.A. (John) Whiting. Like Franklin
Henry in the United States, discussed earlier in the chapter,
Whiting introduced traditional psychological techniques to
the study of human motor control in the performance of sport
tasks. His research on catching performance was the first to
demonstrate that continuous vision of a ball in flight is not
necessary in order to support accurate performance (see
chapter 5). Whiting's book, Acquiring Ball Skill (1969), was
one of the first psychologically oriented texts to be devoted to
the study .of motor skills. Later Whiting turned his attention to
the study of learning in the Bernstein tradition (see chapter
13). During this time Whiting al.so founded and served as the
editor of the Journal of Human Movement Studies and, later,
Human Movement Science. Another important legacy was
the many graduate students Whiting supervised at Leeds
during the 1960s and 1970s, and then at the Free Univer-
sity of Amsterdam in the Netherlands during the 1980s and
FIGURE 1.6 H.T.A. (John) Whiting (1929-
1990s. Many of these scientists continue to carry on Whiting's 2001).
legacy today. Courtesy of Vrije Universiteit.

Selected Bibliography
Vereijken, B., Whiting, H.T.A., & Beek, W.J. (1992). A dynamical systems approach to skill acquisition.
Quarterly Journal of Experimental Psychology, 45A, 323-344.
Whiting, H.T.A. {1969). Acquiring ball skill: A psychological interpretation. London: Bell.
Whiting, H.T.A., Gill, E.B., & Stephenson, J.M. (1970). Critical time intervals for taking in flight informa-
tion in a ball-catching task. Ergonomics, 13, 265-272.

Further Reading
Savelsburgh, G., & Davids, K. (2002). "Keeping the eye on the ball": The legacy of John Whiting (1929-
2001) in sport science. Journal of Sport Sciences, 20, 79-82.
Snyder, C., & Abernethy, B. (1992). The creative side of experimentation: Personal perspectives from
leading researchers in motor control, motor development, and sport psychology. Champaign, IL:
Human Kinetics. ·

and cats. Records from electrodes implanted behaviors and neurological processes in order to
in the brain, spinal cord, or muscle were taken provide a more complete understanding of how
while the animal was engaged in motor activity. movements are controlled. This emphasis marked
Representing this approach are Grillner and his a refreshing change from the earlier research in
colleagues (1972, 1975) and Smith and her col- which the movements per se were hardly consid-
leagues (1986), who studied locomotion in cats; ered. The association between motor behavior and
Evarts (1972, 1973), who studied a number of motor control resulted in several reviews written
separate brain structures in monkeys; and Houk toward the end of the 1970s, such as those by
(1979) and Granit (1970), who studied the gamma Brooks (1975, 1979, 1986), Grillner (1975), Wetzel
motor system in monkeys and humans. and Stuart (1976), and Gallistel (1980). Behaviorists
The essential feature of all this work is the strong and neurophysiologists participated in a number
attempt to find an association between movement of scientific meetings, and the results appeared in
 Evolution of a Field of Study !I'! 17

edited volumes (e.g., Gandevia, Proske, & Stuart, The late 1970s and early 1980s were also char-
2002; Stelmach & Requin, 1980, 1992; Swinnen, acterized by a general decline in interest in motor
Heuer, Massion, & Casaer, 1994). learning, with a corresponding increase in issues
An additional change occurred toward the of movement control or human performance. This
end of the century-one far more subtle than was unfortunate, because the issues involved in
those just mentioned. Rather than remaining a learning have perhaps the most practical appli-
mere blending of two different fields, the field cation to training, rehabilitation, and teaching
of motor control acquired an independent iden- in general. But there was a renewed interest in
tity. It became a field of study in its own right, learning toward the end of the century, sparked
complete with its own journals and methods for in part by counterintuitive findings regard-
asking research questions and collecting data. ing how practice scheduling (Shea & Morgan,
Such methods involve the use of sophisticated 1979; Lee & Magill, 1983b; Magill & Hall, 1990)
techniques for recording and analyzing move- and augmented feedback (Salmoni, Schmidt, &
ments (such as electrophysiological recordings), Walter, 1984; Schmidt, 1991a) could be organized
cinematographic and three-dimensional analyses, to optimize the learning environment. Much of
measurement of the kinematics of movement, and this work is described in chapters 11 and 12.
advanced methods for examining the involve-
ment of brain structures, integrated with the more Motor Control and Learning
traditional techniques for studying learning (e.g., Research Today
Corcos, Jaric, & Gottlieb, 1996; see also chapter 2).
The influence of Bernstein (and others) resur- The integration of motor control research that
faced in the writings of a number of scientists who developed toward the end of the 20th century
conducted motor control research (e.g., Greene, continues today. For example, behavioral studies
1972; Kelso, 1995; Kugler & Turvey, 1987; Reed, of motor control and learning now appear in jour-
1988; Turvey, 1977). According to Turvey (1990), nals that in the past were strictly oriented toward
Bernstein's legacy resulted in two rounds of theo- neuroscience (e.g., Experimental Brain Research,
rizing and experimentation. The first round dealt Neuroscience Letters, Journal of Neuroscience), and
with the degrees of freedom problem-research neuroscience-oriented studies appear in journals
addressing how a system with many independent that were formerly primarily behavioral (e.g.,
parts could be controlled without the need for an Journal of Motor Behavior, Human Movement Sci-
executive "decision maker." The second round ence). Studies of the specific brain mechanisms
extended Bernstein's thinking on coordination involved in the performance of movement tasks
and the degrees of freedom problem to a search have appeared in specialist journals (e.g., Neuro-
for laws and principles of self-organization. Much Image) and represent a rapidly growing approach
of this work uses physical biology as its basis. The to the study of motor control. Studies of motor
dynamical-systems perspective (e.g., Kelso, 1995) control and learning also continue to be published
suggests that coordinated movement evolves in journals devoted to specific professional topics
over time as a function of the interaction among (e.g., Physical Therapy, Human Factors), and this list
the body parts, and between the body parts and continues to expand to include journals that did
the physical world. Also associated with this not previously publish motor behavior articles
view are the ideas that perception and action are (e.g., American Journal of Surgery). New societies
functionally inseparable-that understanding (e.g., International Society of Motor Control) and
the motor system depends on understanding the journals (e.g., Motor Control, Frontiers in Movement
physical principles of our actions and how they Science and Sport Psychology) have emerged. In
interact with biological functions. Advocates short, more research in the area of motor control
of these traditions showed a reluctance to use and learning is being published now than ever
cognitive-psychological styles of inquiry with before. And that research is being conducted by
hypothetically defined brain mechanisms such as researchers who have a greater diversity and
memory, motor programs, schemas, and the like. breadth of knowledge than ever before.
This approach contributed a different emphasis to The computer, employed in relatively few
the attempt to understand motor behavior (e.g., laboratories for research investigations prior
Anson, Elliott, & Davids, 2005). to the 1980s, is now used almost universally
18 Motor Control and Learning

for conducting research and organizing data. less of the changes that occur, however, the lessons
Analyses that used to take many hours can now of history tell us that the scientific disciplines
be performed in a matter of minutes. In short, the combining to define motor control and learning will
capacity to conduct research of very high quality continue to undergo refinement with further accu-
has grown at a dizzying rate. mulation of knowledge and application.
This is an exciting time for research in motor
control and learning, an area of literature now so
broad in scope that it would seem to be more and Summary
more difficult for the student to remain aware of
the most recent publications. Ironically, however, This text is fundamentally concerned with move-
it has never been easier to retrieve research infor- ments of human beings. Some of these movements
mation in this area. The Internet, which did not are probably genetically defined while others
even exist at the time the second edition of this are skills, requiring practice or experience. Even
book was published, has revolutionized the way though most of the field of human behavior deals
in which motor behavior research is made public. in one way or another with movement, in this text
E-mail alerting services deliver the tables of con- we focus primarily (but not exclusively) on those
tents of new journal issues as soon as they are movements in which cognitive involvement is
published and previews of soon to be published relatively slight, and for which the nature of the
papers. Almost all journals are now available in movement itself-rather than the choice of the
electronic formats, and the articles are available movement from already learned alternatives-is
to consumers at the click of a mouse. Moreover, the primary determinant of success. We focus on
many journals are not only publishing current movements that do not have a heavy concentra-
issues in electronic formats but also archiving tion on cardiovascular endurance or strength, as
all of their pre-electronic issues. General search these activities seem to be more closely aligned
engines (such as Google) and specialized search with other fields of study. Finally, the focus is on
engines (such as Google Scholar, PubMed, Web many different movements that fall within the
of Science) make the literature retrieval for a categories mentioned, such as those in musical
particular topic, author, or paper very fast and performance, work, industry, and sport and other
effective. The Internet has also made databases activities of daily living.
easy to access. Carlton and colleagues (1999) The field of movement control and learning,
described a Web-based laboratory, for example, viewed from a historical perspective, emerged
where a student can retrieve and analyze a set of from the separate but parallel fields of motor
data that addresses a particular research topic. behavior and neurophysiology. Both fields
However, the student of motor behavior must showed steady growth through the beginning
also be aware, now more than ever, that what of World War It then increased in growth and
is available online can be of variable quality. sophistication after the war and through the 1960s
Respected journals maintain the peer review and 1970s. The two.fields, however, were largely
process, whereby a paper submitted for publica- separated until the early 1970s, when they began
tion undergoes careful scrutiny by a small com- to share common problems and methods. Motor
mittee of established peers ("referees") before it learning research, largely forgotten during the
can appear in press under the journal's title. This '70s, reemerged as a strong research area toward
scrutiny is maintained regardless of whether the the end of the century. Today the computer (and
published work appears in paper or electronic the Internet) has revolutionized the way in which
format. However, not all articles that one can research is conducted and disseminated.
retrieve on the Internet have undergone this peer
review process. As in the previous centuries, the
best and most influential research articles are
Student Assignments
almost always those that have undergone careful
and rigorous scientific scrutiny. 1. Answer the following questions and bring
It is impossible to know what changes will the information to class for discussion:
emerge in the next decade. Perhaps these words a. How does a behavioral level of analysis
will no longer be available in paper format. Regard- of motor control and learning differ
 Evolution of a Field of Study m 19

from biomechanical and neurophysi- www.i-s-m-c.org

ological levels of analysis?
These journals specialize in publishing research
b. If you could interview one of the people in motor control and learning:
featured in the highlight boxes (Bern-
stein, Fitts, Henry, Keele, Adams, Whit- www.tandf.co.uk/journals/titles/00222895.asp
ing), what would you ask him about
www.humankinetics.com/MC/journalAbout.
motor control and learning?
cfm
c. What has the proliferation in publica-
www.elsevier.com/wps/find/joumaldescription.
tion outlets contributed to our current
cws_home/505584/description#description
understanding of research in motor
control and learning?
2. Find a research article authored by any of Notes
the six people listed in question lb. Are the
findings of that research article still relevant 1 Learning the rules (grammar) of a language would

today? probably not be of much relevance for the area of motor

control, but learning to make the guttural sounds involved
in German or the nasal sounds inherent in French could
logically be included in the area of motor control (Mac-
Web Resources Neilage, 1970).
2 Now called Perceptual and Motor Skills.
These are some of the professional organizations 3 Now called Research Quarterly for Exercise and Sport.
that promote research in motor control and learn-
4 Now subdivided into five different journals (one of
ing through newsletters and annual conferences:
which is entitled Journal of Experimental Psychology: Human
www.naspspa.org Perception and Performance).
www.scapps.org
 major goal of this book is to present not (i.e., the class of task) under consideration. That
only relevant principles and theories about is, the relation between certain independent
the nature of motor performance, motor control, and dependent variables is often different for
and motor learning but also the research evidence one class of task or behavior when compared to
that supports (or, in some cases, refutes) these another. Without classification, the laws of motor
principles and theories. In evaluating this evi- control would be far more difficult to understand.
dence, it is necessary to understand some of the Movement behaviors have been classified
methods involved in the research and the ways in in various ways. Two important classification
which the motor behaviors are measured, so that schemes are the discrete I continuous I serial
we can more effectively establish the relevance of dimension, which is based on the particular move-
the evidence to the particular principle or theory ments made, and the open/ closed dimension,
in question. Later in the book (chapter 10), we which is determined by the perceptual attributes
focus on the methods and paradigms used spe- of the task.
cifically in the study of motor learning.
Discrete, Continuous,
Classification of Motor Skills and Serial Skills
Discrete movements are those with a recognizable
In any field of study, the objects under investi- beginning and end. Kicking a ball, throwing,
gation are usually classified according to some striking a match, and shifting gears in a car are
scheme or framework in order to simplify discus- examples (figure 2.1). The end of the movement is
sion. The field of motor behavior is no exception. defined by the skill in question, not arbitrarily by
Classification of movements and motor tasks is the time at which an observer ceased examining
important for two fundamental reasons. First, in it, as would be the case for swimming or jogging,
the research literature on motor behavior and con- for example. Discrete skills can be very rapid,
trol, various terms are used to describe the tasks requiring only a fraction of a second to complete
and movements. These terms must be understood (e.g., kicking, blinking an eye); but they can also
if we are to communicate about the field. The require much more time for completion, as in
second reason is that the laws of motor behavior writing your signature. Discrete skills can also be
seem to depend on the kinds of performances quite cognitive. For example, one laboratory task

Ill 21
22 Motor Control and Learning

is to press one of four buttons when one of four or an uncertain way on the screen. The second
lights comes on; the problem for the subject is to cursor is moved by the subject via a hand control,
decide which button to press in response to which and the subject's task is to minimize the distance
light. Thus, the decision about which button to (or error) between the two cursors.
push is paramount, and the "how" of pushing the Two kinds of tracking tasks are used com-
button is clearly secondary in importance. While monly in motor behavior research: pursuit track-
many discrete skills have large verbal-cognitive ing and compensatory tracking. In pursuit tracking,
components, there are certainly examples of dis- experimenter-produced actions of the target and
crete skills that are highly "motor" as well. the subject's own movements are both displayed.
Continuous movements-defined as those that The previously mentioned task of steering a car is
have no recognizable beginning and end, with a good example of pursuit tracking. In compensa-
behavior continuing until the movement is arbi- tory tracking, the experimenter-produced varia-
trarily stopped-are at the opposite end of the tions in the track are combined with the subject's
continuum (in figure 2.1). Swimming, running, movements to produce a single displayed value,
and steering a car are examples of tasks that have and the subject's goal is to maintain this value
arbitrary ends. Continuous tasks tend to have at some constant location. Practical examples of
longer movement times than do discrete tasks compensatory tracking are often seen in aircraft
(they might even continue all day). This, however, instruments, such as the glide slope indicator.
should not be taken as basic to their definition. Here only the difference between the proper
A common class of continuous skills, both in altitude and the actual altitude is displayed; and
everyday experience and in the laboratory, involves when the pointer is in the middle of the screen,
tracking tasks. The tracking task is characterized the pilot's altitude is correct. Compensatory track-
by a pathway (track) that the individual intends ing tasks are almost always more "difficult" than
to follow and a device that the person attempts to pursuit tracking tasks, particularly if the behavior
keep on the track via certain limb movements. In of the track is irregular and unpredictable.
steering a car, for example, the track is the road, Tracking tasks also vary in terms of the aspect
and the device is the car, steering wheel, and so of the display that the subject controls. The most
on. A very common laboratory example involves simple is the zero-order, or positional, display. If
two cursors on a computer monitor. One of the the subject moves the handle from one position
cursors is moved by the experimenter (or by the to another and then stops, the indicator on the
computer), and it can move in either a predictable display moves a proportional amount and also

FIGURE 2.1 The discrete/serial/continuous classification for motor behavior.

 Methodology for Studying Motor Performance Ill 23

stops; that is, the handle movements control the starting a car, preparing and lighting a wood
position of the pointer. In first-order, or velocity fireplace, and many tasks involved in production
control, movement of the handle causes changes lines in industry. Serial tasks can be thought of as
in the velocity of the pointer. Moving the handle a number of discrete tasks strung together, and
further in one direction causes the velocity of the order (and sometimes timing) of the actions
the pointer to increase in the same direction, and is important.
stopping the handle movement off center results
in a constant velocity of pointer movement. In a Open Versus Closed Skills
second-order task, the movements of the control Environmental predictability during the per-
produce changes in the pointer's acceleration. formance provides another basis for classifying
Keeping the handle centered produces zero movement skills (Poulton, 1957; Gentile, 2000).
acceleration, but moving the handle to a new Open skills are those for which the environment
position off center accelerates the pointer in the is constantly (perhaps unpredictably) changing,
same direction. Each of these kinds of tracking so that the performer cannot effectively plan
tasks is used in research, and there are real-world the entire movement in advance (figure 2.2). A
examples of each in various control systems (see good example is the penalty shot in ice hockey.
Poulton, 1974, for more details). While skating toward the goalie, the player may
One final type of tracking task is step tracking. make a general decision about whether to go
In this task, the track "jumps" from one fixed left or right, but the final decision may depend
location to another, often unpredictably, and the on what the goalie does. Another example is
subject's task is to move the control as quickly driving on a busy freeway. Although you may
as possible to correct this sudden change in the make a general plan about what you want to do,
track's location. Step tracking tasks can be either such as pass another car, your precise plans must
pursuit or compensatory. be left flexible enough to deal with unexpected
Serial movements are neither discrete nor con- actions of other drivers. Success in open skills is
tinuous, but usually are comprised of a series of largely determined by the extent to which the
individual movements tied together in time to individual is successful in adapting the planned
make some "whole." These types of movements motor behavior to the changing environment.
appear in the center of the continuum in figure Often this adaptation must be extremely rapid,
2.1 because they can be rather long in duration and the effective responder must have many dif-
but are not stopped arbitrarily. Examples are ferent actions ready to implement.

FIGURE 2.2 The open/closed continuum for motor behavior.

24 Motor Control and Learning

Closed skills, for which the environment is pre- Objectivity

dictable, are at the other end of the continuum
shown in figure 2.2. An environment may be pre- The term objectivity is important in measurement
dictable because it is almost perfectly stable-for because of the scientific demand that observations
example, the environment in which one performs be subject to public verification. A measurement
skills like archery, bowling, or signing one's name system is objective to the extent that two observ-
to a check. An unstable yet predictable situation ers evaluating the same performance arrive at
can also arise when the environment is variable the same (or very similar) measurements. For
but the changes are very predictable, or have been example, using a tape measure to determine the
learned as a result of practice, or both; examples distance a javelin was thrown yields very simi-
are juggling and industrial production-line tasks. lar results regardless of who reads the tape. By
Here, the essential feature is that the environment comparison, evaluation of performances such as
for the next few seconds or so is essentially pre- diving, gymnastics, and figure skating is more
dictable, so that the movement can be planned subjective-although elaborate scoring rules,
in advance. Of course, some skills have environ- complete with certification tests for judges, help
ments that are semi-predictable, and these can make it more objective. From the point of view
be classified somewhere between the ends of of research in motor behavior, it is important to
the open/ closed continuum in figure 2.2. Farrell use performances in the laboratory for which the
(1975) has provided additional distinctions that scoring can be as objective as possible; and this
help us to classify movements on this dimension. necessarily limits the usefulness of some tasks for
The observation that open skills seem to require providing an understanding of motor behavior
rapid adaptations to a changing environment, in general.
whereas closed skills require very consistent and A second aspect of objectivity relates to the
stable performances in a predictable environ- sensitivity of the measuring device to changes in
ment, raises interesting questions about how the the skill of the performer. How high did the girl
two classes of skills might best be learned and jump when she set the school record in the high
taught. Should the methods for teaching open and jump? The record books report that she jumped
closed skills be different? Do different individuals 5 ft 6 1/2 in. (168 cm). But it is more accurate
perform better in one of these skill classes or the to say that she jumped at least 5 ft 6 1/2 in. The
other? Are the laws of performance different for measurement system for the high jump (and for
the two kinds of skills? Evidence suggests that the pole vault) indicates only whether or not the
some of the answers to these questions are yes, bar was knocked off during the jump, and it is
and we will discuss these issues in more detail possible that the successful jumper cleared the
in later chapters. bar by 6 in. In such situations, the scoring system
is somewhat insensitive to the variations in the
performer's actual l~vel of skill.
Basic Considerations In this example, the scale of measurement itself
in Measurement was acceptable in terms of sensitivity; in fact, we
often see high jump officials measuring the height
A fundamental issue in any science concerns of the bar very carefully, perhaps to the nearest
how the behaviors of the objects of study are 1/ 4 in. Yet this method is not totally objective, as
measured, and motor behavior is no exception. we have just argued. However, sometimes the
We often need to operationalize skills so that we scale of measurement itself is lacking in preci-
can assign numerical values to certain perfor- sion. Often, continuous measures are artificially
mances based on the quality of the movements. categorized, as in hit/ miss scoring for basketball
Scientists must be able to measure the degree of and golf putting. For these examples the details of
skill exhibited by a performer in scientifically the movement, together with information about
acceptable ways. Some of the criteria for good the performer's skill level, can be lost in the over-
measurement systems are (a) the objectivity of simplified measurement method. Such methods
the system, (b) the reliability (or stability) of the make it more difficult to determine whether an
measuring system, and (c) the validity of the individual has improved on the task with prac-
system. tice, or to identify which of two individuals is
 Methodology for Studying Motor Performance 25

the more skilled performer-both of which may practical situations, but the alternative of study-
be critical questions in the study of movement. ing skills during a game adds sources of variation
that reduce the reliability of the measures.
Reliability The procedures mentioned can reduce variabil-
A second aspect of the measurement system that ity in experimental settings. But even when the
is important to motor behavior is reliability-the task is well learned and simple, when the experi-
extent to which the measurement is repeatable mental situation is well controlled, and when the
under similar conditions. A lack of reliability can subject is trying to do well, there remains a great
result from random technological error, such as deal of variability because biological systems
the stretch in measuring tapes, errors in clocks, are inherently somewhat unstable. Experimen-
and human errors in reading instruments. These tally, the best method for countering this type
errors, while they might seem to be important of variability is to record many observations of
sources of unreliability, probably contribute the "same" behavior on the same subject, taking
very little to unreliability as long as a researcher the average of a large number of measurements
uses quality recording apparatus and careful under essentially identical conditions. With this
procedures. The most important source of unreli- procedure the variations in the subject's perfor-
ability manifests itself when the performer does mance tend to "average out," raising the reliabil-
not perform the same action twice in exactly the ity of the measurement system, so that the mean
same way. Some of these intrasubject variations of a large number of observations more closely
are caused by momentary changes in the internal represents the construct being measured.
state of the subject (degree of attention, fatigue, or
boredom, for example), while others are caused
Validity
by systematic changes, such as alterations in strat- Another aspect of the measurement process is
egy, the amount of practice, and the like. Both of validity, the extent to which the test measures
these factors tend to obscure the constructs that what the researcher intends it to measure. An
scientists are attempting to measure. important aspect of validity (called construct
Experimenters seek to minimize these sources validity) is the extent to which the measures
of variability through experimental control in taken actually reflect the underlying construct
the testing situation. Researchers typically use of interest. We would be reasonably comfortable
written or prerecorded instructions in order to with a 10 min typing test to operationalize typing
eliminate variability in what and how informa- skill, but we would perhaps be less comfortable
tion is presented; they use testing rooms that are with a measure of finger tapping speed to assess
either silenced or sound deadened; subjects are typing skill. There are, on the other hand, situa-
tested one at a time to eliminate variability due tions in which validity does not seem to present
to another person's presence in the room; and the much of a problem. One of these involves what
entire experimental session is often quite formal are often called face valid tests, which are so obvi-
and impersonal. This is the primary reason some ously measures of the concept of interest that
researchers tend not to measure skills in every- they usually are not questioned. For example, if
day settings-at a ball game or on an industrial we wish to determine which member of a group
production line, for example. In these situations, of individuals has most skill in javelin throwing
the environment is not well controlled; there are (a construct), we might have them all throw the
many sources of variation from other players or javelin as a standardized test.
workers, from changes in the score of the game, Another class of measurement situations in
from the level of proficiency of opponents, from which the importance of validity is minimal is in
day-to-day changes in the weather, and so on. experiments on motor learning (Schmidt, 1989a).
Primarily for reasons of experimental control, In these situations, an arbitrary task is created
motor behavior tends to be most profitably stud- that represents a motor performance novel to
ied in the laboratory, away from these sources of the subject, and the experimenter studies how
variability. To be sure, there is a trade-off in this the subject attempts to learn it or what variables
approach: Laboratory studies tend to make the influence that learning. The particular constructs
situation less natural and more artificial, and the being measured (e.g., balance, timing, movement
measures taken are not quite as directly related to speed) frequently are not very important to the
26 Motor Control and Learning

experimenter, because the primary focus is on act at a particular time (e.g., hitting a baseball). A
the variables that affect performance and learn- particular force, distance, speed, or time can be
ing generally. defined as the subject's target; then, deviations
of the subject's performances with respect to this
target are measured. The level of analysis that is
Measuring Motor Behavior least sensitive comprises dichotic outcomes, such
as when the performances are scored as hit/miss
In the field of motor behavior and control, mea- or right/wrong, as in shooting a basketball or
surement can be approached in essentially three judging which of two lifted weights is heavier.
different ways. At the most general level, we can We can refine the accuracy score by dividing the
describe how well a movement achieved some possible outcomes into hit/almost hit/miss or
environmental goal that was inherent in the task by dividing a bull's-eye into 10 or more zones,
(e.g., whether or not a target was struck). Here for example. But motor performance is complex,
the emphasis is on the outcome of movement. and more sophistication in the measurement of
At a more specific level, we may be concerned accuracy is usually required.
with quantifying the actual movements the In the discussion that follows, assume that a
person made. In this case, the focus of analysis is single performer is striving for accuracy in arriv-
describing the movement itself. The third level of ing at some target (e.g., a force, a speed, a location
analysis entails the study of the brain and central in space) and that the movement outcomes can be
nervous system prior to and during the produc- placed along some measurable dimension (e.g.,
tion of movement. At this level, researchers are kilograms, centimeters per second, centimeters)
interested in the neural activities involved in as in figure 2.3. Let the correct value along this
planning and executing movements. dimension-the target-have the value T. The
values that the performer actually achieves are
Describing the Outcome abbreviated by xi' where i is a subscript notating
of Movements a particular trial (i.e., the ith trial). For example,
The first aspect of measurement in motor behav- x23 is the score on the 23rd trial. In the simple for-
ior is quantification of the extent to which a given mulas that describe these fundamental statistical
movement achieved the goal that was intended accuracy scores, the symbol~ means "the sum
or instructed. For example, did the movement of." For example, ~xi means to add up all of the
result in striking the target, or was the movement values xi' where i ranges progressively from 1
made at the right time? Such measures generally through n, with n = 5 in the following example:
concern the movement in relation to some object (2.1)
or to another performer in the environment,
although some movements (e.g., modern dance, In the following explanation, assume that the
diving) may not be so closely associated with target (T) is 100 units ~nd that the individual does
other environmental elements. The achievement not always achieve this target score. In figure 2.3
of such environmental goals can be assessed in there are five scores: 93, 103, 99, 105, and 96 units
essentially four fundamental ways-through for trials 1through5, respectively.
measures of (a) error, (b) time and speed, (c) It is obvious that no single trial will be very
movement magnitude, and (d) performance on effective in describing the subject's behavior, as
secondary tasks. the scores possess a great deal of variability. One
solution is to combine these scores to achieve a
Measures of Error for a Single Subject more representative measure of the subject's capa-
Many performances require the subject to per- bility. In the study of motor behavior, researchers
form some action with maximum accuracy. Thus, have typically focused on five methods for com-
the performance measures represent the degree to bining scores into measures of "error," and each
which the target was not achieved-a measure of one has a slightly different meaning in terms of
error. The accuracy goal can be imposed in many the performer's capability. These methods are
ways; for example, subjects can be asked to move described in the next sections as (a) constant error,
with a certain amount of force, hit a certain spatial (b) variable error, (c) total variability, (d) absolute
target, move at a certain speed, or perform some error, and (e) absolute constant error.
 Methodology for Studying Motor Performance 27

FIGURE 2.3 An arbitrary measurement scale, showing locations of a target (T) and of five hypothetical movement
attempts (x1 , . • . , X5 ).

Constant Error (CE)-Computation The first ference scores are shown in column C, headed
statistic to be considered as a measure of the (xi - T). It is important to notice in this column
subject's accuracy is the constant error (CE), which that for CE, the sign (+ or - ) of the difference
measures the average error in responding. Its is retained. Next, the summation sign calls for
formula is adding the values (using the signed values) for
Constant error= CE= 2. (xi - T) In (2.2) each of the trials (in this case for each of the five
trials), and this sum is presented at the bottom
where xi is the score on trial i, Tis the target, and n of column C (-4.0). Then the formula calls for
is the number of trials the subject performed. It is dividing by n, the number of trials, in order to
very easy to compute this measure from table 2.1, get the average CE over trials. The final CE score
which can serve as a work table for computing all is -4.0/5, or -0.80.
of the statistics presented in this section on error
measures. The trial numbers are listed in column Interpretation of Constant Error The CE score
A; the scores obtained (x) are given in column B. of -0.80 indicates that, on average, the subject fell
All other values in the table are computed from slightly short of the target (by 0.80 units). Notice
these initial values (remember that T = 100 in that the CE is given in units that represent the
this example). amount and direction of deviation relative to the
To compute the CE, the numerator calls for target, sometimes called bias. One could also ask
finding the difference between each of the scores for the subject's scores on the average by consult-
on the test (x) and the target (T = 100); these dif- ing the mean for column B. Thus, the average
28 Motor Control and Learning

score was 496 I 5 = 99.2 units, meaning that the are given in column E. Next, obey the summa-
subject fell short of the target by 99.2 -100 units, tion sign and add the squared values, the sum of
which is also -0.80 units. The CE represents the which (96.80) is shown at the bottom of column
average magnitude of the movement and mea- E. Then divide this sum by the number of cases
sures the direction of the errors on the average. (n = 5) to get 19.36 and take the square root to
While a measure of average error bias might, arrive at the final answer of 4.40 units.
at first, seem satisfying to students as a measure
Interpretation of Variable Error The VE reflects
of accuracy, notice that the value computed for
the variability, or inconsistency, in movements,
the subject (-0.80) was far smaller than the error
as can be seen from the "ingredients" in the
for any of the single movements that contributed
formula. The important feature is the difference
to the average. The movements were scattered a
between the subject's score on each trial and his
great deal, with the center of movements being
own average score. Thus, if one subject always
roughly the target that was the goal. What the
moves very consistently, the VE will tend to be
CE does not consider is this amount of scatter,
small. If the subject always receives the same
variability, or inconsistency in performance of
score, even though it is not the correct one (such
the movements. Consider a second hypothetical
as a score of 99 on all five trials), then the VE will
subject with scores of 99, 99, 99, 99, and 100. These
be zero. This is so because the subject's average
scores represent a very small scatter but would
result in precisely the same CE score as for the score will be 99, and the difference between each
subject we have just been considering (-0.80). of the scores and the average will always be zero
For this reason, another measure of error, the as well.
variable error, is used to describe the subject's Thus, VE does not depend on whether or not
inconsistency. the subject was close to the target, since it is the
measure of spread about the subject's own aver-
Variable Error (VE)-Computation The vari- age. To illustrate, the VE for the set of scores 43,
able error (VE) measures the inconsistency in 53, 49, 55, and 46 achieved during aiming at a
movement outcome. It is the variability of the target of 100 units will be precisely the same (4.40)
subject's performances about the mean value and as that calculated in the previous example. (We
is calculated by the formula obtained these five new values by subtracting 50
Variable error= VE=~~ (x; - M)2 I n (2.3) from each of the raw scores in table 2.1.)
where x; and n are defined as in the previous Using Constant Error and Variable Error An
example. The Mis the subject's average move- additional aspect of error scores is important
ment, measured in the same units as the scores from the point of view not only of research but
for the task, so that for this example the M has also of practical application. Compare two rifle
the value of 99.2 units. To compute the VE for this marksmen: Marksman Ahas a large VE and small
subject, use table 2.1 once again. Notice that the CE, whereas marksman B has a small VE and
formula indicates first to compute the difference large CE. This situation was described years ago
between the performance score and the subject's by Chapanis (1951) and is illustrated in figure
own mean (M), so the first step is to compute the 2.4 (see "The Relative Importance of Constant
subject's M. As noted in the previous section, the and Variable Errors"). Which marksman, A or B,
computed M for these trials was 99.2 units. Now, appears to be the more skilled? 1
the values in column D of table 2.1 represent the The study of motor learning will show that
differences between each of the scores on the the measure of error that is most sensitive to the
trials and 99.2 (that is, the difference between each effects of practice is consistency (VE); bias (CE)
individual score and the mean of all that person's often changes quickly in the first several trials and
scores). For example, 93.0 - 99.2 equals -6.2, the remains near zero thereafter, even after years of
first entry in column D. Since the deviations from practice. There are some situations, however, in
the mean for each of the individual trials must which CE is preferred to VE; but these are special-
cancel each other out, by definition the sum of the ized applications. Thus, these two measures of
values in column D must equal zero. So, the next error, CE and VE, seem to represent two distinct
instruction from the formula is to square each of aspects of performance-bias and variability,
the values in column D, and these squared values respectively. But sometimes it is more desirable
 Methodology for Studying Motor Performance 29

THE RELATIVE IMPORTANCE

OF CONSTANT AND VARIABLE ERRORS

Alphonse Chapanis was one of the pioneers in the emergence of human factors research. His
analysis of errors in movement is as important today as it was over a half century ago.
"Having defined constant and variable errors, we might ask: Which is the more important? Let
us return for a moment to the target patterns shot by the two riflemen [figure 2.4]. At first glance,
you might say that B is a very inaccurate shooter. And yet any rifleman will tell you that this is
not the case at all. B is a much better shooter than A The reason is this: The large constant
error in the trial shots fired by B can be compensated for very easily by simple adjustments in
his sights. With suitable corrections in elevation and windage, rifleman B will turn in a perfect
score. In rifle shooting, then, constant errors are not the important ones, because they can be
very easily adjusted for by changing the position of the sights on the gun. The really important
errors are the variable errors. No correction of the sights on P.s gun will make all of his shots
fall in the center. He is inherently much too variable" (Chapanis, 1951, p. 1187).

Marksman A Marksman B

FIGURE 2.4 Distribution of rifle shots. Marksman A has a small constant error (CE) and large vari-
able error (VE). Marksman B has a large CE bias, but a small VE.
Reprinted, by permission, from A. Chapanis, 1951, "Theory and methods for analyzing errors in man-machine systems," Annals of the New York
Academy of Sciences 51: 1181.

to have a single measure of "overall error" that E2 = VE2 + CE 2 (2.4)

combines both of these performance indicators E can also be computed directly from the
rather than using separate measures of accuracy formula:
and variability.
Total variability= E = ~ ~ (x; - T)2 I n (2.5)
Total Variability (E)-Computation The total
variability around a target (or error) for a set where X;, T, and n are defined as before. To apply
of movements (labeled E by Henry [1975] and the formula, we can use table 2.1 again. Notice
sometimes called root mean square error) can be that the major "ingredient" is the difference
thought of as the measure of "overall error" (see between the score and the target (x; - T), and
also Chapanis, 1951). E can be defined as the this difference (with the sign included) is given
square root of the sum of VE 2 and CE2, or in the in table 2.1, column C-the same values used to
following formula when expressed as E2: compute CE. Next, square each of these values;
30 Motor Control and Learning

these squared values are given in column F. The Interpretation of Absolute Error In providing
summation sign then says to add the squared an interpretation of the AE, it will be helpful to
values, and the sum (equal to 100) is given at the consider the ways in which it is similar to E. First,
bottom of column F. Next, divide by n (5), which notice that the numerator is essentially the same
results in a value of 20; then find the square root. for the two statistics, each numerator having a
The final value for E is fiQ, or 4.47. difference between the obtained score (x) and the
Interpretation of Total Variability The total target (T) as the major "ingredient." Second, the
variability, E, is the total amount of "spread" of xi values for E and AE (4.47 and 4.0, respectively)
the movements about the target, so it represents are very similar; the two values will be equal only
an overall measure of how successful the subject in special circumstances but will be very close
was in achieving the target. The key to under- in most situations. Third, both of the formulas
standing this formula is the expression in the involve methods for eliminating the sign of the
numerator (xi - T)2. Eis based on the sum of a difference between the score and the target; for
group of squared differences, where each differ- the AE, the method is to take the absolute value,
ence is the amount by which the subject missed while for E, the method is to square the values
the target. This contrasts with VE, where the in the numerator.
numerator (xi - M)2 represented the deviations Absolute Error Versus Total Variability The
from the subject's own average, which is not neces- AE is a very "logical" measure to use to describe
sarily equal to the target. In cases when CE is close the subject's overall accuracy in a task because it
to zero (i.e., the mean of the trials approximates is sensitive to the extent to which the subject was
the aimed-for target [Mand Tare nearly equal]), "off target." It was used far more commonly than
then E and VE come to represent very similar E in the early research and for many different
aspects of the subject's performance. But, in cases applications. A controversy, however, has arisen
when the CE is very different than the target, then about the use of AE (Schutz & Roy, 1973). The
E will represent the combination of this error in mathematical properties of AE have been shown
bias plus the variability about the CE (i.e., E is a to be a complex combination of CE (accuracy or
combination of CE and VE). bias) and VE (variability), and it is difficult to
Absolute Error (AE)-Computation A statistic be certain of the relative contribution of each.
closely related to the total variability (E) is abso- Because of the precise relation among E, CE, and
lute error (AE), which can also be thought of as VE (namely, E2 = VE 2 + CE2), Eis always an exact
a measure of overall accuracy in performance. It combination of the variability and bias, and thus
is the average absolute deviation (without regard is preferred to AE (Henry, 1975). The tendency
to direction, or sign) between the subject's move- today, when a researcher wishes to present a
ments and the target, and its formula is as follows: combined measure of accuracy and variability,
is to prefer E, for two reasons: First, E measures
Absolute error= AE = 2, IX; - TI In (2.6) essentially the same -component of movement
where X;, T, and n are defined as before. The as AE, and second, E is more easily interpreted
important difference here is the presence of the since it represents a simple combination of CE
vertical bars (I I), which are the symbol for and VE. However, we will use AE a great deal
"absolute value of" and mean that we should take in this text, because much of the earlier research
away the sign of the difference before summing. reported only this measure.
To compute AE, refer again to table 2.1. The Absolute Constant Error (I CE I) One final
first step is to compute the values for the numera- measure of accuracy is merely a transformation
tor terms, and this is done in column G, headed of constant error, CE:
IX; - TI · Notice that the value in this column is
Absolute constant error = ICE I (2.7)
the same as the corresponding value in column
C, except for the sign. The summation sign ~ is Thus, for a single subject, the absolute constant
an instruction to add up the values from each of error ( ICE I) is just the absolute value of the CE,
then trials (recall that n = 5 in this example), and which is simply 0.80 in the present example. Be
the sum is given at the bottom of column Gas 20. careful to note, however, that ICE I is not cal-
The next step is to divide by the number of trials culated in the same way as AE. For AE we take
included (n = 5), and so the final answer is 4.0. the sign away immediately after calculating the
 Methodology for Studying Motor Performance 31

difference between a single score and the target, direction, the measures of overall error (E and
prior to summation. For ICE I the sign is not taken AE) tend to represent the magnitude of the bias,
away until after the average over a series of trials and VE alone represents the variability.
has been calculated. Now consider subject 2 in table 2.2. This subject
Interpretation of Absolute Constant Error The has the same spread of outcomes as subject 1, but
situation sometimes arises (e.g., Newell, 1976a) in with much less bias. We obtained these scores
which approximately half the subjects in a group by adding 15 to each of subject l's scores to get
have positive CE scores while the other half have subject 2's scores: 95, 105, 100, 97, 102. Table 2.2
negative CE scores. If one characterizes the aver- gives the error measures of this set of scores. Now
age bias for this group of subjects by calculating notice that the measures of overall error tend to
a group mean score (the average score of all the be very close to the VE, all around 3. The CE (and
subjects), the positive and negative signs will ICE I), however, is now nearly zero. Here the
tend to cancel" each other out. This could give
/1 measures of overall error (E and AE) represent
rise to the misleading conclusion that the average the variability of the movements (VE), exactly the
bias for the group was nearly zero when in fact opposite of the situation with subject 1.
every subject in the group might have shown a Therefore, when CE is large in either direction,
bias of considerable size. In this case, it is useful the measures of overall error (E and AE) tend
to also compute ICE I, which tells the researcher to represent the amount of bias in the scores.
the mean amount of bias for the group of sub- When CE is small, E and AE tend to represent the
jects without regard to its direction, and which amount of variability (VE) in the scores. When
will not fall prey to this canceling" effect of the
/1 CE is intermediate in value (with some bias, but
positive and negative CEs. In cases in which with scores falling on both sides of the target), the
group data have been the focus of a researcher's measures of overall error represent an unknown
experiment, there has been a trend to report ICE I combination of bias and variability. This should
(as the measure of accuracy) along with VE (as make clear why simply examining overall error
the variability score) (see Schutz, 1977, for more statistics does not provide a very complete picture
statistical details). of performance.
Relationships Among the Error Scores One Other Measures of Accuracy There are many
way to evaluate the relative contributions of the tasks in the motor behavior literature that could
various measures of error is to consider the fol- not be scored so simply. A task for which accu-
lowing cases. At one extreme, when CE is very racy is important is the tracking task; in this case,
large (an extreme is the situation in which all the performance is ongoing, thus preventing the
person's movements lie on one side of the target), computation of a discrete performance error. A
then the absolute error (AE), the total variability commonly-used tracking task is the pursuit rotor,
(E), and the constant error (CE) all tend to mea- shown in figure 2.5a. There are many varieties of
sure the same component of performance-the pursuit tracking tasks, and figure 2.5a illustrates
bias or directional deviations of the errors. In the just one type. Here, a target (e.g., a small circle)
following case, the target is again 100, but subject is embedded in the surface of a turntable-like
1 produces five movements, 80, 90, 85, 82, and structure that rotates at various speeds. The
87-each smaller than the target. Table 2.2 gives subject holds a stylus in the preferred hand and
the measures of error for this subject. Notice that attempts to keep its tip in contact with the target
the statistics E, CE, AE, and ICE I are all around as the turntable rotates. A trial might last from
15, but that the VE is very much lower at 3.54. 10 s to 1 min, and performance is scored in terms
This suggests that when the CE is large in either of the amount of time in the trial that the subject

Subject 1
Subject 2 3.55 -0.2 3.54 0.2
32 Motor Control and Learning

maintained contact with the target. The perfor- scored in terms of the number of correct matches
mance measure is usually called time on target that can be achieved in a trial of fixed duration.
(TOT) and can range from zero (if the subject The stabilometer is shown in figure 2.Sc. The
never touched the target) up to a value equal to standing subject attempts to keep an unstable
the duration of a trial (if the subject was always platform level; the scores denote either time in
in contact with the target). But notice that time balance or the number of times the platform edge
on target can represent a complex combination touches the floor (indicating extreme loss of bal-
of bias (if the subject is consistently behind the ance) during a trial of perhaps 30 s. The number
target, for example) and variability (if the subject of times a stylus touches the sides of a maze is
is alternately ahead of and behind the target). another example of this kind of measure. Figure
Other common variations of tracking tasks are 2.Sd shows the two-hand coordination task, in which
shown in figure 2.5. The Mashburn task is shown the subject attempts to follow a target by moving
in figure 2.Sb. It was designed to simulate certain a pointer with two crank handles. One handle
features of airplane controls. The control panel controls the right-left movement, and the other
contains three double rows of lights. One row of controls the forward-backward movement (as in
each pair is controlled by the movements of the the "Etch-a-Sketch" toy). The score is again TOT,
subject (left-right and forward-backward move- or the amount of time in a trial that the subject was
ments of the stick, and right-left movements of the over the target. All these measures, including TOT,
pedals), while the other row of each pair is con- are measures of overall error, and they tend to
trolled by the experimenter. The subject attempts confound the bias with variability in performance.
to match the experimenter-determined lights with In each of the foregoing examples, the experi-
appropriate movements of the controls. The task is menter does not need to keep a record of the sub-

Mashburn task

a Pursuit rotor b

coordination task
c Stabilometer d

FIGURE 2.5 Four movement tasks frequently used in motor behavior research: (a) pursuit rotor, (b) Mashburn task, (c)
stabilometer, and (d) two-hand coordination task.
 Methodology for Studying Motor Performance 33

ject's actual behavior. However, there are other can accomplish more in a given amount of time,
tracking tasks for which a continuous record of or who can produce a given amount of behavior
the movements of the subject and the target is in less time, is the more skillful. These two kinds
kept. From these data a measure of overall accu- of measures are essentially the same, since a time
racy, the root mean square (RMS) error, can be measure (time/unit) can easily be converted to a
computed. (Notice that RMS error is analogous speed measure by taking the reciprocal; that is, 1
to E, described earlier. Both are root mean square I (time I unit)= units I time, which is a measure
deviations of the behavior from some target, com- of speed. Both speed and time measures have
puted on successive "trials.") Essentially, the RMS been used a great deal in motor behavior research.
error is based on taking small "slices" of time Reaction time and movement time are common
and measuring the deviation of the subject's line examples, described next.
from the target at each of these times, as shown
Reaction Time Reaction time (commonly abbre-
in figure 2.6. Depending on the capacity of the
viated as RT) is a measure of the time from the
recording system, these slices can be taken every
arrival of a suddenly presented and unanticipated
few milliseconds or so over the entire course of
signal to the beginning of the response to it. In
a 20 s trial, providing many measures of error.
the RT paradigm shown in figure 2.7, the subject
To compute the RMS error, square each of these
is given a warning signal, and after a randomly
deviations from the track, add up the squared
deviations, divide by the number of measures, determined foreperiod (perhaps ranging from 1 to
and then take the square root, giving a measure 5 s), the stimulus is presented. Using the variable
of the amount of deviation over the course of foreperiod represents an attempt to prevent the
the trial. Root mean square error in each of these subject from anticipating when the stimulus will
cases represents essentially (but not exactly) the arrive (temporal anticipation). Sometimes" catch
area between the subject's movements and the trials" are used, in which the stimulus is not
target, as shown by the shaded portions of figure presented at all; this allows the experimenter to
2.6. As with TOT, the RMS error is a measure of "catch" a subject who is anticipating, and thus its
overall error and is sensitive to both the bias and use tends to prevent anticipation. The introduc-
the variability in performing. tion of catch trials, given randomly perhaps on
15% of the total number of trials in an experiment,
Measures of Time and Speed improves the experimental control. Also, subjects
The second fundamental way of assessing skills can be prevented from anticipating which move-
is by measures of time and speed. Basic to this ment to make (i.e., spatial or event anticipation)
idea is the assumption that the performer who through the use of two or more choices, so that the

c:
0
:;::;
'iii
0
c..

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Time (s)

FIGURE 2.6 Hypothetical record from a tracking task, showing the basis for computation of root mean square (RMS)
error.
34 Motor Control and Learning

Stimulus Response Response

Warning presented begins ends

Response time

1 - - - - - - RT _ _ _ ____,,_,_..,,___ _ __ MT --------c..-

_.___ Premotor ---)!- -<Oii(- Motor ___..._

RT RT

EMG

FIGURE 2.7 Critical events involved in the reaction-time (RT) paradigm.

proper response is signaled by the stimulus itself the signal to the first change in EMG is termed
(e.g., red light means move left, blue light means "premotor RT" and is thought to represent central
move right); this is termed the choice-RT method processes involved in making the response (e.g.,
(when only one option or "choice" is available, perception, decisions). The interval from the first
this is termed a simple-RT method). change in EMG to finger movement is termed
Reaction-time measures are common in many "motor RT" and represents processes associated
sport settings; an example is the interval between with the musculature itself. Such methods are
the starter's gun and the first movement in a useful in gaining further information about the
swimming race. This is an example of a simple-RT effect of an independent variable on RT (e.g.,
task in which there is only one response to make. Fischman, 1984).
The starter varies the time between the "ready" Reaction-time measures are very common in
command and the auditory signal to start in order research on skills, for two basic reasons. First,
to reduce temporal anticipation. The decision of RT measures are components of real-life tasks
a soccer goalie to dive to the left or to the right (e.g., sprint starts), so they often have high face
in a penalty kick is an example of a choice-RT validity. Amore imp0rtant reason (which we will
task in sport. Reaction-time measures are also amplify in chapters 3 and 4) is that RT presumably
used extensively in the laboratory as measures measures the time taken for mental events, such
of information-processing speed (see chapter 3). as stimulus processing, decision making, and
One variation of the RT method is to partition movement programming. These two motivations
the latency of the response initiation into "cen- for using RT measures differ considerably. In the
tral" and "peripheral" components (Weiss, 1965). first case, RT is a measure studied for its own
The bottom of figure 2.7 shows a hypothetical sake; in the second case, RT allows the researcher
electromyographic (EMG) trace taken from a to understand the kinds of mental processes that
muscle involved in the movement to be made lead to movement (e.g., Posner, 1978). Regardless
(EMG indicates the electrical activity in a muscle). of the motivation, the measurement of RT is the
The EMG is mostly silent during a substantial same.
part of the RT, indicating that the command to
move the finger, which is initiated in the brain, Movement Time Movement time (commonly
has not yet reached the finger musculature. The abbreviated as MT) is usually defined as the
muscle is activated late in the RT, but no move- interval from the initiation of the response (which
ment occurs for 40 to 80 ms. The interval from defines the end of the RT) to the completion of
 Methodology for Studying Motor Performance 35

the movement (figure 2.7). Clearly, MT can be just via instructions, so that accuracy can be assessed
about any value, ranging from a few milliseconds (e.g., Quinn, Schmidt, Zelaznik, Hawkins, &
for a very quick movement to several weeks if McFarquhar, 1980; Schmidt, Zelaznik, Hawkins,
the movement being studied is jogging from Frank, & Quinn, 1979).
Los Angeles to Chicago. Some sport skills have
minimal MT as a goal (e.g., time to run 100 m, Measures of Movement Magnitude
or time for a quarterback to "set up" for a pass), A third way of measuring skills is by the mag-
and MT is used a great deal in skills research as nitude of behavior that the performer produces,
a result of its overall external validity in these such as the distance that a discus was thrown
practical settings. Sometimes researchers use RT or the amount of weight that was lifted. These
and MT tasks together in the same performance, measures have particularly important applica-
as in requiring the subject to lift a finger from a tions to sport settings, as many sports use such
key and move to a button as quickly as possible measures as the primary determinants of success
after a stimulus. in the activity. One example of these measures in
The sum of RT and MT is termed response time research is the Bachman (1961) ladder climb task
(figure 2.7). Research has consistently shown that (figure 2.8), which uses a specially constructed
very different processes or abilities are required ladder. At the beginning of, say, a 30 s trial,
in reacting quickly as opposed to moving quickly the subject begins to climb the (unsupported)
once the reaction is over, and this has justified ladder without skipping rungs until balance is
separating response time into RT and MT. What lost and the subject topples over. The subject
is frequently called "brake reaction time" in the quickly returns to the starting position and begins
field of automobile accident analysis is really climbing again, and so on until the trial has been
response time, because it consists of the time used completed. The score is the number of rungs
to initiate the foot movement from the accelerator accumulated in a given trial. A variant of the task
pedal plus the time required to move the foot to is to climb as high as possible in a single attempt.
the brake pedal and press it. The ski simulator (figure 2.9) is a task in which
Often the degree of accuracy in the task must large rubber bands keep a platform centered. The
be taken into account when measures of speed subject's task is to displace the platform as far
are used. A well-known phenomenon in motor as possible from side to side using whole-body
behavior is the speed-accuracy trade-off, mean- movements.
ing simply that when performers attempt to do
something more quickly, they typically do it less
accurately. In most measures of speed, therefore,
accuracy requirements are kept to a minimum
so that speeding up the movement (which is the
major goal for the subject) does not seriously
affect accuracy. In some situations, though, mea-
sures of speed are confounded with measures of
accuracy, and the speed with which the subject
performs is dependent on the amount of error
she is willing to make or the amount of error
the experimenter will tolerate. Such trade-offs
are particularly troublesome for experimenters,
because it is not always clear to subjects how
much error will be tolerated, and experimenters
are unsure about how to interpret an indepen-
dent variable that produces increases in speed
but decreases in accuracy. One solution to this
problem is to hold accuracy constant by various
experimental techniques so that a single depen-
dent variable of speed can be assessed (e.g., Fitts,
1954). Another solution is to hold speed constant, FIGURE 2.8 The Bachman ladder climb task.
36 Motor Control and Learning

FIGURE 2.9 Ski simulator being used by Gaby Wulf.

Reprinted, by permission, from G. Wulf, 200~ Attention and motor ski/I learning (Champaign, IL: Human Kinetics), 9.

While it might seem that tasks with goals of measurement will be sensitive to differences in
maximum movement magnitude are consider- skill among individuals or to differences in skill
ably different from those requiring speed or caused by some independent variable. Generally
accuracy, their fundamental determinants may these situations involve tasks for which differ-
not be all that different. At first glance, producing ences in performance are not evident because
maximum movement magnitude would seem the tasks are well learned (driving a car down
simply to be a matter of generating more force. an open road), or involve tasks that do not "tax"
But these skills certainly require precise timing the motor system very much because they are so
of the forceful contractions and accurate coordi- simple (drinking a glass of water without spill-
nation among the various participating limbs. ing). How are skills assessed in such situations?
Therefore these precise muscular activities might One method is to use some measure of critical
be essentially the same as those required in tasks incidents. In the driving example, accident rates
that seem to necessitate only accuracy. Of course, from statistical databases for a particular kind
inconsistency (in terms of VE) in these processes of car, or for certain types of people (e.g., drunk
will degrade performance, and such inconsis- drivers), might be used as measures of skill; with
tency is probably related to the VEs that are seen pilots, "near misses" (midair near-collisions)
in the outcomes of simpler tasks. might be used. But these techniques are difficult
to utilize in the laboratory because (fortunately)
Measures of Secondary Tasks such critical events occur so infrequently. Thus,
There are instances in both practical and research they are far more useful for groups of people
settings in which none of these basic methods of taken over relatively long stretches of time.
 Methodology for Studying Motor Performance 37

Another useful technique is to employ perfor- (as a measure of overall effort), or even EMC
mance on some sort of secondary task, performed from the muscles of the forehead (to indicate the
simultaneously with the primary task, as a mea- level of concentration) can be used, depending
sure of the skill in the primary task. For example, on the particular situation. Neurophysiological
Brown (1962) used a verbal task in which the indi- techniques, such as the recording of event-related
vidual was presented with eight-digit numbers at potentials from the scalp, are also used. One such
4 s intervals. Each number contained seven of the potential has been called P300, as it occurs as a
same digits as the previous one, and the subject's positive voltage about 300 ms after the presenta-
task was to detect the digit that was different and tion of a stimulus (Duncan-Johnson & Donchin,
to provide a response. Errors were counted as 1982). The latency of the P300 and the magnitude
omitted responses, incorrect responses, and late of its amplitude appear to be highly correlated
responses. In this experiment, Brown obtained with RT, thus providing an unobtrusive way to
measures of performance when subjects per- measure processing activities (e.g., Radlo, Janelle,
formed only this verbal-numerical task, and also Barba, & Frehlich, 2001). In all these cases, the
when subjects performed this task while driving secondary measures become the focus of the
under various conditions. The mean percentage investigator, especially when the main task does
of correct responses when the verbal task was not provide sensitive measures of the subject's
performed alone was 90.6%. When the task was performance.
performed during driving in quiet residential A variation of this technique is to use a sec-
areas, the mean percentage dropped to 83.8%. ondary task as a distracter in order to increase
And when the task was performed in heavy traf- the overall "load" on the performer. Normally,
fic conditions, the percentage again dropped, to fatigue may not have any obvious effect on the
79.5%. Yet it was very difficult to see any differ- well-learned task of driving. However, if the
ences in vehicle control in light and heavy traffic, driver is required to perform a simultaneous
largely because driving is so well learned by most mental arithmetic task at a predetermined level,
people. This secondary task provided evidence then large differences between fatigued and
about the difficulty of the driving conditions rested driving may be seen. In this situation,
when the driving task itself would not have pro- unlike the others, the major interest is in the per-
vided such a measure. formance of the main task, and the secondary task
Some experimenters have used RT tasks has increased the sensitivity of the measurement
(so-called probe tasks-e.g., Kerr, 1975) inserted system for the main task. However, care must be
during a performance of the primary task. Others taken with these techniques, as Brown (1962) has
have used measures of finger tapping regularity shown. When truck drivers were fatigued, their
(e.g., Michon, 1966) as the secondary task. In performance on the secondary digit detection
these latter cases, the implication is that these task actually improved, suggesting that they were
tasks require some of the subject's limited capacity devoting less capacity to the driving task and
to process information; presenting the second-
overcompensating by devoting more capacity to
ary task simultaneously with the primary task
the secondary task. While these techniques can
necessitates the use of some of this capacity, and
be somewhat tricky to use, they have served well
lowers the performance on the secondary task in
in a number of situations.
relation to the amount of capacity demanded by
the primary task. (See chapter 4 for more on this Describing Characteristics
assumption and general method.)
Rather than a task, secondary physiological mea- of Movements
sures of effort can be used during the performance Countless methods could be employed to
of a main task. One technique is to measure pupil describe movements, depending on the charac-
diameter by one of various recording methods. teristics of the movement that are of interest to the
Pupil dilation is associated with circumstances observer. At the most fundamental level, one can
in which effort, arousal, or information process- use verbal descriptors to characterize movement.
ing is demanded (e.g., Beatty & Wagoner, 1978; For example, movements have been described in
Kahneman, 1973). Similarly, measures of heart dance notation created by Laban (1956) and in
rate, heart rate variability, oxygen consumption terms of units of work behavior called "Therbligs"
38 Motor Control and Learning

in early industrial time-and-motion studies (Gil- 1979) using series of still photos (see "The Move-
breth, 1909; note that "Therblig" is the author's ment Photographs of Eadweard Muybridge,"
name spelled backward, almost). Another way is figure 2.10), or with videotape or digital images.
to illustrate movement with photographs, as was Such methods are of some use in describing or
done over a century ago by Muybridge (1887, illustrating the basic forms of movement, but

The Movement Photographs

of Eadweard Muybridge
In 1887 Eadweard Muybridge published what probably was the first photographic analysis of human
and animal locomotion. In some series, Muybridge used 36 separate shutters, arranged so that
three different angles could be photographed simultaneously, positioned at the side, front, and rear
of the subject. At each angle a series of 12 shutters was electrically arranged to open in a timed
sequence after a signal had been sent to the first shutter. Muybridge determined the length of the
time intervals on the basis of the speed at which the subject performed the activity (e.g., picking up
an object vs. running at full speed). These methods resulted in a sequence of still photographs that
produced simulated motion, which one could study further by examining the changes in the action
as it evolved over time. In all, Muybridge published 781 plates, each plate a series of photographs of
the subject. The subjects were both. humans and animals. Humans of all ages, males and females,
and of many different body types, performed actions such as walking, running, jumping, skipping,
lifting objects, and hammering. Many of the humans were photographed in the nude, which provided
rather explicit details of body actions. Animals from a wide range of species were photographed in
various natural and contrived settings. It is rumored that Muybridge settled a bet concerning whether
or not all of a galloping horse's feet are ever off the ground at the same time. According to his photos,
they are. Muybridge's series of still-sequence photographs remains as a magnificent legacy in the
history of human and animal motion analysis.

FIGURE 2.10 A famous series of Muybridge photos, clearly revealing a period of time when the horse has no
contact with the ground.
 Methodology for Studying Motor Performance 39

have limited value in detailed assessments of or the entire body. The locations of various parts
performance. Improvements in photographic of the body during the movement, the angles of
techniques occurred early in the 20th century; the various joints, and the time relations between
some were pioneered by Bernstein (Bernstein & the movement in one joint and the movement in
Popova, 1930/2003). In this section we focus on another are examples of the many ways move-
kinematics, a branch of mechanics in physics that ment kinematics can be recorded.
involves the description of "pure" motion without
regard for the forces and masses that produced the Location Perhaps the most common of the kine-
motion. The devices that can be used to collect this matic methods entails recording the locations of
information also vary widely, and we will describe the limbs during a movement. An early example
some of the more common ones. of research using movement kinematics is that
of Lindahl (1945; see "Lindahl's Study Using
Movement Kinematics Movement Kinematics"). Early in the history
As applied to movement behavior, kinematic of motor behavior and biomechanics, research-
measures describe the movement of the limbs ers used high-speed cinematography to record

Lindahl's Study Using Movement Kinematics

One of the early investigations using movement kinematics as a research tool was published by
Lawrence Lindahl in 1945. His study concerned a very practical problem: an industrial task in which
a factory worker operated a machine that cut very thin slices off tungsten rods. The key aspect of
the job involved using a foot pedal action that moved an unseen cutting wheel through the tungsten
rod. Success at the task required carefully coordinated displacement-time foot actions. Failure
to coordinate these actions resulted in wasted material (improperly cut discs) and damage to the
equipment (broken wheels). Skilled machine operators made very stereotypical actions of the
foot, which Lindahl captured nicely using a writing apparatus that recorded the foot displacements
over time (see figure 2.11 ). These recordings were then used for comparison with the recordings
that beginning and less skilled workers produced in their disc-cutting performances. This type of
recording, using kinematics as a means to provide objective feedback and goal-related movement
information (so-called kinematic feedback), predated modern investigations by about 40 years (see
discussion in chapter 12).

Machine table

Paper tape

FIGURE 2.11 Experimental setup used by Lindahl (1945) to record kinematics of the disc-cutting task.
40 Motor Control and Learning

movements. Often the subject being filmed wore from the output of a device called a potentio-
tape markers over certain landmarks (e.g., the meter that signals angular position. This trace,
wrist or ankle) so that the locations of these body read from left to right, represents an arm move-
parts could be studied frame by frame. These ment of about 17 cm. The movement began at the
positions on successive frames were separated time when the trace left the horizontal axis. The
by nearly fixed periods of time, so a graph of the largest amplitude (about 20 cm) was achieved
position of the landmark against time could be about 125 ms after the movement started, and then
generated from the data. the limb stabilized its position at the final location.
Figure 2.12 shows an example of this type of But examining the location of a limb in space
graph, taken from Wadman, Denier van der Gon, may mask some of the more subtle factors that
Geuze, and Mol (1979). For now, consider only determine its control. For this reason motor
trace a, which represents position; it is derived behavior researchers often examine variables that

0.3

:[ 0.2
c:
~
"(ij
0 0.1
D..

a O
3.0

-!!! 2.0
.§_
~
"ij
0 1.0
Qi
>
0

+50.0

N'
i c:
0
~.
Q)
Qi
0

0
0
<(

-50.0

0.0 0.1 0.2 0.3 0.4

c Time (s)

FIGURE 2.12 Position, velocity, and acceleration traces representing a rapid 17 cm elbow extension movement.
Reprinted from W.J. Wadman et al.. 1979, "Control of fast goal-directed arm movements;· Journal of Human Movement Studies 5: 5. By permission of W.J. Wadman.
 Methodology for Studying Motor Performance Ill! 41

can be derived from location information-veloc- actions of one body part are controlled together
ity and acceleration. with the movements of another body part. Many
Velocity Trace b in figure 2.12 is a record of types of coordination exist, such as the coordina-
velocity of the movement at each moment in time, tion of two or more joints in one limb (as when
placed on the same time scale as the positional one is reaching for a cup) or of different limbs
trace for easy comparison. The velocity trace was simultaneously (as of the arms and legs during
determined by computer, which read in the posi- walking), or even more subtle coordinations (such
tion information from the potentiometer and then as the movements of the eye and head while
calculated the slope or inclination of the line at one is reaching for a glass in the cupboard). For
each moment. The slopes (called derivatives) of movements that are oscillatory, one measure of
the positions at each moment in time represent the coordination is to describe the temporal phasing
velocities at corresponding moments, indicating between the two body parts.
the rate of change in position. Then this information Consider the simple action of tapping two
is output onto the same record as the position fingers on a table. Suppose we plotted the dis-
information. Such a trace is useful here in showing placement records of the up-and-down tapping
that the maximum velocity (Vm) was about 2.7 ml s cycles of the right finger along the abscissa and
and that the achievement of maximum velocity the cycles of the left finger along the ordinate.
occurred at about 75 ms through the movement. Plotted separately, each would be represented as
Also shown are a gradual increase in velocity until a back-and-forth, overlapping straight line along
the peak velocity (the midpoint of the movement) its respective axis. However, to assess how these
is reached and then a decline toward the end. Such two fingers are coordinated, the time records
a trace gives a more complete description of the of one finger can be plotted relative to the time
movement than does positional information alone. records of the other finger. At any point in time,
the position of the left hand and the position of
Acceleration Trace c in figure 2.12 is a record the right hand are represented as a single data
of the acceleration at each moment of time. This point on the graph. When the points are combined
record was also obtained by the computer, which over time they produce one continuous trace, as
calculated the slope or inclination of the velocity illustrated in the two examples in figure 2.13 (see
curve at each moment. The slopes of the velocity also Winstein & Garfinkel, 1989). In figure 2.13a,
curve at each moment yield the accelerations and the two fingers tap the table at approximately
represent the rate of change in velocity. This output the same time, are at maximum height above the
is plotted along with the other two traces on the table (in the "up" position on the graph) at
same time scale. Initial acceleration lasts about about the same time, and seem to be moving
100 ms until the acceleration trace returns to zero. within their respective cycles at about the same
Then there is a deceleration (a negative accelera- time-the coordinated motions are simultaneous.
tion trace) that lasts for about the same length of The example in figure ~.13b illustrates alternate
time. Also, the peak velocity of the movement is tapping-one finger taps the table at about the
achieved at the point at which the acceleration same time that the other finger reaches maximum
changes to deceleration (where the acceleration height, then vice versa. This figure illustrates
curve crosses the zero baseline). just two of the types of temporal coordination
These kinematic variables, with simultaneous patterns that exist between oscillating effectors
recording of position, velocity, and accelera- (see chapter 8).
tion as a function of time, provide a reasonably One can also obtain a quantitative measure
complete picture of these movements. Scientists of temporal coordination by considering the
often search for changes in these kinematic displacements of each cycle over time. This is
variables when certain independent variables represented in figure 2.14 in two ways. In figure
are changed-for instance, instructions to the 2.l4a, displacement of one finger is plotted over
subject or the size of a target to which the person time-position A represents the finger at the time
is moving. Examples of this kind of research are of a tap, position Bis about halfway up, position
provided later in the text. C represents the "up" point, and position D is
Coordination Kinematics In chapter 8 we halfway back down again. Figure 2.14b repre-
will focus on movement coordination-how the sents the same data by plotting these one-finger
42 Motor Control and Learning

Up

Tap

Tap Up Tap Up
a Right finger b Right finger

FIGURE 2.13 Sample displacement plots of two fingers moving simultaneously. (a) In-phase coordination; (b) anti-phase
coordination.

displacements against their own velocities (called phase angle (<I>) indicating the progress through
a phase-plane representation). Now positions A a cycle, or a circle, containing 360°.
and C represent zero velocity, and positions B Since the phase planes of each finger can be
and Dare the maximum upward and downward determined independently, the measure of coor-
velocities, respectively. The value of phase-plane dination is simply the difference between the
representations is that the position of each finger phase angle for the left finger (<PL) and the phase
at any point within its cycle can be described as a angle for right finger (<PR). This measure is called

c:
0
E
Ill
~

Time

Position

a b

FIGURE 2.14 Repetitive movement of a finger: displacement (a) relative to time and (b) relative to its velocity.
Figure 2.14a Adapted from Kelso et al. 1985.
Figure 2.14b Adapted from Burgess-Limerick, Abernethy, and Neal 1991.
 Methodology for Studying Motor Performance 43

relative phase, because the measure represents the Goniometers can be wired with potentiometers
position of one finger within its cycle relative to that send voltage information (proportional to
the position of the other finger within its cycle. the joint angle); this can be accumulated and
From figures 2.13a and 2.14, it is clear that at analyzed by a computer. Potentiometers are also
any one time, the two phase angles are about the used in many other direct measurement devices,
same. That is, when both fingers are down," each
/1
such as a computer mouse, that signal changes in
phase angle is 0°; when both fingers are "up," position when physically moved by the subject.
each phase angle is 180°, and so on. Thus, the A graphics tablet is another device that records
relative phase (<I\ - <l>R) will always be close to the changes in contact positions with the surface.
0° whenever a sample is measured. This simul- Muybridge (see "The Movement Photographs
taneous pattern of coordination is often referred of Eadweard Muybridge" on p. 38) is often cred-
to as moving in-phase. However, for figure 2.13b, ited with initiating the analysis of human move-
notice that the temporal phasing of one finger is ment through imaging techniques. High-speed
always nearly exactly opposite that of the other cinematography introduced a way to capture
finger. That is, when one phase angle is at 0°, the images of moving limbs many times each second.
other is at 180°; then when the first is at 180°, the However, frame-by-frame analysis methods were
other has reached 360° (or 0°). The relative phase tedious ways of examining changes in locations
for this series of estimates produces an average over time (Bernstein & Popova, 1930/2003).
relative phase of about 180°, which is sometimes Fortunately, technology has introduced more
referred to as moving in an anti-phase pattern of automated ways of performing these analyses.
coordination. Such instruments have relieved the scientist of the
But notice also that there is some variability in time-consuming job of reading locations from each
the plots presented in figure 2.13. Thus, research- frame of film, a method that also made the cost of
ers often calculate the standard deviation of these such analysis systems prohibitive for many years.
relative-phase samples in order to supplement the Computer systems that analyze such data are now
description of a pattern's average relative phase more powerful and much cheaper, and the use of
with an estimate of its stability. The measure of these measurement techniques is now the norm.
relative-phase variability is determined in a way The most common imaging devices use video
similar to that for the measures of VE, discussed and optoelectric methods. Video methods were
earlier in the chapter. Both the mean and standard straightforward, as movements could be cap-
deviation measures of relative phase will be used tured on relatively inexpensive videotape using
to describe important features of coordination in VHS and 8 mm formats. Typically, the subject
chapter 8. was recorded on videotape wearing pieces of
tape or fluorescent "markers." The locations of
Movement Measurement Devices these markers were later digitized for analysis.
Measurement recording systems, used to collect For optoelectric method_s, tiny light bulbs (called
data about human movement, have undergone light-emitting diodes or LEDs) are attached to
many changes over the years. Just as "moving the subject on various body parts. Light-sensing
picture" technology evolved from the efforts devices then detect and record automatically the
of Muybridge and others to capture on film the locations of the LEDs during the movement. As
moment-to-moment changes in body positions, illustrated in figure 2.15, an advantage of these
new developments in measurement technology newer systems is the capability to record infor-
have often been created to satisfy the needs of mation in three dimensions, providing greater
researchers to make recordings with increased information about movements and greater flex-
precision and economy (in terms of time and ibility in the types of actions that can be studied
effort). in experiments.
One of the simplest methods to collect body The most recent advances in movement record-
position information is to acquire the information ing technology have combined the use of robotics
directly through various means. For example, a and virtual reality environments (see figure 2.16).
goniometer is a hinged device that, when strapped These systems not only record movement with
to the side of a body joint, physically changes fine precision, but also have great flexibility to
in angle along with changes in the joint angle. provide diverse perceptual inputs. For example,
 FIGURE 2.15 Three-dimensional optoelectric system.
Photo courtesy of Northern Digital Inc.

FIGURE 2.16 Robotic system in a virtual environment.

© SensAble Technologies, Inc. www.sensable.com/documents/images/premium_3.0_6doUarge.jpg

44
 Methodology for Studying Motor Performance 45

the robotic system can induce different types perform a four-phase movement involving
and magnitudes of haptic feedback (information rotation of the right wrist. The record at the top
about touch) during movement. When combined of the figure shows the clockwise (supination)
with virtual reality feedback information, these and counterclockwise (pronation) movements.
robotic systems can simulate the look and feel of The EMGs are from the pronator teres muscle,
complex skills (e.g., surgical techniques) without which acts as the pronator (counterclockwise),
the dangers that might otherwise be present if and the biceps muscle, which acts as the supina-
practice were undertaken in the "real world" tor (clockwise). Once the movement begins, one
(e.g., on a patient). can see marked activity in the various muscles;
the activity is dependent on the particular action
Electromyography
being performed. The pronator teres is the first
Another method for describing movement char- muscle to act, throwing the wrist into pronation;
acteristics is to measure the involvement of a then the pronator is turned off and the biceps
muscle in a movement by recording the electrical acts to brake the action and reverse it; then the
activity associated with its contraction. The sim- pronator brakes and reverses that action, and so
plest method is to attach (with adhesive collars) on. These records describe the temporal patterning
recording electrodes to the skin surface over the of the movement segments. Information about
involved muscle; then, this weak signal from the the intensity of contraction is also provided by
muscle is amplified and recorded on a polygraph the amplitudes in these records, with larger EMG
recorder or computer for later analysis. Occa- amplitudes being generally indicative of larger
sionally subcutaneous electrodes are used; the forces. However, while the relation between
electrode is placed just under the skin but above EMG amplitude and force under static, controlled
the muscle belly. Or a small wire electrode can conditions within a given muscle is good, many
be embedded within the muscle so that electrical situations arise that can degrade this relation,
activity in small portions of the muscle can be so that the amount of force produced is usually
recorded. not accurately reflected by the amount of EMG
A recording using surface electrodes, taken being produced.
from a study by Carter and Shapiro (1984), is A record of transformed EMG activity, taken
shown in figure 2.17. Subjects were asked to during a rapid elbow extension (from Wadman

FIGURE 2.17 A typical electromyogram tracing taken during a movement.

Journal of Neurophysiology by M.C. Carter and D.C. Shapiro. Copyright 1984 by AMERICAN PHYSIOLOGICAL SOCIETY Reproduced with permission of AMERICAN
PHYSIOLOGICAL SOCIETY in the format Textbook via Copyright Clearance Center.
46 mi Motor Control and Learning

et al., 1979), is depicted in figure 2.18. A number other records of kinematic information, so that the
of changes were made in the raw EMG signals changes in the muscle actions can be associated
before they were plotted. First, the EMGs were with the resulting actions of the limbs.
rectified; that is, the negative voltage values were
Measures of Eye Movements
given positive signs so that the resulting record
would be completely positive. (Notice that figure Motor skills involving the use of the upper
2.18 has two such records, with the biceps record limbs constitute a large proportion of our daily
inverted so that the two patterns can be com- activities. How we use our hands and fingers to
pared more easily.) When the EMG is rectified, press the buttons to make a phone call, pick up a
the pattern of electrical activity can be seen more cup from the table, or move a mouse-controlled
readily than with the raw signals shown in figure pointer across a computer monitor depends con-
2.17. Second, these records were averaged for a siderably on our visual system. We use vision to
number of similar movements, mainly so that determine where we want the hand to go and
the important patterns of contractions could be then to update the relative success of that move-
seen over and above the trial-to-trial variations. ment as it approaches the target. Researchers
These patterns are more reliable than are those have used eye movement recording systems in
for a single trial. their investigations to determine where someone
is looking (that is, where the eyes are directed),
for example during an aiming movement of the
hand or in making a left turn in an automobile.
These recording systems can provide accurate
measures of what the subject sees, through two
video imaging techniques. One camera mounted
on the head or helmet of the subject provides an
image of the subject's line of vision. This visual
gaze is then coordinated with an eye-tracking
device that measures the movements of the eyes
by means of corneal reflection. The calibration of
the two recording devices provides an accurate
measure of where the eye is directed during the
performance of a motor task, sometimes referred
to as point-of-gaze information. One caveat here
is that what the person "sees" (i.e., perceives visu-
ally) is not perfectly related to where the eye is
pointed. In some situations, even though the eye
is directed at an object, the object is for various
reasons not actually "seen." The mere fact that a
FIGURE 2.18 Rectified and averaged electromyogram person is looking at an object does not guarantee
signals from the triceps and biceps muscles during a rapid
that he actually perceives it.
elbow extension.
Reprinted from W.J. Wadman et al., 1979, "Control of fast goal-directed arm
Figure 2.19 provides a good example of how
movements," Journal of Human Movement Studies 5: 10. By permission of W.J. the information from eye-tracking devices can be
Wadman.
coordinated with other information to provide
a more informative description of the motor
Such records are useful in that they provide behavior of a subject in a particular task. In the
one kind of description of what the central ner- example in figure 2.19, the subject is moving a
vous system "tells" the muscles to do. In the hand forward to point at a target that is 40 cm
example shown, it appears that the triceps muscle away from the initial "home" location. The top
contracted for about 100 ms; then it turned off half of the figure provides data from the eye-
and the biceps muscle contracted for about 50 tracking measurement system, showing that the
ms; and then the triceps muscle came on again for eye performs a rapid relocation from the home
another burst of about 100 ms. These records are position to a point about 4 cm short of the 40 cm
even more helpful if they are superimposed on target (indicated by the arrow denoting point of
 Methodology for Studying Motor Performance II! 47

associate the movements of body parts with the

'E perceptual information that may be received
~ 50 during action. Such techniques may be informa-
-
CJ
c.. tive when one is determining, for example, how
c 40
CD visual information is used during action (chapter
E
CD
u 5), or how athletes of varying levels of skill use
ltl 30 visual information in performance (chapter 9).
c.
Ill
:a Figure 2.20 illustrates the use of eye tracking to
-
iii
N
c
0
·;::
20

10
measure point-of-gaze information during the
performance of a golf putt.
0
::c
Measures of Brain Activity
15,000 The most dramatic advances in measurement in
recent years have been the various methods of
10,000 measuring brain activity. Their use in describing
N'
~
the brain activity that underlies motor perfor-
E 5000 mance is still in development; and the prom-
.§.
c ise of faster, more flexible, and more accurate
0
;;
0 advances in technique remains a technological
ltl
.... challenge. Each of these techniques has advan-
CD
Cii -5000
u
u tages and disadvantages, although changes in
ct methodology are occurring at a rapid pace and
-10,000
many of the current disadvantages may yet be
-15,000 overcome.
200 400 600 800 Electroencephalography (EEG), which has
Time (ms)
been around the longest, involves the recording
of electrical changes that occur in the brain as
FIGURE 2.19 Acceleration traces of the finger, elbow,
recorded from the scalp, more or less as EMG
and shoulder, relative to the changes in horizontal dis-
placement in point of gaze, during a rapidly aimed move- measures electrical activity of a muscle via the
ment. skin adjacent to the muscle. Another type of
Adapted from Helsen et al. 2000 and 1998. encephalography is known as magnetoencepha-
lography (MEG). As the name suggests, this tech-
nique measures changes in the magnetic fields
gaze). The bottom half of the figure illustrates involving brain activities. Both techniques have
kinematic information regarding the accelera- the advantage of working very quickly, providing
tion profiles of the finger, elbow, and shoulder precise measures of the timing of mental events.
as the upper arm moves toward the target. These However, their capacity to allow inferences about
kinematics were provided by an optoelectric localized anatomical structures is significantly
measurement system that, while separate from less than that of other, albeit slower, methods
the eye-tracking system, was calibrated in real (Haynes & Rees, 2006).
time in order that these measurements could In contrast, methods such as (a) positron
be coordinated later to provide the kinds of emission tomography (PET), (b) single-photon
data illustrated in figure 2.19. The dotted line emission computed tomography (SPECT), and
in the figure illustrates that the end of the initial (c) functional magnetic resonance imaging (fMRI)
rapid-eye movement toward the target (called are all neural imaging techniques that provide
a "saccade") corresponded very well with the much more detailed information about the local-
point of peak acceleration of the finger. A second ization of brain structure and activity. The latter
saccade of the eye, initiated about 100 ms later, technique (fMRI) is based on the fact that when a
again corresponded with an increase in the specific part of the brain is active in the process-
acceleration profile of the finger. Methods such ing of information, neural activity increases and
as these have provided a rich source of informa- consequently oxygenated blood to that region of
tion about motor behavior, as researchers can the brain increases. The fMRI records the "BOLD"
48 Motor Control and Learning

FIGURE 2.20 Eye-tracking setup.

Reprinted, by permission, from J. Vickers, 2007, Perception, cognition, and decision training: The quiet eye in action (Champaign, IL: Human Kinetics), 59.

(blood oxygen level dependent) signal as vectors TMS). Recent investigations have combined
of three-dimensional pixels (called "voxels"), TMS with £MRI so that the effects of stimulation
allowing researchers to map the location and on both behavioral and brain responses can be
volume of brain regions that are actively process- observed (Sack & Linden, 2003).
ing information during task performance. The These neural imaging and simulation tech-
use of new brain recording techniques in motor niques are useful in localizing the parts of the
control and learning experiments has increased brain that are active during the performance of
dramatically in the past decade, although some many activities. Indeed, many motor behavior
challenges remain, such as temporal resolution researchers are including these techniques as
and the nature of tasks that can be used with routine components of their investigations to
these devices. provide neural-anatomical evidence that aug-
Although not a brain activity "measurement" ments the motor behavior data obtained in their
device, an approach to studying brain function experiments.
that is frequently used today is called transcranial
magnetic stimulation (TMS). The TMS method
sends a brief pulse through a magnetic coil Measuring and Evaluating
that has a temporary effect on the region of the
brain above which the coil has been positioned. Relationships
Depending on where the magnetic coil is posi-
tioned, the brief pulse can either excite or inhibit An important process for evaluating the out-
the activity of that brain area. In contrast to what comes of various experimental procedures begins
happens in nonstimulated trials, the TMS-stim- after the main performance measures are gener-
ulated locations of the brain are assumed to be ated from the experiment. This process involves
active if a change in task performance is observed determining the relationship between some inde-
(see Hallett, 2007, for an excellent primer on pendent variable and a dependent variable on the
 Methodology for Studying Motor Performance 49

basis of the empirical data. One important kind of In this equation, Y represents the values on
relationship is linear. In a graph, the dependent the y-axis (error), X represents the values on the
variable plots essentially as a straight line with x-axis (average velocity), and a and b are con-
the independent variable. stants (figure 2.21). The constant a is termed the
y-intercept, and it refers to the value of Y when
linear Relationships the line crosses the y-axis; here the value is about
In figure 2.21 we have plotted data from an 2 mm. The constant bis called the slope and refers
experiment by Schmidt and colleagues (1979), to the amount of inclination of the line. The slope
for which the error (measured as VE and called can be positive (upward and to the right, as in this
effective target width, We) in hitting a target with example) or negative (downward to the right),
a handheld stylus is plotted as a function of the associated with either positive or negative values
average velocity of the movement. A quick exami- of b, respectively. Once these values are specified
nation of the plot in figure 2.21 indicates that from a given set of data, the empirical equation
the relationship between these two variables is that describes the linear relation between values
essentially linear. A line has been placed through of X and Y can be written.
the points that seems to represent their general
direction, and this line is called the line of best fit. Computation of Constants
The actual placement can be done accurately by Computation of the constants needed for the
various statistical techniques (e.g., regression) or empirical equation is simple. After the line of best
can be done "by eye" for a rough approximation. fit has been applied to the data points, extend it
The goal in this section is to express this line of leftward until it crosses the y-axis and read off
best fit in terms of what is known as an empirical they-intercept, or a. In the data shown in figure
equation, a kind of shorthand that enables us, with 2.21, a equals 2.08 mm.
but two numbers, to convey information about a Next, draw two lines, one perpendicular to the
linear relationship for an empirically determined y-axis and one perpendicular to the x-axis, form-
set of data points. These two numbers also will ing the shaded triangle as shown. The length of
have special meaning in terms of various theo- the line forming the base of the triangle will be
ries; that is, these numbers will be measures of called AX, and the length of the line forming the
certain hypothetical constructs. We begin with side of the triangle will be called AY. The symbol
the general equation for a line: A means "change in"; measures of the changes
Y = a + bX (2.8) in Y (6.7 mm) and the corresponding changes in

8
~7
E
E 6
c:

is
0.... 4
w3
2

50 100 150 200 250

Average velocity (cm/s)

FIGURE 2.21 Graphical method for determining constants for linear empirical equations.
Reprinted, by permission, from A.A. Schmidt et al., 1979, "Motor-output variability: A theory for the accuracy of rapid motor acts,'' Psychological Review 86: 427.
50 Iii Motor Control and Learning

X (201 cm/ s) can be seen. Then, the slope of the Interpreting Empirical Equations
line is defined as In addition to the benefits provided by empirical
b=~Y I ~X (2.9) equations in terms of description of experimental
results and prediction of new findings, the values
That is, the slope is defined as the change in Y of the constants a and b often have special theoreti-
divided by the corresponding change in X. Here, cal meaning, depending on the nature of the data
the slope (b) is computed as 6.7 I 201 = +0.033. collected and the kind of independent variable
The interpretation of this slope is that each time studied. In the present example, the meaning of the
the value of X increases by 1 cm/ s, there is a 0.033 constant a (the intercept) is related to the amount
mm increase in the Y value (error). of error for the slowest movement possible, and
Uses of Empirical Equations thus the intercept seems to represent a kind of
The slope and intercept are the only two values "background" or "baseline" error. On the other
needed to determine the linear relationship. Put- hand, the value of the slope (b) refers to the amount
ting the slope and the intercept together into the of increase in error as the velocity increases, and it
general equation for a straight line results in the represents a measure of the "difficulty" of the task.
empirical equation for these data: In this and other similar situations to be discussed
later, the slope and intercept describe two distinct
Y = 2.08 + 0.033X features of the task or the subject's behavior.
Having been provided the calculated values of
a and b found by a person in California, someone Correlation and Regression
in Munich can reconstruct the line of best fit by A statistical tool that is used often in motor
using the linear equation. This is done by picking behavior research, and that has mathematical
any two arbitrary values of X (say, 50 and 250 properties similar to those of the simple linear
cm/s) and calculating the values of Y for these regression method, is the correlation. The correla-
values of X: tion and regression methods are used to establish
Y = 2.08 + 0.033 (50) = 3.73 mm the degree of association between two measures,
as seen in the study of individual differences,
Y = 2.08 + 0.033 (250) = 10.33 mm for example (in chapter 9). For a relatively large
Then, on a new graph, these data points (X = group of subjects (e.g., 50), we begin with two
50, Y = 3.73; and X = 250, Y = 10.33) can be plot- different tests administered to each person. The
ted, and the line in figure 2.21 drawn between degree to which the performances by individuals
them. Thus, saying that the intercept was 2.08 on one test are related to the performances of the
and the slope was 0.033 can convey a great deal same individuals on the other test is reflected in
of information about the experimental results to the size of the correlation coefficient. The correla-
someone who does not have access to the nine tion coefficient expresses the amount of shared
actual data points. association that exists between the two data sets,
In addition, this relation can be used to predict with no implications about whether or not one
new values of error before they are found. If we variable caused the other to change.
wanted to choose a velocity value so that the error Scaffergrams
was only 5.00 mm, we could take the empirical
equation and substitute the value of the error as One of the ways in which data from two tests can
follows, then solve for the value of the velocity: be described is by a special kind of graph called
a scattergram. Consider the data shown in table
5.00 = 2.08 + 0.033X 2.3, which have been plotted on the scattergram
x = (2.08 - 5.00) I (-0.033) in figure 2.22. The two axes in figure 2.22 are the
scales of the two tests, respectively; and each
X = 88.48 cm/ s
of the subjects is represented as a dot, located
Thus, if we wanted the error to be about 5 according to his scores on the two tests. The data
mm, we would use a velocity of about 88 cm/ s. are hypothetical scores that might be obtained
Having an empirical equation makes it possible on a common playground and consist of age (in
to predict this result without actually going into years) and the time for a 100 m dash (in seconds).
the laboratory to measure it directly. In figure 2.22, the scores for these 10 people are
 Methodology for Studying Motor Performance 51

12.8
12.0
10.5
16.1 22.4
15.2 23.7

7 16.0 22.5

8 14.1 25.5

9 13.0 27.7

10 18.0 20.0
Note: Speed data were computed from the time data.

plotted, so that each of the 10 dots on the graph between these two variables, and a line could be
represents each person's joint scores on the two drawn through the points to represent the "direc-
variables (age and running time). tion" in which the "cloud" of points is oriented.
A relationship apparently exists between the This "line of best fit" can be determined in exactly
score on the age variable and the score on the run- the same way as discussed earlier for empirical
ning test, indicating that as the age score becomes equations, and often involves computing regres-
larger, the number of seconds on the running test sion equations.
tends to become smaller, with some exceptions. Direction of the Relationship In this example, as
In general, the 10 subjects showed a relationship the value of the age variable increases, the value of
the running time variable tends to decrease. This
kind of relationship is called an inverse, or negative,
relationship, and the equation representing this
18
relationship has a negative slope constant (b). In
other situations, we might find that as one of the
variables increases, the value of the other variable
16
tends to increase as well; this is a direct, or positive,
~
Cl> relationship. In such cases, the slope constant of
:S 14 the regression equation has a positive value, with
Cl
c: the line of best fit sloping upward to the right.
·~ 12
::i
The direction of the relationship shown in such
a: data is often dependent on the scoring system
10 used. Consider the data in table 2.3. In the fourth
column we have expressed each subject's 100
8 m dash scores as average running speed (km/h)
rather than as the time required to travel 100 m.
9 12 15 18
This change in scoring system "inverts" the group
6
Age (years) of scores, so that the person who had the largest
time score has the smallest speed score, and so
FIGURE 2.22 A scattergram showing the relationship
on. When the age data are plotted against the
between age and running time. (Data are from table 2.3, average running speed scores in figure 2.23, the
and each dot represents one of the 10 subjects.) relationship becomes positive, and the empirical
52 Motor Control and Learning

is its strength. By strength, we mean the extent

35
to which the relationship is perfectly linear, or
the extent to which all of the subjects' points fall
exactly on the line of best fit. Figure 2.24 shows
two scattergrams that represent relationships of
different strengths. In figure 2.24a, the relation-
ship can be considered quite strong because
nearly all the points fall close to the line of best
fit. In figure 2.24b, however, the relationship
is not very strong, because the points tend to
fall away from the line of best fit. These two
aspects, direction and strength, are the primary
descriptors used to characterize relationships.
20 They are independent, in that a relationship can
be either positive or negative and either strong or
weak.
6 9
Age (years) Predicting From a Relationship One of the
most important reasons scientists want to know
FIGURE 2.23 A scattergram showing the relationship the nature of the relationship between two vari-
between age and running speed. (Data are from table 2.3, ables is for the purpose of prediction. For example,
where speed is computed from running time; each dot rep-
resents one of the 10 subjects.) if we know that the relationship between age and
running time has been found to be as described
in figure 2.22, then on a new group of children
equation would now have a positive slope we can estimate (or predict) the 100 m time given
constant (b) rather than a negative one. Thus, a the age of a person not in the original data set
positive relationship is not any "better" than a without actually measuring running speed, just
negative one; the sign simply indicates the direc- as with the linear equations discussed earlier.
tion in which the line of best fit is sloped. Such procedures are used extensively in every-
Strength of the Relationship A second charac- day situations, as in predicting the probability of
teristic of the relationship between two variables having an automobile accident from one's age,

a
x b
x

FIGURE 2.24 Hypothetical scattergrams for (a) a strong and (b) a weak relationship.
 Methodology for Studying Motor Performance im 53

or predicting success in graduate school from ship, exactly as described in the previous sections.
achievement test scores. The absolute size of the correlation indicates the
As the strength of the relationship increases, strength of the relationship and hence is critical
the predictability of one variable from the other for evaluating the extent to which one can use
increases as well. When the relationship is perfect the relationship to predict. Figure 2.25 shows five
and all the individual data points fall exactly on hypothetical examples of correlations and the
the line of best fit, perfect predictions with no associated scatterplots. Both a +0.90 and a -0.90
error can be made. When the data are related less correlation are strong relationships, as all the data
perfectly, as in the example shown in figure 2.24b, points fall almost exactly on the lines, although
then more error is introduced into the predictions. the lines are sloped in opposite directions. Cor-
Thus, the strength of the relationship-but not the relations of +0.50 and -0.50 are moderate in
direction-is the primary determinant of the extent strength, and the points fall considerably away
to which a relationship can be used to predict. from the lines of best fit. A correlation of zero is
weakest, indicating that no predictive capability
Correlation Coefficient is possible between these two variables. Formulas
These concepts of strength and direction of for the calculation of correlations can be found
relationships can be quantified using a statistic in most statistics textbooks (e.g., chapter 8 in
called the correlation coefficient. The correlation, Thomas, Nelson, & Silverman, 2005; chapter 7
abbreviated r, ranges from +1.0 through zero to in Vincent, 2005).
-1.0. The two important aspects of the correlation A convenient method for comparing the
are the sign and its absolute size. The sign of the strength of relationships between tests is to square
correlation indicates the direction of the relation- the correlation coefficient and multiply by 100 to

y y

x x

y y

x x

FIGURE 2.25 Scattergrams for hypothetical data showing high, moderate, and low relationships.
54 Motor Control and Learning

convert it into a percentage score. Generally, the error (AE, also a measure of overall error). Each
square of the correlation coefficient indicates the error measure has a different meaning and is used
extent to which two tests can be considered to in different aspects of the measurement process.
measure the same thing, and represents the per- Measures of speed are used when accuracy is
centage of shared variance (or, the percentage "in less important (or is controlled) and when rapid
common") between the two variables. This fits actions are critical. Measures of magnitude are
nicely with the earlier discussion of the strength used when the amount of behavior is critical. A
and direction of relationships. The shared vari- fourth but related measure is based on the analy-
ance for two tests that correlate positively (+0.90) sis of simultaneous secondary tasks, providing a
or two others that correlate negatively (-0.90) is measure of the spare capacity of the performer
the same (e.g., +0.902 x 100 = -0.902 x 100 = 81 %), after she has devoted attention to a primary task.
indicating that the direction of the relationship Movements can be measured in many ways,
(positive or negative) has no effect on the amount but common methods involve the calculation of
of shared variance. kinematic variables (position, velocity, accelera-
The r2 value is useful for interpreting the tion) and the recording of the electrical activity
strength of two relationships. For example, sup- from muscles (EMG). Methods used to assess
pose two variables correlate with a value of r kinematics have changed dramatically with digi-
= 0.50 and two other variables correlate with a tal technology, and precise measures of sensory-
value of r = 0.25. In this case, the r2 value of the motor control are now the norm. Studies of brain
first relationship (0.50 2 = 0.25) indicates approxi- activity are also becoming increasingly popular,
mately four times more shared variance than for and their use in motor behavior research contin-
the second relationship (0.25 2 = 0.063), or that the ues to grow. Linear empirical equations provide a
first two tests had approximately four times more description of a linear relationship between two
in common than did the second pair. variables. The parameters of the equation can be
easily estimated, and they provide a means by
which the relationship can be used to predict facts
Summary that are yet unknown. The parameters can also
be useful in describing the direction and strength
Motor behavior can be classified according to of the relationship between two variables as indi-
several dimensions, such as (a) continuous/serial/ cated by the correlation coefficient.
discrete, referring to the extent to which the move-
ment has a definite beginning and end, and (b)
open/closed, referring to the extent to which the Student Assignments
environment is predictable. Most of the tasks used
in motor control work fall into one or the other of 1. Answer the following questions and bring
these basic categories. In measuring movement, the information to class for discussion:
experimenters devote attention to a measure's a. Compile a new list of examples from
objectivity (the extent to which two independent activities of everyday life that fit each
observers achieve the same score), its sensitivity to of the following task categories: dis-
changes in skill, its reliability (the extent to which crete/open, discrete/closed, serial/
the score is repeatable), and its validity (the extent open, serial/closed, continuous/open,
to which the test measures what the experimenter continuous/closed.
intends it to measure).
The outcome of movements in terms of the b. Close your eyes and draw 10 lines as
environmental goal can be measured in essen- close as possible to 10 cm in length.
tially four ways: in terms of errors, speed or time Do not open your eyes until you are
(or both), or magnitude, or with various second- finished. Calculate the following error
ary tasks. There are many ways to measure errors measures: CE, ICEI, VE, E, and AE.
in movement; chief among these are constant error c. Discuss the relative importance of
(CE, a measure of average error or bias), variable reaction time in the following Olympic
error (VE, a measure of inconsistency), total vari- events: 100 m run, 10,000 m run, 100
ability (E, a measure of overall error), and absolute m butterfly swim, 400 m relay.
 Methodology for Studying Motor Performance 55

2. Find a research article published in the past This site provides some examples of Muybridge' s
five years that uses functional magnetic photographs:
resonance imaging (fMRI) or transcranial
magnetic stimulation (TMS) to better www.masters-of-photography.com/M/muybridge/
explain basic motor control processes. muybridge.html

Web Resources Notes

1 A quantitative analysis of error scores in two dimen-
This resource is related to reaction-time method- sions is provided by Hancock, Butler, and Fischman
ology and the history of RT research: (1995), but is beyond the scope of the present discussion.
www.chss.montclair.edu/psychology/museum/
mrt.html
 CHAPTER 3
·. · •·. · . · ·•. ·.·• .· •·. •·. N·.·•.• ··.o
HUMAN. .•··. ·•. ·.• · . · ·.••·R
·.·.···.·.·.1.···
... F .·• .··.••··. •:·. ·.~l ·!·;f~,
. · . · •. •.· .·• .·•·••.-.•
.· .··. ·.· ·. . •.·•. •. . •.
1

P.ROCESSml~~Ji,.o~:
.•···.•.·····..•.•·.·.•
. ..··•·...•.•.• ·...· .•••·. . .•··•..····:·.··i·····················l····.. .·.:
. .•..•
•·.•.•. ·...·•.•.·.· •· •· · ...•.·· •.•.•.•. .:·.· ·.....·.· •...·•.•.•.·. ·...·.•..•·.·.·.m
•. ....••.·..•.•.•·.·..•.•.·....••.·.•.·
..·•·.·..··.··•···...
·....•·..··.·..··.··...·......
• ·... . •

uman functioning in the environment can cessed information can result in various kinds
be conceptualized and studied in many of movement, just as a computer displays the
ways; one of the most popular is based on the results of its processing on a monitor or sends it
fundamental notion that humans are processors elsewhere as an output. The goal of this chapter
of information. It is assumed that information is is to discuss information processing as it relates
available in the environment; that the individual to human motor behavior (see also Marteniuk,
accepts this information into various "storage 1976)-specifically how information is processed
systems" called memory; and that the informa- for the specific purpose of producing skilled
tion is then "processed" for the purposes of movement.
perception, decision making, and action. Human
information processing is based on a computer
metaphor. According to this metaphor, we "take
The Information-Processing
in" information from outside sources, just as a Model
computer takes in information via input devices.
That information undergoes transformations, The model begins with the input of informa-
uses other information stored in memory, and tion from the environment through one or more
is subject to certain limitations based on the of the sense organs, and then considers what
amount processed and the speed of processing happens to this information once it is inside
that is available; again, this is similar to how the system. A typical "black box" model of the
a computer processes information. And lastly, process is shown in figure 3.1. The individual is
the process by which information is "output" considered to be the box, and information enters
has many analogies to the computer-the pro- the box from the environment. This information

Output
Input
(Signals) --+ --+ (Motor
response)

FIGURE 3.1 The simplified information-processing model.

57
58 Motor Control and Learning

is then processed in various ways, eventually viewpoint, the general concept of stages of pro-
resulting in output as observable motor activity. cessing is quite old, dating back to the research of
This black box model prevailed in the stimulus- Danders (1868/1969; see "Danders' Subtractive
response (S-R) tradition (see chapter 1), in which Method"). This thinking, coupled with efforts
researchers were primarily concerned with the from cognitive psychology (e.g., Kellogg, 2003;
relationship between what went into the box Sanders, 1980; Schweickert, 1993; Sternberg,
(the information, or stimuli) and the output 1969), has led to the view that various processing
from the box (the response). With the emergence stages can be defined and that these stages can
of cognitive psychology, however, interest began be either serial (sequential) or parallel (simultane-
to focus on the processes that occur within the ous) in nature.
box. Obviously, this is an abstract way to study Danders argued that with use of these
human behavior because it concerns processes methods, the durations of the stages of process-
and events that are not directly observable. ing involved in stimulus discrimination and
Knowledge about these processes is inferred response selection, which are not directly observ-
from the overt behavior of the human under able, could be estimated in the laboratory. It is
various experimental conditions. Events occur- remarkable that these insights were developed in
ring inside the box can be studied in a variety of 1868, long predating the cognitive revolution in
ways through use of the cognitive-psychological experimental psychology that made use of these
perspective. and similar concepts.
The most common approach to the study of Danders' general methods and assumptions
information processing is to consider the dura- would later be revived in a classic paper by Saul
tions of these various processes. This chronometric Sternberg (1969) that generated considerable
approach (see Posner, 1978) makes considerable interest and further research, some of which
use of the reaction-time (RT) method, whereby was involved in understanding motor behavior
the chief measure of the subject's behavior is the (e.g., Sanders, 1980). Later thinking and research
interval between the presentation of a stimulus identified a number of flaws in Danders' subtrac-
and the beginning of the response. Many dif- tive method (see Massaro, 1989, for an analysis
ferent information-processing activities occur of Danders' and Sternberg's logic and methods).
during RT; but if the experiment is designed But even so, Danders' basic idea that we could
properly, so that the durations of most other examine the duration of stages by subtracting the
processes are held constant, one can usually RTs in various conditions was remarkable, given
infer that an increase in RT resulting from some the time of his work, and served as the foundation
experimental variable was caused by the length- for more modern analyses of human information
ening of the duration of a particular process. This processing. 1
chapter presents many examples of this type of
research and thinking, and the chronometric The Nature of Serial
approach to studying information processing and Parallel Processing
constitutes a large portion of it. Later in the
Imagine an automotive plant as a model of infor-
chapter we consider other aspects of informa-
mation processing, as in figure 3.2. Some stages
tion processing that are not necessarily studied
occur at the same time in different places, such as
using the chronometric approach but are par-
the fabrication of the electronics, the assembly of
ticularly important in the performance of motor
the engine, and the assembly of the chassis. But
skills.
at various times during the overall process, these
components are combined to complete the final
Three Stages assembly stage. After final assembly, imagine a
test-drive stage, in which the drivers search for
of Information Processing problems before the car is sent to the dealers, as
yet another stage. This simple analogy contains
Although the notion of separate stages or pro- an example of both parallel processing (simulta-
cesses between a stimulus and a response has neous assembly of the electronics, engine, and
been popularized by the cognitive-psychological body stages), and serial processing (successive
 Donders' Subtractive Method
Over a century ago, the Dutch physician F.C. Donders (1868/1969) made the first attempts to measure
the time required to complete certain thought processes. Donders' subtractive method assumed the
existence of a series of separate, nonoverlapping information-processing stages between a stimulus
and a response. The notion was that the processing that occurs in stage 1 is separate and distinct from
the processing that occurs in stage 2, and that stage 2 processing cannot begin until stage 1 process-
ing is completed. Donders studied this idea using three RT methods that differed in systematic ways.
In the simplest method, termed the a-reaction task, the subject was presented with a single,
unanticipated stimulus that required a single response (e.g., pressing a key using the right hand
in response to the illumination of a red light). This reaction task is now commonly referred to as a
simple-RT task. In a more complicated task, termed the c-reaction task, a subject was presented
with two different light stimuli and asked to respond by pressing a key with the right hand if a specific
stimulus was illuminated. For example, if the two stimuli were red and blue, then the task might be
to respond if the red stimulus was illuminated but to not respond to the blue stimulus. This reaction
task is today called a go/no-go task because the subject is to respond if the specific stimulus is
illuminated ("go") but not if the nonspecified stimulus is illuminated ("no-gd'). In the third task, called
the b-reaction task, the subject was again presented with more than one stimulus and asked to
make a response that depended on which particular stimulus was presented. For example, a red
light might require a key press with a finger on the right hand whereas a blue light would require a
left-hand response. This reaction task is known today as a choice-RT task.
The logic of Donders' subtractive method was that the three tasks differed in the number of stages
involved. All three tasks require stimulus detection (e.g., noting that a stimulus light has appeared).
In addition, however, the b-reaction involves the processes of discriminating the stimulus that was
detected from the possible alternatives that could have appeared, as well as the selection of a specific
response. The c-reaction, however, requires only discrimination of the stimulus, and no response
selection (because the specific response is always the same in the c-reaction). Thus, the difference
in RT between a c-reaction and a b-reaction reflects the time to perform response selection. In a
similar way, the c-reaction involves stimulus discrimination whereas the a-reaction does not; only
one stimulus is presented in the a-reaction, and both of these tasks involve only one response (i.e.,
no response selection is needed for either). Therefore, the difference in RT between the a-reaction
and the c-reaction should reflect the time for stimulus discrimination. This can perhaps be seen
more clearly in table 3.1, which presents an example using these ideas.

Choice RT 2 2 ' Stimulus detection, stimu- 285 ms

lus identification, response
selection, response execution ,
2 Stimulus detection, stimu- 230 ms
lus identification, response
, execution

, 285 - 230 = 55 ms

59
60 Motor Control and Learning

Parallel processing Serial processing

FIGURE 3.2 Examples of serial and parallel processing in an automobile assembly plant.

stages of final assembly and test drive). Thus, in the decisions needed to respond to a yellow
this system has both serial and parallel process- (caution) light at an intersection. Although
ing, but at different times in the total sequence. many of the experimental methods and research
In human behavior, many have thought that the issues discussed in this chapter may seem quite
total RT has various stages, some of which can removed from this example, it is probable that
be performed in parallel and some of which are the processing of information in lab tasks and
performed serially. many examples of everyday life involves similar
At least three stages can be proposed that inter- activities.
vene between the presentation of a stimulus and
the production of a response (see figure 3.3). First, Stimulus-Identification Stage
the individual must sense that a stimulus has Think of the stimulus-identification stage as
occurred and identify it. This stage is frequently beginning with the presentation (or onset) of an
called the stimulus-identification stage. Second, environmental stimulus that must be detected;
after a stimulus has been properly identified, the this stimulus information must then be recog-
individual must decide what response to make. nized as a part of an identifiable pattern. These
The decision can be to do one of a number of two substages are considered in the following
possible actions, or the stimulus can be ignored sections.
in favor of no action at all. This stage is usually
called the response-selection stage. Finally, after Stimulus Detection
the response has been selected, the system must The occurrence of an environmental stimulus
be prepared for the appropriate action and then (e.g., light entering the retina of the eye or sound
initiate that action. This stage is frequently called entering the ear) results in neurological impulses
the response-initiation stage, but we will use the that are sent toward the brain. The stimulus is
term response-programming stage to represent the presumably processed further at different levels
preparations of the motor apparatus and the of analysis until it contacts memory, meaning
initiation of the action. More detailed discussion that some memorized aspect of the stimulus is
of these stages is presented in the remainder of aroused, such as its name or an attribute with
the chapter. which it has been associated in the past (e.g.,
"Information Processing and Traffic Intersec- its red color, as with a stop sign). Considerable
tions" (p. 62) presents an example of the poten- processing must occur in order for the stimulus
tial information-processing activities involved to arouse the proper association in memory
 Human Information Processing 61

FIGURE 3.3 An expanded information-processing model.

(rather than an improper associate or all pos- Pattern Recognition

sible associates). Scientists working in this area
In more realistic tasks, the stimuli that enter the
have assumed that the variables affecting the
system are seldom as simple as they are in lab RT
stimulus-identification stage relate specifically
tasks, and we must usually extract a pattern or
to the nature of the stimulus. For example, an
feature from the stimuli presented. Often these
experimental variable called stimulus clarity,
patterns have to do with such things as the shape
which refers to the extent to which the visual
of a face-or in sport, with where a baseball is
stimulus is well defined and "sharp" (vs. out of
focus) has been used. With increased clarity, the going, how fast it is traveling, and what kind of
overall RT is shorter, and this change is attrib- spin has been put on it. Some of these pattern
uted to the increased processing speed in the detections are genetically defined (e.g., related
stimulus-identification stage. A variable called to survival). Others may depend heavily on
stimulus intensity (e.g., the brightness of a light learning, such as recognizing a developing play
stimulus or the loudness of a sound stimulus) in volleyball.
is also assumed to affect information-processing Important studies of chess players demon-
time in this stage. Early research that has been strate the influence of learning on pattern detec-
replicated many times also revealed that RT is tion (deGroot, 1946/1978; Chase & Simon, 1973).
greatly affected by the modality of the stimulus- In one such study, deGroot asked chess masters
the latency in responding to a visual stimulus and good-to-average chess players to reconstruct
is somewhat slower than the time to respond the locations of the chess pieces in a half-finished
to an auditory or a tactile stimulus. Further, game after viewing the board for 5 s. As one might
stimuli presented simultaneously in more than imagine, the chess masters were far superior to
one modality (e.g., combined visual + audi- the good-to-average players. It could be argued
tory stimuli) will shorten RT relative to either that the superiority of the chess masters in this
modality presented alone; this is called inter- task was not necessarily evidence that they had
sensory facilitation (Nickerson, 1973; Schmidt, learned to remember chess patterns, but rather
Gielen, & van den Heuvel, 1984). Many more that they were just superior, generally, in their
effects involving the physical properties of the inherent perceptual abilities. This last hypothesis
stimulus have been investigated, much of this is doubtful, however, as a later experiment by
research having been conducted many years Chase and Simon (1973) included an important
ago. An excellent summary of the effects of the condition in which the chess pieces were placed
properties of the stimulus on RT by Woodworth on the board in random fashion. With this random
and Schlosberg (1954) remains informative arrangement, the chess masters and the average
today. players were about equal in their ability to
 e.¢'~0~··1;19~ tb•J€lSpond when an
··. · :~~~rflnformation processing
···a~soll:rc(:)s of environmen~
etfa~i~~ighaLfrorn green .to yellow is
· tl\a decision: lnthe simplest case,
• jth~rto.continue on through the
•.. 6afcieci~i()l1, the driver quickly
·:fo. l;>e.c;;overed (Senders, 1998).
frnu$falsoaccess i.nformation in
,pf the car behind.More remotely
this pa,rticular traffic intersection (such
~ll:O'W.p~fiSeof the traffic signal) and the
'n istc;> carry o'n through the intersection,
at~~ti~~$a,J}:le seeed or even to a.ccelerate. If the deci-
rn tl;i~ accelera,tor and apply it to the brake.
lYe..irii b'rakef a.n action is required that must
f'flusc~lat(Jre.for implementation (see Green,

.··1cleo~fyi~~tl;iatifie,.si9f1al has changed color, select-

. • ~ ~sp~pse to be made-tal<e time, during
ti~n;,Jt~i$.:qJ.lite:u.oder~tandable, therefore, why. a
.• };Jilp~~~tcofYiponents of information processing,
'mijtr19~pl'Jroa,cl;i {eosner, 1978).
R'~··a.r¢'.a.fso il)"lgp~1~1nf c9mpo11ents of ·information-
.Jta Qne
•.!•.·· pf these .entails anticipation, the use.
Of c~itjcal stimulus~ For example, a driver who
.S90614\ly.a11tic.ip.ate'ti:.ata, YelJoW lightwill not occur
~erU~i#'Jnf()rr'flatibn is 4'dmbined with prior informa-
\i$.fl\liflytbng, ~e other factor relates to the potential

62 Ill!
 Human Information Processing 111 63

drivermakesanr~c~~
results if the
was the. appropriate respon§~;·~(~C>
rect response). fy1ak.irtg.thedeclsf
outcomes,·one correct andoneiri•
the outcome of this decision is the:3
All of these va.riousissues(prdee~~ln~
features that influence indirectlY,how·W J.

reposition the pieces (see also Vicente & Wang,

1998). One interpretation is that the processes
involved in the stimulus-identification stage
were improved in the masters through years of
experience in game situations and that consider-
ably more information could be accumulated in a
single view of the board, but only as long as that
information conformed to the normal patterns
of chess play.
Analysis of such static situations seems
important to many activities; but even more
important from the point of view of motor
behavior is an ability to extract patterns of move-
ment from the environment. In many situations,
how the environment changes from moment to
moment will determine which action is most
appropriate.
A great deal of environmental movement infor-
mation is represented as changes in the visual FIGURE 3.4 Optical arrays vary as the location of the
field, and it seems clear that an individual can observer changes in the environment.
use this information to provide an unequivocal
analysis of movements in an environment or of
These interpretations are important for
the environment's movements. Gibson (1966),
many fast-action sports and games. Patterns of
for example, referred to "optical-flow patterns,"
opponent position or action arouse meaningful
which are the patterns made up of the rays of light
responses that lead to fast action in highly skilled
that strike the eye from every visible part of the
players, while they may go nearly unnoticed by
environment (see figure 3.4). As the individual or
novices. For example, certain patterns of linemen
the environment moves, the angles of these rays
and backfield movement in American football
change predictably to allow the subject to extract
mean the difference between a running and a
a pattern of movement from the changing visual
passing play, and an effective response to this
array. For example, you know that a ball is coming
action by the defensive player often depends on
directly toward your eye if the rates of change of
the angles of the rays of light from all edges of the recognizing such a pattern quickly and accurately
ball are the same. This looming can elicit strong (for reviews see Ericsson, 1996; Starkes & Allard,
av?idance reactions even in children and young 1993; Starkes & Ericsson, 2003; Williams, Ward,
& Smeeton, 2004).
arumals who have presumably never been hit in
the eye, suggesting that at least some kinds of
Response-Selection Stage
pattern recognitions may be genetically defined.
We will discuss these various aspects of visual As a product of the stimulus-identification stage,
information processing in more detail in chapter 5. the information in the stimulus input has been
64 Motor Control and Learning

analyzed, and the individual now has a basis for stimuli was actually presented. (On this set of
"knowing" what happened in the environment. trials, only 3, 5, and V would be presented, and
In the next stage, response selection, the subject none of the other stimuli would be involved.)
decides what response to make. A baseball out- Other sets of trials might involve 1,3, 5, and VI
fielder must make rapid decisions about whether (four possible stimuli) or 2, 4, VII, VIII, and X
to attempt to field a batted ball either before or (five possible stimuli). Note that, on any given
after it bounces, what direction to move in order trial, only one of the set of possible stimuli would
to catch it, and what to do with the ball if it is occur. Merkel studied the relationship between
fielded successfully. Such decisions are important the number of possible S-R pairs (ranging from
and apply to a wide variety of activities in sport, 1 to 10) and the choice RT.
industry, driving, and so on. His basic findings are presented in the left
side of figure 3.5, which plots the choice RT
Number of Stimulus-Response against the number of S-R alternatives (or N). As
Alternatives the number of alternatives increased, so did the
For over a century (since Danders' contributions), choice RT taken to respond to any one of them.
scientists have believed that the processing of This relationship was curvilinear, as is clearly
information relevant to the selection of a response illustrated in the data on the left side of figure 3.5.
requires more time when the number of possible Note, for example, that as the number of alterna-
alternatives is larger. The idea is that if increasing tives was increased from 1 to 2, the increase in
the number of alternatives causes an increase in choice RT was about 129 ms, whereas when the
the choice RT, then the increased RT is associated number of alternatives was increased from 9 to
with changes in the way the information was 10, the increase in choice RT was only about 3
processed in the response-selection stage. ms. Even though there was an increase by one
In the choice-RT paradigm, the subject might in the number of alternatives in both cases (i.e.,
be presented, say, with four stimulus lights and l-72 vs. 9-710 alternatives), the effects on choice
instructed that one of the four will be illuminated RT were very different.
randomly on a particular trial. Each of the four This relationship between number of alterna-
lights is associated (via instructions from the tives and choice RT has been studied a great deal
experimenter) with one of four different responses since Merkel made his original observations. The
(e.g., the four stimulus lights are assigned, or overall conclusion has not changed, although
"mapped," to buttons located under the four there have been some refinements in technique
fingers of the right hand). The task is to press the and much additional theorizing about the causes
appropriate button as quickly as possible after of the relationship. The most widely known
one of the stimulus lights is illuminated. Usu- findings and explanations of the effect were
ally subjects are not able to predict exactly when apparently produced by two people at about the
the stimulus will occur and are thus prevented same time-Hick (1952) and Hyman (1953). The
from initiating the response in advance. The time relation they discovered between the number of
from stimulus to response will be sensitive to the S-R alternatives and RT has since been termed
speed of information processing responsible for Hick's law, or sometimes the Hick-Hyman law
the selecting the appropriate finger. (Keele, 1986; Proctor & Dutta, 1995).
One of the earliest studies addressing this Hick's Law Hick (1952) and Hyman (1953) stud-
question was done by Merkel in 1885 (described ied the relationship between choice RT and the
by Woodworth, 1938). The digits 1through5 were number of S-R alternatives in much the same way
mapped to the fingers of the right hand, and the Merkel had, using various numbers of lights that
Roman numerals I through V were mapped to were associated with an equal number of keys to
the fingers of the left hand. In one set of trials, be pressed when the appropriate light appeared.
Merkel used all 10 possible stimulus-response As Merkel had found, choice RT increased as the
(S-R) pairings, so there were 10 possible stimuli number of possible S-R alternatives increased.
for these trials. On other sets of trials the subject The RT values, as well as the overall shape of the
knew which, from the set of 10 stimuli, would be function, were consistent with Merkel's findings
possible (e.g., if there were three possible stimuli, in figure 3.5. However, what Hick and Hyman dis-
they might be 3, 5, and V), yet only one of these covered was that choice RT appeared to increase
 Human Information Processing 65

700 700

600 600

500 500
en en
§. 400 §. 400
I- I-
a: a:
GJ GJ
0 0
·5 300 ·5 300
..c: ..c:
(.) (.)

200 200

100 100

2345678910 2 3
N (number of S-R alternatives) Log2 (N)
a b

FIGURE 3.5 (a) Choice reaction time (RT) as a function of the number of stimulus-response (S-R) alternatives and (b)
the logarithm of the number of S-R alternatives.
Data from Merkel 1885.

by a nearly constant amount (about 150 ms) every est at this point in our discussion are the open
time the number of S-R alternatives was doubled. circles, referring to experiment 1, in which the
This suggested that the relationship between number of S-R alternatives was varied (plotted as
choice RT and the logarithm of the number of S-R the Log2 [NJ, or the number of bits of information
alternatives should be linear. to process-more on that later). For each subject
The interpretation here is that the logarithm of there was a strong linear trend between the Log 2
the number of S-R alternatives is a measure of the (N) and choice RT. The individual linear equation
amount of information that had to be processed, is shown on each subject's graph; for example, for
suggesting that more alternatives required more subject G.C., the intercept (a) was 212 ms and the
information processing. In the right side of figure slope (b) was 153 ms. Notice that these empirical
3.5 we have replotted Merkel's data as a function "constants" were considerably different for dif-
of the logarithm of the number of alternatives, and ferent subjects. Nevertheless, in each case the RT
we see that these data demonstrate a linear fit too. was linearly related to the amount of stimulus
The formal relation that has come to be known as information (or Log2 [NJ).
Hick's law states that choice RT is linearly related That the relationship between the choice RT
to the logarithm to the base 2 (Log2) of the number and the logarithm to the base 2 of the number
of S-R alternatives. In equation form, of alternatives should be so clearly linear is of
Choice RT= a+ b[Log2 (N)] (3.1)
considerable interest in its own right. A linear
relationship is the most simple of relationships,
where N is the number of (equally likely) S-R and scientists become excited about the possi-
alternatives and a and b are the empirical con- bility that complex behaviors of human beings
stants. Notice that equation 3.1 is somewhat can be described by such simple expressions.
different from the "typical" linear equation dis- But of even more importance is one interpreta-
cussed in chapter 2, where Y =a + bX; but if Xis tion of this relationship: that the time required
[Log 2 (N)], then equation 3.1 means that choice to make a decision (or to make a choice) about
RT is linearly related to Log2 (N). a response is linearly related to the amount of
The data from four subjects in Hyman's (1953) information that must be processed in coming
study are presented in figure 3.6; the data of inter- to that decision.
66 Motor Control and Learning

800

(jj' 600
.s
400

200

2 3 2 3
Stimulus information (bits) Stimulus information (bits)

800 800

(jj' 600 (jj' 600

.s.... .s....
a: 400 a: 400

200 200

2 3 2 3
Stimulus information (bits) Stimulus information (bits)

FIGURE 3.6 Choice reaction time as a function of stimulus information.

Reprinted from R. Hyman, 1953, "Stimulus information as a determinant of reaction time," Journal of Experimental Psychology 45: 192.

Measuring the Amount of Information For uncertainty to the logarithm involved in Hick's
Hick, and other scientists at that time, the notion law. First, the logarithm to the base 2 of some
of information was used in a very special way. number N is defined as the power to which the
Information is related to uncertainty and to the base 2 must be raised in order to obtain that
amount of uncertainty that is reduced by the pre- number. For example, the Log2 (8) is 3, since 2
sentation of a stimulus. Specifically, the amount of must be raised to the third power to obtain 8 (23 =
information contained in some signal is measured 2 X 2 x 2 = 8). Other examples are shown in table
in bits (short for binary digits). One bit is defined 3.2 (a more comprehensive table of log values
as the amount of information needed to reduce appears in the appendix). Notice that every
the original uncertainty by half. For example, time N is doubled (say from 4 to 8, or from 8 to
your friend tells you that she is thinking of one of 16), table 3.2 shows that the Log2 of the number
four numbers (1, 2, 3, and 4), and you are to guess increases by 1.
which it is. Then she tells you that the number is Now, as applied to the RT situation, if N is
even. This last message has reduced the number the number of equally likely S-R alternatives,
of choices from four to two (i.e., by half) and has resolving the uncertainty about N things can be
conveyed 1 bit of information. See "Uncertainty interpreted as requiring Log2 (N) bits of infor-
and Information Theory" and figure 3.7 for mation. In terms of Hick's law, we can say that
more details and an example about measuring choice RT is linearly related to the amount of
information. information needed to resolve the uncertainty
Interpreting Hick's Law Now we will connect about N S-R alternatives. Stated differently, every
the notion of information as the reduction of time the number of S-R alternatives is doubled,
 Uncertainty and Information Theory
Scientists who study information processing use the term "information" to refer to the amouht of
uncertainty that has been reduced by a signal that was presented. If during a walk in the rain your
friend says that it is raining, that "signal" conveys little information because there was little original
uncertainty available to be reduced further. But if your friend says that it is raining in the Sahara
desert, that signal conveys a great deal of information because (a) the probability that it is raining
there is low and (b) you had no previous knowledge that it was raining there. Thus, the amount of
information transmitted is affected by both (a) the amount of uncertainty prior to the signal's being
presented and (b) the amount of reduction of uncertainty. Generally speaking, the amount of infor-
mation (H) is given by a simple equation:
(3.2)

where Pi is the probability that a given

event (i) will occur. As the probability of an
event (P;) decreases, the amount of infor-
mation conveyed by a signal describing Task: Find the "x" in as few
that event increases; this is why a signal x questions as possible.
about a rare event (it is raining in the
desert) carries more information than a
signal about a common event (it is raining
in Eugene, Oregon). Figure 3.7 provides a Question #1. Does the x appear
in the left half of the matrix?
concrete example about how the informa-
Answer. No.
tion in a given situation can be reduced Information transmitted: The 8 squares
by asking a series of uncertainty-reducing on the left side are eliminated and 8
questions (from Attneave, 1959). In the remain (shaded); uncertainty is reduced
by half, so 1 bit of information has
figure is a 4 x 4 matrix of squares, com-
been transmitted.
prising 15 empty squares and 1 square
with an "X" inside. If you did not know Question #2. Does the x appear in the
which of the 16 squares the "X" appeared bottom half of the remaining squares?
in but wanted to find out, you could do so Answer. Yes
Information.transmitted: The upper 4
by asking four questions (because Log 2
squares are eliminated and the bottom
16 = 4 bits, and because 2 x 2 x 2 x 2 4 squares remain viable; again, uncertainty
= 16), designing each question to reduce reduced by half, and 1 bit is transmitted.
the uncertainty by one-half (1 bit). Notice
also that it does not matter whether the Question #3. Does the.x appear in one
answer to each of the questions is yes of the two squares in the right side?
or no-the information gained allows the Answer. No.
uncertainty to be reduced by one-half, Information transmitted: The right two
squares are eliminated and now only
regardless of the answer. two squares remain; again, 1 bit
transmitted.
Question #4. Of the two remaining squares,
is the x in the bottom position?
Answer. No.
Information transmitted: Since we can eliminate
FIGURE 3.7 Guess the location of the X the bottom square and no more alternatives
in only four questions (from Attneave, 1959). remain, we do not need to ask another question -
Adapted from F. Attneave, 1959, Applications of information the x must be in the only square that has not
theory to psychology: A summary of basic concepts, been eliminated; one more bit transmitted, fora
methods, and results (New York: Holt, Rinehart & Winston).
total of 4 bits in the whole task.

67
68 Motor Control and Learning

slope (b) in Hick's law? From the earlier discus-

sion of empirical equations (in chapter 2), recall
that the intercept (a) was that value of choice
RT associated with Log2 (N) = 0, or the value of
choice RT when the line crossed the vertical axis.
In the data from various subjects in Hyman's
(1953) study (figure 3.6), the average of these
intercepts was 179 ms (mean of 212, 165, 180,
and 160 ms). Also, recall that when Log2 (N) =
O the value of N must be 1 (table 3.2), denoting
1
a situation with only one alternative, in which
there is no uncertainty about what to do (this is
usually called simple RT). Consequently, it has
been reasonable to interpret the intercept (a) of
Hick's law as a measure of the overall "speed"
of the perceptual and motor system exclusive
of any time required for a decision about which
response to make. Comparing subject L.S. in
figure 3.6 (a= 160 ms) with subject G.C. (a= ~12
ms) might be roughly analogous to companng
the amount of information to be processed is
a telephone dial-up modem with broadband to
increased by 1 bit, and the time required for choice
surf the Internet (Seow, 2005)-the basic process-
RT is increased by a constant amount. This con-
ing speed is different.
stant amount is the slope of the Hick equation, b.
Also from Hyman's (1953) data, the slopes
To this point, we have considered only the
of the relation ranged from about 127 ms/bit to
circles in figure 3.6 (Hyman, 1953, experiment I);
215 ms/bit for the various subjects (figure 3.6).
in that experiment, the number of S-R alterna-
Remember that the slope (b) is a measure of the
tives was varied in order to change the amount
amount of inclination of the line; alternatively b
of information to be processed. In Hyman's
represents the amount of increase in choice RT as
experiment II (the filled circles), the amount of
Log (N) is increased by one unit (1 bit). So, one
information was varied by changing the prob-
additional bit of information resulted in from
ability of the stimulus. Remember, as the event
127 to 215 ms of additional choice RT for the dif-
becomes less probable, having to process a signal
ferent subjects. Thus, the slope is the "speed" of
about it conveys more information, as in equation
decision making by the response-selection stage
3.2 (see "Uncertainty and Information Theory").
of processing, measured in units of milliseconds
When information was increased by decreasing
per bit. Seen in this way, the slope and intercept
the stimulus probabilities, RT again increased
measure two different underlying processes in
linearly. In experiment III, Hyman varied infor- human performance.
mation by changing the sequential dependencies
The practical implications of the slope are
(the probability that a given event [or stimulus] important and probably obvious. For games in
is followed by another event [or stimulus]). The which rapid reaction is important, if the player
effect was to make a particular stimulus more or can double the number of likely alternatives
less probable just before the presentation of the to which the opponent must respond (and the
stimulus for a given trial. The squares in figure opponent cannot anticipate them), then the player
3.6 again show a linear relationship between increases by 1 bit the amount of information that
amount of information and choice RT. If this inter- must be processed in order for the opponent to
pretation is correct, the response-selection st~ge respond, and thereby increases by a constant
can be thought of as being involved in reducing amount the opponent's choice RT in initiating the
uncertainty about alternative responses when a appropriate response. These kinds of effects are
given stimulus is presented. seen in most fast ball games, as well as in reaction
Interpreting the Intercept and Slope How situations involved in driving a car, performing
should we interpret the intercept (a) and the various industrial tasks, and so on.
 Human Information Processing 69

Exceptions to Hick's Law Whereas Hick's law and were instructed to press as quickly as pos-
does hold very generally over a wide variety of sible on the one key that vibrated. The relation-
situations and people, one needs to take a number ship between the stimulus (vibrations) and the
of other variables into account in order to accu- response (pressing that finger on the key) was
rately predict choice RT, and a few situations exist very direct. As the number of S-R alternatives
in which the law does not appear to hold at all. increased from two to eight, Leonard found no
The key variable is the subject's familiarity with further increase in choice RT. That is, the relation-
responding to a particular stimulus by means of ship was flat (with b = 0). When the finger was
a specific response. These familiarity effects have vibrated, the relation between it and the response
been studied in various ways, most notably by was so direct that no additional time was required
examining (a) practice or experience with the task for decision making about which finger to acti-
and (b) the nature of the relationship between the vate (see also Keele, 1986, for a good review).
stimuli and the associated responses. Another exception to Hick's law is seen in
In one of the first studies to investigate the role responses involving rapid movement of the eyes
of practice in relation to Hick's law, Mowbray and to targets. In experiments by K veraga, Boucher,
Rhoades (1959) used a two- and four-choice RT and Hughes (2002), subjects responded to stimuli
task and found, as had been observed previously that occurred at from one to eight locations by
by many researchers, that early in practice the making either a manual key press, an eye move-
four-choice task showed a much slower choice ment (saccade) toward the target, or a saccade
RT than did the two-choice task. However, their away from the target. The authors found that
study is unique in that they provided their sub- RTs to initiate manual key presses and saccades
jects with an incredible 42,000 trials of practice! away from the target corresponded well to Hick's
After this amount of practice, the four-choice law. However, saccades toward the target were
RT was reduced to a level essentially equal to completely unaffected by the number of stimulus
that of the two-choice RT. Thus, the slope of the alternatives. This finding agrees well with the
choice-RT function between 1 and 2 bits (i.e., two ideas just discussed, as the saccades toward tar-
and four alternatives) was reduced by practice, gets are much more "natural" (and biologically
eventually becoming essentially flat (where b = important) than either manual responses or sac-
0). Similar findings were also reported by Seibel cades away from a stimulus.
(1963). These effects of practice have been the These findings illustrate the importance
basis for the important notion of automaticity that the relationship between the stimulus and
in responding, which we discuss more fully in response can have to response selection. The
chapter 4 (see Schneider & Shiffrin, 1977). exceptions to Hick's law suggest that highly over-
Another set of data that can be similarly inter- learned S-Rrelationships, or relationships that are
preted was generated by Mowbray (1960). When in some sense biologically "natural," can facilitate
the task involved digit naming (the number of response selection when multiple alternatives are
possible digits was varied, and the RT was mea- available. In the next section we examine how this
sured by a voice microphone), increasing the same influence of S-R familiarity can also have
number of possible digits to be named did not detrimental effects on response selection.
increase the choice RT. The interpretation is that
the names of digits are so highly practiced that Stimulus-Response Compatibility
the association between the digit and the name The association between (or the degree of "natu-
is nearly direct (i.e., automatic), not requiring ralness" between) a stimulus (or set of stimuli)
further reduction in uncertainty (see also Fitts & and the response (or set of responses, called the
Seeger, 1953; Hellyer, 1963). S-R ensemble) has a very important influence on
Similar findings of essentially zero slope for information processing, and was given the term
highly overlearned S-R relationships were pro- stimulus-response compatibility (or simply S-R
vided by Leonard (1959). In this study, choice RTs compatibility) in 1951 by Small (see Small, 1990).
with one, two, four, and eight alternatives were For example, when a right-positioned stimulus
examined, but the situation was quite different requires a right-hand response and a left stimu-
from those studied earlier. The subjects placed lus signals the left, the situation is said to be S-R
their fingers on the appropriate number of keys compatible, as in figure 3.8a. However, when a
70 Motor Control and Learning

left signal indicates a right-hand response and

a right signal a left-hand response (figure 3.8b),
this is said to be S-R incompatible, or at least less
compatible than the arrangement in figure 3.8a.
The pioneering work on S-R compatibility
was conducted by Fitts and his colleagues. An
example is illustrated in figure 3.9, a and b (Fitts
& Deininger, 1954; Fitts & Seeger, 1953). Figure
3.9a illustrates the basic experimental setup used
in these studies. The subject was to move rapidly
and accurately from a center home position to one
of eight target locations available on the response
panel. The specific target was identified by the
stimulus pattern on the display visible to the a b
subject. Figure 3.9a illustrates the general layout,
where the stimulus information about which
FIGURE 3.8 Stimulus-response (S-R) compatibility
response to make is provided via lights, and the is defined by the relationship between the stimuli and the
subject makes a response with a hand movement responses to which they are associated.
on the response panel. Fitts and Seeger used three
different stimulus patterns and three response
patterns in all nine combinations, illustrated in figure 3.9a). In the second row, the stimulus was
figure 3.9b. presented in a different way. If the upper light
The organization of the stimulus patterns (the was lit, it meant "up," and the right light meant
left side of figure 3.9b) to indicate which of the "right." But if the lower and right lights came on,
eight responses the subject was to make worked this meant" down-right." The third configuration
as follows. For the top stimulus pattern in figure was again different. Here, if the upper light of
3.9b, the light that was illuminated indicated the the vertical pair came on, it meant "up," and the
response-such as "up-left," or" down-right" (see right light of the horizontal pair meant "right."

Response patterns

o 0 o
0 0
en
o 0 o
...s:::
!!<a 0
a.
en
::s
0 0
:i
E 0
i5'l
0
o-o I
0
a b

FIGURE 3.9 (a) A subject performing in Fitts and Seeger's (1953) experiment and (b) stimulus-response (S-R) compat-
ibility effects for various stimulus arrays and response panels.
 Human Information Processing 111 71

Here, the left light in the horizontal pair and the the fastest response pattern depended on which
upper light in the vertical pair meant "up-left." stimulus pattern was paired with it.
The working of the response panels is shown on This is the essential idea of spatial S-R com-
the top of figure 3.9b. In the example at the left, patibility. The pattern of the stimulus arrange-
the subject always started in the middle. If the ment seems to be closely linked with the pattern
display signaled "up," the response was up; if it required for the response. The pattern with lights
signaled "left," the move was left. If"down-right" arranged in a circle was fastest when the hand
was signaled, then the movement was along the had to make responses radially, but was relatively
path downward to the right. slow when a two-hand response had to be made.
In the second response panel, the subject again That is, the fastest responses were always on
started in the middle, but this time could move the diagonal in this matrix, where the stimulus
only along the perpendicular pathways. "Up" requirements seemed aligned with the response
meant to move straight up to the center of the requirements.
upper line, and "left" meant to move to the center Next, consider the percentage errors, shown
of the left line. "Down-right" meant to move to also in figure 3.9b, for each of the nine combina-
the lower-right corner, either by moving down tions. Again, the stimulus pattern that had the
and then right or by moving right and then down. fewest errors depended on the response that was
Finally, the response panel on the right used paired with it. Also, the response pattern with
two hands, the right hand on the right half and the fewest errors depended on which stimulus
the left hand on the left half. Each hand started at pattern was paired with it. Note again that the
its respective center point. The stimulus pattern smallest errors were on the diagonal in the matrix
"up" meant to make a vertical response up with where the stimulus and response requirements
the right hand (and not move the left hand), and seemed to be matched. And notice also that some
"left" meant to move the left hand to the left (and combinations resulted in a very high percentage
not move the right hand). Here, though, "down- of errors.
right" meant to move the right hand down and the Finally, note that these effects cannot be
left hand to the right, and similarly for "up-left." explained by a speed-accuracy trade-off, with
Now, consider the data shown in the body of some of the combinations having low RTs only
the table in figure 3.9b-the RTs for the correct because errors happened to be large. To the con-
responses under that combination of stimulus trary, with only minor exceptions, the combina-
and response arrangements, and the percent- tions that made the subjects fastest also made
age errors where the subject moved initially in them the most accurate-the opposite of a speed-
the wrong direction. Consider first the RTs. It is
accuracy trade-off. This was truly a remarkable
interesting to ask which stimulus display pattern
demonstration of the power of having the stimu-
is fastest across all the combinations of response
lus and response patterns "go with" each other.
arrangements. Examining the RTs, it is clear that
This general principle of S-R compatibility has
the fastest stimulus pattern is different for each
become one of the cornerstones of the field of
of the response arrangements; the shortest RT for
human factors and ergonomics, and it pervades
the first pattern (390 ms) is for the first response
the design of countless real-world devices for
pattern; the shortest RT for the second pattern
which subjects must make responses based on the
(410 ms) is for the second response pattern; and
patterning of stimuli. In the following sections we
the shortest RT for the third pattern (480 ms) is for
describe some of the areas in which S-R compat-
the third response pattern. The fastest stimulus
ibility research has been conducted.
pattern depended on which response pattern was
paired with it. Population Stereotypes Many S-R relationships
Next, what is the fastest response pattern? that were originally quite arbitrary "become
The shortest RT for the response pattern on the natural" through practice and experience (e.g.,
left (390 ms) is for the top stimulus pattern; the moving a light switch up to turn a light on) and
shortest RT for the response pattern in the center are called population stereotypes. Different popu-
(410 ms) is for the center stimulus pattern; and lations often have their own unique stereotypes,
the shortest RT for the right response pattern however, resulting in S-R relationships that are
(480 ms) is for the lower stimulus pattern. Again, quite different from others and sometimes even
72 Motor Control and Learning

opposite, such as the movement of the light

switch down to turn the lights on in Germany. 320
This relationship seems clumsy to Americans, but
it is perfectly natural to Germans who have expe- 310
rienced this relationship throughout their lives.
0 300
Spatial and Anatomical Relationships A high .§,
degree of S-R compatibility seems to suggest that Cl)
290
there is an overlap of the mental representation of ~
c:
the task dimensions associated with the stimuli .2 280
't)
and the mental representation of the possible l1l
Cl)
0: 270
responses (Kornblum, Hasbroucq, & Osman,
1990; Weeks & Proctor, 1990). Data of the type 260
shown by Fitts and Seeger (1953) and many others
suggest that S-R compatibility occurs because of 250
spatial similarities in the stimulus and response
sets. One possible reason for the S-R compat- Compatible Incompatible
ibility effect in figure 3.8 could be that the right Spatial S-R compatibility
limb responds to the right light faster than to the
left light because of some anatomical or neural FIGURE 3.10 Effects of stimulus-response (S-R) com-
advantage. If this is so, then imagine what would patibility and hand location on reaction time (RT).
happen if the hands were crossed, so that a sub- Data from Brebner et al. 1972.

ject's right hand was used to press the left button

and the left hand used for the right button. This resents another type of S-R compatibility effect.
crossed condition is now compatible with respect Reaction times are smaller and errors are fewer
to the spatial mapping of the stimulus and the key when the task is to make the natural response to
to select (as before), but incompatible with respect look at the source of a visual stimulus as compared
to the mapping of the stimulus and hand to select to an anti-saccade task, which can be thought of
(i.e., the right stimulus is responded to with the as an incompatible S-R relationship (see Munoz
right key, but using the left hand). & Everling, 2004, for a review). In the case of the
Experiments that have addressed this issue anti-saccade, the natural tendency-that is, to
(Anzola, Bertolini, Buchtel, & Rizzolatti, 1977; look at the source of the visual stimulus-must
Brebner, Shephard, & Cairney, 1972; Wallace, 1971) be actively suppressed, or inhibited, before the
reveal that it is the spatial relationship between correct response can be made. Presumably, this
the location of the stimulus and the location of the inhibition requires time.
response key that is crucial. Regardless of which Stimulus and Response Intensity Another
hand is used (i.e., crossed or uncrossed condi- rather straightforward type of S-R compatibil-
tions), performance is faster in the spatially com- ity relationship has been found when the force
patible conditions (ensemble a in figure 3.8) than of a response is mapped to the intensity of the
in spatially incompatible conditions (ensemble b stimulus. Su!::>jects in a study by Romaiguere,
in figure 3.8). This conclusion is supported well Hasbroucq, Possama'i, and Seal (1993) were asked
by the findings of Brebner and colleagues (1972), to make a weak or strong isometric thumb press
shown in figure 3.10. In general, the crossed-hands in response to a visual stimulus that varied in
procedure resulted in slower overall RT than did intensity. The authors found a 50 to 75 ms RT
the uncrossed condition. However, regardless advantage in the compatible condition (in which
of the position of the hands, when the stimulus stimuli low in intensity were paired with weak
light was paired with its spatially compatible thumb movements and strong stimuli with strong
response-key position, the RTs were faster than movements), compared to the incompatible S-R
in the incompatible positions (figure 3.10). condition (low intensity paired with strong move-
Earlier we discussed an exception to Hick's ments). Note, however, that the relation between
law-making saccades toward visual stimuli the stimulus and the response here is more
(Kveraga et al., 2002). Directing the eyes in abstract than the more direct physical compat-
response to the presentation of a visual target rep- ibility relationship that we have considered so far.
 Human Information Processing 73

A related S-R compatibility effect occurs when (see Umilta, Priftis, & Zorzi, 2009, for a review).
different grip actions are made in response to Some of these types of S-R compatibility effects,
numbers that differ in their magnitude. The left which appear to be more "conceptual" than
side of figure 3.11 illustrates the two types of physical (by definition), are considered in the
responses made by subjects in an experiment by next sections.
Lindemann, Abolafia, Girardi, and Bekkering
Simon Effects We can also see the influence of
(2007)-a precision grip (a pinching action used
S-R compatibility on response selection when the
to pick up small or light objects) or a power grip
spatial dimension of the stimulus is irrelevant to
(a whole-hand grip used to move large or heavy
objects). Lindemann and colleagues used a clever response selection, not unlike the S-R compatibil-
design in which these specific grips were used to ity effects discussed in the previous sections. In
respond in a choice-RT task regarding the features an early demonstration of what were later called
of a number, presented as the visual stimulus. For Simon effects, subjects were asked to respond by
example, if the number 2 or 8 was presented (both pressing a right key whenever the verbal stimu-
even numbers), subjects in one condition were to lus "right" was presented via earphones, and by
respond with a precision grip; if the number 3 or pressing the left key when the word "left" was
9 was presented (odd numbers), then a power presented (Simon & Rudell, 1967). The auditory
grip was required. Quite incidental to making stimuli were sometimes presented only to the
the response was the effect of the magnitude of left ear and sometimes only to the right ear. Note
the number and its "compatibility" with the that regardless of the ear to which the stimulus
response. As illustrated in the right side of figure was presented, subjects were to respond to the
3.11, RT was lower to initiate a precision grip content of the message; the spatial location (ear)
in response to a low number (2 or 3) compared to which the message was sent was irrelevant to
to a high number (8 or 9). Conversely, RT was making the correct response. However, the find-
longer to initiate a power grip to a low number ings revealed that the irrelevant (spatial) feature
compared to a high number. The compatibility of the stimulus had a profound effect on response
of specific grips to the magnitude of a number selection and could not be ignored. As figure 3.12
may be an example of a more general type of shows, choice RT was faster when the message of
conceptual S-R compatibility effect known as the the signal and the response were compatible with
SNARC effect (the "spatial-numerical association the spatial origin of the stimulus. These findings,
of response codes" effect), which describes a rela- along with others reported since Simon published
tionship between space and number magnitude his initial series of studies, suggest that there is

Precision grip 640

(j)
.§.. 630
Q)
E
~ 620
0
~
ca
Q)
610
a:
600

Power grip
Small Large
a b Number magnitude

FIGURE 3.11 (a) Subjects in the study by Lindemann and colleagues (2007) were cued with a low or high number to pick
up an object using either a pinch grip or a power grip; (b) RTs were fastest for large numbers using a power grip, but were
fastest for small numbers using a pinch grip.
Reprinted, by permission, from 0. Lindemann et al., 2007, "Getting a grip on numbers: numerical magnitude priming in object grasping,'' Journal of Experimental Psychology:
Human Perception and Performance 33: 1400-1409.
74 Motor Control and Learning

Interference is likely to occur when the response

430
is inconsistent with the learned association (for
in-depth discussion of theoretical interpretations
420 of various S-R compatibility effects see Cho &
Proctor, 2003; Hommel & Prinz, 1997; Kornblum
410
et al., 1990; Proctor & Reeve, 1990; Proctor & Van
! 400 Zandt, 1994).
l-
a: 390 Compatibility and Complex Actions When
people respond to stimuli that require complex
380 actions, the issue of S-R compatibility and
response selection seems to be related to the inten-
370
tions of the action. For example, Rosenbaum and
colleagues (1990) describe the everyday task of a
waiter who wants to turn over an upside-down
Left Right glass with one hand so that he can pour water
Ear presented with message from a pitcher into it with the other hand. What
will be the initial grasp of the hand-thumb up
FIGURE 3.12 The Simon effect-the irrelevant stimulus or inverted with the thumb down? Experiments
(which ear receives the message) interacts with the infor- show that the waiter's initial hand position is
mation.
usually inverted so that when the glass is turned
Adapted, by permission, from J.R. Simon and AP. Rudell, 1967, ':Auditory S-R
compatibility: The effect of an irrelevant cue on information processing," Journal upright the hand is in the correct holding position,
of Applied Psychology 51: 302. ready to receive poured liquid. The response ini-
tially selected by the waiter "trades off" an awk-
interference in selecting a response when the pre- ward early posture in order to achieve a posture
sumably irrelevant stimulus is incompatible with that is more suited to filling the glass with water.
the response. The spatial dimensionality of the In this case, compatibility between the stimulus
stimulus tends to attract the response toward it, (the glass) and the waiter's response (initial hand
which must be inhibited (thus taking extra time) posture) is defined not in terms of the initial
before the correct response can be selected (see interaction, but rather in terms of the efficiency
also Simon, 1969a, 1969b, 1990). of the intended final position (what Rosenbaum
Although the spatial layout of the S-R ensem- and colleagues have termed the "end-comfort
ble has a critical effect on response selection, this effect"). Also, selection of the method that maxi-
appears to be only a partial explanation of what is mizes the efficiency at movement completion is
happening (Umilta & Nicoletti, 1990). For exam- initiated faster than of one that is more efficient at
ple, choice RT to the stimulus word "right" or the action's start (Rosenbaum, 2009; Rosenbaum,
to a rightward-pointing stimulus arrow is faster Vaughan, Barnes, & Jorgensen, 1992; Rosenbaum,
when the subject responds with a button press van Heugten, & Caldwell, 1996), especially if the
with the right hand or with the word "right" in final location has a stringent precision require-
comparison to arrangements with incompatible ment (Short & Cauraugh, 1999). Thus, for more
relationships (e.g., McCarthy & Donchin, 1981; complex actions, S-R compatibility effects may
Weeks & Proctor, 1990). Thus, similar to the effects influence the response-selection stage in terms
of number size, verbal stimuli that implicate a of how an action is to be performed, rather than
spatial relation and responses appear to have simply about where to move.
dimensions that produce compatibility relation-
ships affecting response selection. Response-Programming Stage
Together, these findings suggest that the The performer can organize and initiate an action
response-selection stage is susceptible to com- only after having identified the stimulus and
patibility effects when the relationship between selected a response. After response selection,
the set of stimulus alternatives and the set of the performer must translate this abstract idea
response alternatives has a highly learned (or into a set of muscular actions that will achieve
"natural") association (as in the word "right"). the desired action. These processes are thought
 Human Information Processing 111 75

to occur during the response-programming stage. series of triggers on a conveyer belt (as in a player
Like the processes in the earlier stages, the piano), with more triggers required for more
events occurring in response programming are complex movements (Klapp, 1996). Although
probably very complex-requiring that some the language and analogies have changed over
program of action be retrieved from memory, that the years, it is interesting and useful to read the
the program be prepared for activation, that the theoretical arguments that Henry and Rogers
relevant portions of the motor system be readied (1960) presented in their paper and to understand
for the program (called feedforward, or tuning), why RT was considered a valuable measure of
and that the movement be initiated. It is helpful motor program retrieval and preparation (see
to view the response-programming stage as the "Henry and Rogers' 'Memory Drum' Theory of
final set of processes that allows the individual N euromotor Reaction").
to communicate with the environment-just as
the stimulus-identification stage is viewed as the The Henry-Rogers Experiment
first stage that allows the environment to com- Henry and Rogers (1960) studied the nature of the
municate with an individual. movement to be made using a simple-RT para-
Information about the response-programming digm, in which the subjects knew, on any given
stage has developed more recently than research trial, exactly which response was required. Over
regarding the other two stages. It was not until many blocks of trials, Henry and Rogers had the
1960, when Henry and Rogers performed an subjects make different movements while keep-
experiment on the nature of the movement to be ing the stimulus for the movement, as well as the
produced in RT situations, that conceptualization number of response alternatives, constant (i.e.,
about the response-programming stage began. there were no stimulus-response alternatives, or
The underlying motivation for their experiment N = 0). The apparatus used for the Henry and
was based on Henry's idea that motor programs Rogers experiment is illustrated in figure 3.13. The
are stored in memory. When a well-learned action first task version (movement A) involved merely
is retrieved from memory, it is "read out" like a lifting the finger from a key a few millimeters

:. :. ti~h$ttir,e
.· .n}~t6t ~kill~.~~'.... . ~ thi~ :.~re iti~ate. o~µromotor' .. 9rt:lil'l~tron . . ..·~re::m1p9~tra!itlJ;] rnot6i:aet$: ·: ·
the ~EIPPihg Ott~~ !$tc}r13 rna:V be thpugof of broadly.Els.~ rnefl1ory. s\orEIQe cirµn;i p~e~oJTl~non;,t9 '·
us~ tb~. ani:dogy of the electronic compute!-. The neuralpattern for a .sp139ific andwe11~cc)9rciir1ated
motor acfis. controlled by a stored program that is used to direct the neuromotor details oHts
performance . ~ . there.should be a longer reaction latency fora compHeated movementthan
for a simpler movement. This .is because a more comprehensive program,. i.e., a larger amount
of stored information, will be needed, and thus the neural impulses will require more.time for
coordination and direction into the eventual motor neurons and muscles" (pp. 449 and 450).
It was not until years later that Henry and Rogers' ideas became widely cited (Cardinal &
Thomas, 2005; Ulrich & Reeve, 2005). These ideas about motor programs are quite prevalent
and useful today, as will be evident in later chapters here.
76 Motor Control and Learning

versions was in the nature of the movement to be

made. The primary measure, as before, was the
(simple) RT, or the interval from stimulus onset
until the movement began.
Henry and Rogers' data from the adult subjects
are presented in table 3.3. Movement B (single-
ball grasp) resulted in a 36 ms longer RT than
did movement A. The most complex movement
(C) (ball strike, button press, ball grasp) resulted
in an additional 13 ms increase in RT compared
to movement B. Because the stimuli were not
changed for the different movements, nor was
the number of choices, these data suggest that
FIGURE 3.13 This apparatus, from Howell (1953),
the increased RT as the movements increased in
was almost identical to the apparatus used by Henry and complexity was due to an increased amount of
Rogers (1960). Explicit details about the apparatus and the time required to program the movement in some
three movement tasks are reported in Fischman, Christina, response-programming stage. Henry and Rogers'
and Anson (2008). (A= tennis balls, B = nylon strings, C =
base, D = reaction time, E = button, F =friction contacts, G
original idea was that a more complex set of
= sidearm, H = stimulus lights.) motor commands, such as would be necessary to
Reprinted with permission from Research Quarterly for Exercise and Sport, Vol. 24, control the limb through several movement rever-
22-32, Copyright 1953 by the American Alliance for Health, Physical Education,
Recreation and Dance, 1900 Association Drive, Reston, VA 20191.
sals and to produce striking and grasping actions,
would require more neuromotor activities to be
coordinated, in turn requiring more time for all
and had essentially no accuracy requirement. of this neurological complexity to be organized
For the second version (movement B), the subject during RT. In slightly modified form, this idea still
lifted the finger from the key and moved approxi- has a great deal of support (see Christina, 1992;
mately 33 cm forward and upward to grasp a Henry, 1980; Klapp, 1977a, 1996).
tennis ball suspended on a string, which stopped What is going on in the Henry and Rogers
a timer that measured movement time (MT). experiment? The obvious conclusion is that RT
The third task version (movement C) involved a increased as the complexity of the movement to be
second suspended ball mounted 30 cm to the left made increased. But what was actually involved
of the first ball. The subject lifted the finger from in making the movement more "complex"? The
the key, moved upward to the right to strike the careful reader will notice a number of factors that
first ball with the back of the hand, moved for- came into play in this regard. Table 3.3 also sum-
ward and downward to push a button, and then marizes a number of potentially relevant factors
moved upward to the left to grasp the second that differed according to the three experimental
suspended ball. 2 Remember, for all three task conditions in the study. The differences between
versions, there were no stimulus and response movements A, B, and C included additional move-
alternatives for a block of trials. Thus, the pro- ment parts, increased accuracy demands, and
cessing speed in the stimulus-identification and longer movement durations (and perhaps other
response-selection mechanisms should always be factors as well). Research conducted since this
the same; the only variation among the three task pioneering experiment suggests that all of these
 Human Information Processing 77

factors may have played a role in producing the requirements of the task increased (Fitts & Peter-
results seen in the Henry and Rogers experiment. son, 1964). Findings that RT increases as the target
Number of Movement Parts and Reaction Time size decreases (thereby increasing the accuracy
Results similar to the Henry and Rogers (1960) demands) have since been reported in a number
data have been obtained in a number of replica- of experiments by Sidaway and his colleagues
tions of their study, implicating the addition of (e.g., Sidaway, Sekiya, & Fairweather, 1995).
movement parts as the primary reason for the Movement Duration Effects on Reaction Time
increase in RT (e.g., Christina, 1992; Fischman, In addition to complexity and accuracy
1984). Corroborating evidence was obtained from demands, the three movements studied by
experiments using different tasks. For example, Henry and Rogers (1960) varied in at least one
Sternberg, Monsell, Knoll, and Wright (1978) other important respect-their duration. From
observed that the latency in speaking the first Henry and Rogers' data (table 3.3), notice that
word of a sequence increased by about 10 ms for the most complex movement required much
each word that was added to the sequence (i.e., more time to produce (465 ms) than the simpler
strings of one, two, three, four, or five words), and one (95 ms). This observation has led various
by 5 to 15 ms for additional letters to be typed scientists to suspect that the duration of the
(see also Canic & Franks, 1989). It would appear, movement to be produced might be a major
however, that the additional movement must be variable in the response-programming stage.
performed as part of a whole ballistic response in Klapp and Erwin (1976) asked subjects to make
order to have an effect on RT. A study by Franks, 10 cm movements of a slide along a track, with
Nagelkerke, Ketelaars, and van Donkelaar (1998) goal MTs of 150, 300, 600, or 1,200 ms. The
compared the effect on initiation latency of a number of actions and the accuracy demands
single-movement task (elbow extension) and a of the movement were held constant, but the
two-movement task (elbow extension and flex- duration of the movement was varied. As the
ion) in which there was either no pause, a short movement duration increased, the RT to initiate
pause (75 ms), or a long pause (260 ms) between the response increased as well, especially when
extension and flexion. Elevated RTs were found the movement durations were below 600 ms.
when the flexion movement followed without Similar effects were reported by Rosenbaum and
delay or with a very brief delay, but not when Patashnik (1980), who varied movement dura-
there was a long delay between the two move- tion of the time a button had to be depressed,
ments. These results suggested that when there and by Quinn, Schmidt, Zelaznik, Hawkins, and
was sufficient time to program the second move- McFarquhar (1980), who varied the movement
ment following completion of the first movement time for aiming responses of a stylus to a target
(and thus not during the original RT), the subject (see Klapp, 1996, for a review).
needed only to plan the first movement prior to Response Complexity
movement onset, regardless of whether or not and Motor Programming
it was followed by a flexion. However, with no
The effect of movement complexity on RT occurs
pause or a brief pause between the two parts, both when the person knows in advance what
the entire response needed to be programmed in movement will be made (in simple RT) and when
advance, resulting in the observed effect on RT of the choice of movement is not known until it,
the additional movement part (see Klapp, 1996, is indicated by the stimulus (in choice RT) (see
for further discussion). Klapp, 1995, 1996). Regardless of the variations in
Movement Accuracy Effects on Reaction Time method and movements, the effect of movement
At about the same time as Henry and Rogers' complexity on RT has been interpreted as relating
work (but quite separate from it), Paul Fitts was to the time necessary to prepare and initiate the
investigating movement complexity effects as movement during the response-programming
well, using aiming tasks that varied in target stage of RT. Despite over 50 years of research on
distance and accuracy demands. Although Fitts' this topic, the ideas of Henry and Rogers about
primary concern related to the ongoing move- the process of movement organization during the
ment (as revealed by movement time; see chapter motor-programming stage remain remarkably
7), he also found that RT increased as the precision valid today.
78 Motor Control and Learning

Anticipation own movement will take, and often he must

allow for this interval in initiating the actions;
the batter must time the bat swing's initiation so
So far in this chapter, we have discussed that the bat and ball meet at the proper location
information-processing activities in which the over the plate. This is called effector anticipation,
performer's task is to respond to generally unan- because the duration of the effector's movement
ticipated stimulus information. The processing (the body, mainly the arms and bat) must be pre-
speed, as measured by RT, tends to be relatively dicted. Finally, Poulton described what he called
slow, giving the impression that humans have perceptual anticipation; here the environmental
severe limitations in information processing. events are not perceived directly, but they are still
One problem with this line of thinking, though, predictable because the performer has had a great
is that RT is usually studied in highly unrealistic deal of practice with them, such as the regular
situations, and especially in situations in which timing of beats in music or in military drill. In the
the subject is seldom allowed to anticipate envi- next sections, we consider important principles
ronmental information. In fact, experimenters related to a performer's capability to anticipate.
go to elaborate lengths to prevent the subjects
from anticipating. They often use "catch trials," Spatial (or Event) Anticipation
in which the stimulus is sometimes withheld to
"catch" a subject who is anticipating; often use One way in which performers can anticipate
choice-RT tasks, preventing the subject from future activities is by knowing what kinds of
knowing what stimulus is going to occur; or ran- stimuli could be presented and what kinds of
domize foreperiods (the interval from the prepa- responses will be required for each of the antici-
ratory signal to the stimulus presentation) so that pated stimuli. This class of anticipation has been
the subject cannot predict when the stimulus will called spatial, or event, anticipation. In an experi-
appear. By employing these methods, research- mental study of these processes, Leonard (1953,
ers try to ensure that subjects are responding to 1954) and Jeeves (1961) used an apparatus that
an "unanticipated" signal. Clearly, under these had trackways arranged as spokes of a wheel.
conditions, humans find information processing Subjects were asked to move repeatedly from
very difficult indeed. the center position toward lights at the ends of
But in many "real" skills, suddenly presented the spokes and back again as quickly as pos-
and unexpected stimuli are the exception rather sible. The light indicated the spoke along which
than the rule. Of course, unexpected events do the subject was to move. If the next spoke was
sometimes occur-and sometimes we even try indicated only after the subject had arrived back
to make them occur (e.g., faking in sports); but at the center position from the previous spoke,
many of the stimuli to which we respond in subjects could not anticipate which movement to
our daily activities are very predictable. During make next, and performance was slow, jerky, and
walking or driving, the stimuli emerge from a labored. But if subjects were informed about the
generally stable environment that allows us to next spoke when they were at the peripheral end
preview upcoming events with plenty of time of the previous spoke, they could plan the next
to do something about them. And when a signal movement while they were moving back to the
does arrive, it is often not a discrete event, but center. Overall, performance was smoother, less
rather a pattern of sensory information that jerky, and more rapid. Analogous effects were
unfolds before us. Gibson (1966) has emphasized found by Leonard (1953), whose subjects used
that visual information is really an optical flow of advance information to reduce a six-choice RT
stimuli, from which the performer detects impor- task to a three-choice task, with associated gains
tant future environmental events (see chapter 5 in speed as would be expected from reducing
for a discussion of this idea). the number of S-R alternatives from six to three
Poulton (1957) has described three different (conforming to Hick's law). Many other experi-
kinds of anticipation. One obvious type is receptor ments led to a similar conclusion (see Schmidt,
anticipation, in which the performer detects the 1968, for a review).
upcoming events with various sensory receptors Some experimenters have sought to determine
(e.g., the batter sees the ball coming). However, what kinds of information about the upcom-
the performer must also estimate how long his ing movement can be used in advance and how
 Human Information Processing 79

much time can be "saved" by using it. Using work, using a different paradigm, makes a simi-
the precuing technique, Rosenbaum (1980, 1983), lar point.
Goodman and Kelso (1980), Zelaznik and Hahn One way to think of these effects is illustrated in
(1985), and others have examined tasks in which figure 3.14. If the subject receives advance informa-
various aspects of the response could be specified tion about a certain feature of the movement, some
in advance, leaving other aspects unspecified of the processing operations normally done during
until the stimulus actually arrived; the stimulus RT can be done in advance, and can be "bypassed"
then provided the remainder of the information when the reaction stimulus finally arrives. If suf-
needed to produce the action, as well as the "go" ficient information is given so that all the aspects
signal. For example, Rosenbaum (1980) used a can be selected in advance (e.g., knowing arm,
task in which the response involved up to three direction, and extent in Rosenbaum's situation),
types of movement choices: (a) with the right then one can think of "bypassing" processing in
or left arm, (b) toward or away from the body, response selection completely, resulting in what is
and (c) to a target that was near to or far from essentially simple RT. The work of Rosenbaum and
the starting position-an eight-choice RT task others suggests that portions of response program-
(2 X 2 X 2 = 8 alternatives). Rosenbaum found ming can be done in advance as well.
that providing advance information about any It is also interesting to note in Leonard's (1953,
one of the three movement features (arm, direc- 1954) and Jeeves' (1961) studies that advance
tion, or extent) reduced the RT by about 100 to information allowing partial or complete selec-
150 ms. Apparently, when the subjects had this tion of upcoming actions was presented while
advance information they could engage in some the subject was already moving back to the center
response processing before the stimulus arrived, position. The result should not be particularly
thus "saving" processing time during RT. There surprising, but it does show that one movement
seemed to be a greater advantage in receiving can be planned while another is being executed.
information about which arm was to be used This is a very important feature of motor control,
(150 ms reduction in RT relative to the no-precue as it provides a mechanism by which a series of
condition) as compared to information about rapid movements can be executed with consider-
the extent of the movement (100 ms reduction), able speed and accuracy.
suggesting that the situation is somewhat more Anticipating the spatial location of a directed
complex than simply reducing the number of penalty kick in soccer is one example of a situation
alternatives from eight to four. Klapp's (1977b) in which advanced information processing is

No advance information

Advance information available

FIGURE 3.14 Bypassing the response-selection stage by processing information in advance.

80 Motor Control and Learning

necessary if a goalkeeper is to have any chance to (provided that the subject knows which response
be successful (Franks & Harvey, 1997). Analyses to produce). Quesada and Schmidt (1970) showed
of the soccer penalty kick have shown that, if a that the average RT with a constant 2 s foreperiod
goalkeeper waits until the ball is contacted before was only 22 ms! It seems likely that the person
deciding which direction to dive, it is virtually anticipated the temporal onset of the stimulus and
impossible to intercept the ball, since the travel began the process of programming and initiating
time for the ball to reach the goal line (about the response before the stimulus came on so that
600 ms) is roughly equal to the goalkeeper's the overt movement occurred at about the same
own movement time (500-700 ms) after the deci- time as the stimulus. Thus, if the foreperiods
sion to move has been made (Franks & Harvey, are both regular and short, and all aspects of the
1997). Therefore, if she is to be successful, the response to be made are known in advance, then
goalkeeper must pick up advance cues from the subject can (with very little practice) perform
the kicker regarding the direction in which the the needed processes in advance and emit the
ball will travel and must use that information to overt response essentially simultaneously with
decide on a response prior to the actual delivery the stimulus. This has been termed early responding
of the stimulus (contacting the ball). Analyses of in the literature because the response is triggered
this task have revealed that expert goalkeepers early-before the stimulus actually arrives.
focus on the position of the stance foot (the non- On the other hand, when the foreperiod is
kicking leg) and the relation of both feet to the regular but very long (a few seconds or more),
ball as advance cues, whereas novice goalkeep- and various features of the response are known in
ers use visual search strategies that are much advance, subjects apparently cannot shorten RTs to
less reliable and useful (Franks & Harvey, 1997; zero even with extensive practice. Under these con-
Savelsbergh, van der Kamp, Williams, & Ward, ditions, Mowrer (1940) found RTs of about 230 ms.
2005; Williams, 2000). The RTs in these situations with long but regular
foreperiods seem to be similar to those in which the
Temporal Anticipation foreperiod is short but irregular (thus preventing
The evidence just reviewed on spatial anticipa- early responding). When the foreperiods are long
tion suggests that the performer can, by knowing (e.g., 12 s), early responding is prevented because
some specific information about the response to be the subject cannot anticipate the exact stimulus
produced, bypass or at least shorten some of the onset when it is so far in the future. The reason is
stages of information processing. This shortening that the internal "timing" for short durations is
of RT is rather modest, though, and the responder much less variable than that for long durations. 3
still has difficulty processing environmental stimuli Attempts to anticipate the stimulus onset following
quickly. The evidence presented in the next sections a long foreperiod result in too many (very) early
suggests that if the person can anticipate when the responses (by a second or so), which is usually not
stimulus is going to arrive, rather large reductions allowed by the instructions or by the experimenter.
in RT can occur. Under the proper circumstances, All the subject can do is engage in preparation and
the performer can eliminate RT altogether! respond very quickly when the stimulus does
arrive. Therefore, in situations involving anticipa-
Constant-Duration Foreperiods tion, a common question is whether it is prepara-
Imagine a subject in a simple-RT situation (one tion or early responding that is taking place.
stimulus and one response) in which a warning
signal is followed by a foreperiod, at the end of Variable-Duration Foreperiods
which is the stimulus onset. Foreperiods may be of In simple-RT situations, there is a great deal of
a constant duration (e.g., always 3 s), or they may evidence that irregular (variable duration) fore-
be variable and unpredictable (e.g., 2, 3, or 4 sin periods averaging about 1 s produce shorter RTs
random order). It seems obvious that a constant- than do longer ones of 2, 3, or 4 s (e.g., Klemmer,
duration foreperiod will result in the shortest RTs. 1956; Welford, 1968). This effect seems to be quite
If these constant foreperiods are relatively short small, however, and it is overshadowed by a larger
(e.g., less than a few seconds), evidence shows effect that is apparently related to when the sub-
that the subject can respond essentially simul- ject expects the signal. In these latter situations,
taneously with the stimulus after some practice the fastest RT is not associated with the shortest
 Human Information Processing 81

foreperiod as would be expected from Klemmer' s include no catch trials (again, in catch trials, the
results, but rather either with the most probable stimulus-though expected-never arrives;
foreperiod, or, if the foreperiods are all equally e.g., Drazin, 1961; Rothstein, 1973; Salmoni, Sul-
probable, with the center of the range of forepe- livan, & Starkes, 1976). Drazin's data are shown
riods (Aiken, 1964; Mowrer, 1940; Poulton, 1974). in figure 3.16, which plots the RT against the
Subjects in Mowrer's (1940) study responded average foreperiod duration. Here there were
to tones presented every 12 s, but occasionally variable foreperiods and no catch trials, and the
Mowrer presented tones at greater or less than 12 RT decreased as the stimulus was presented later
s intervals. The data in figure 3.15 show that the and later in the group of foreperiods. At first, this
shortest RTs were at the interval that was most result seems contradictory to the earlier conclu-
expected (12 s), and that longer and especially sion that a stimulus presented at the center of
shorter intervals resulted in slower RTs. It appears the foreperiods elicits the most rapid RT (Aiken,
that as the end of the expected foreperiods draws 1964; Mowrer, 1940).
near, the subject begins to prepare for the stimulus
The critical difference is that the studies by
and response. Because maintaining a prepared
Drazin (1961), Rothstein (1973), and Salmoni and
state is effortful (and perhaps not possible during
colleagues (1976) did not employ catch trials.
these longer intervals), this readiness begins
Without catch trials, the subject can become
to increase only when the first of the group of
increasingly expectant for the stimulus as the
stimuli is expected, reaches a maximum at about
the center (the most frequently occurring forepe- foreperiod ages" toward the last possible time of
/1

riod), and declines toward the end. Presumably, stimulus presentation. Consider an example with
the subject is most expectant for the signal when four possible (and equally probable) foreperiods
it is presented with an average foreperiod, and (1, 2, 3, and 4 s) and no catch trials. Note that when
the RT is somewhat faster as a result. Clearly, this only 0.5 s of the foreperiod has elapsed, the prob-
expectancy speeds RT. ability that the signal will appear at the 1 s point is
one in four (0.25). After the 1 s interval has passed
Aging Foreperiods without presentation of the signal, the probability
We can see a notable exception to the findings that the signal will arrive at 2 s is one in three
on foreperiod duration in experiments that (0.33), and so on, until beyond the passage of 3
s the probability of the signal's arriving at 4 sis

300

290 230

280 220

270 210
Ui' Ui' 200
.§. 260 .§.
I- I-
CC cc 190
250
180
240
170
230
160

3 6 9 12 15 18 21 24
0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
Intervals between successive tones (s)
Average foreperiod duration (s)

FIGURE 3.15 Minimum reaction time occurs at the most FIGURE 3.16 The aging-foreperiod effect.
probable interstimulus interval.
Reprinted from D.H. Drazin, 1961, "Effects of foreperiod, foreperiod variability,
Reprinted from O.H. Mowrer, 1940, "Preparatory set (Expectancy): Some methods and probability of stimulus occurrence on simple reaction time," Journal of
of measurement," Psychological Monographs 52 (233): 12. Experimental Psychology 62: 45.
82 Motor Control and Learning

1.0. Thus, the subject has a basis for becoming prevent early responding by using a variable-
increasingly expectant as the foreperiod "ages." duration foreperiod together with some "catch
trials," which requires that the athletes wait for
Temporal and Spatial Anticipation: the signal before initiating their response (see
Implications "Foreperiods and False Starts in Sprinting").
On the other hand, the dance instructor or the
Essentially, anticipation is a strategy to reduce drill leader in the military uses predictable fore-
the time, or even the stages, of processing that periods so that all the performers can respond
would normally be involved in responding simultaneously with the count and with one
to an unanticipated stimulus. For this reason, another. Dance instructors present a 1-2-3-4 count
in some situations people make attempts to that the pupils know not to respond to, then a
either prevent or enhance such anticipation. For second 1-2-3-4; the "l" of the second count is the
example, the starter in a sprint race attempts to stimulus to begin. Drill leaders issue a command

Foreperiods and False Starts in Sprinting

Arare event occurred in the 1996 Summer Olympic Games~a sprinter false-started twice and was
disqt:1alified from the 100 m final. In the second of his false starts, British sprinter Linford Christie
appeared to explode out of the blocks with an excellent and legal start. Moments later the start was
ruled invalid because Christie had apparently "jumped the gun." His RT, as indicated by the time
from the starter's signal until the sudden rise in force of his toot against the blocks, was shorter than
the 100 ms minimum value allowed by Olympic standards. Some raised arguments at the time that
such standards place too fine a constraint on what might otherwise be an exceptional performance.
Was)his a fair judgment?
The starter's goal in the sprint is to have all the athletes reacting to the sound of the gun, and not
responding before it. Forthis reason the starter attempts to prevent anticipations by using variable
foreperiods and by sometimes aborting a start when the foreperiod ages too much (a type of catch
trial). Whatthe starter observes in the sprint start is forward movement of the athlete in relation to
the sound. When that movement occurs before the sound or coincides with it, it is easy to inter that
the athlete has anticipated, and the sprinter is rightfully penalized with a false start. The more dif-
ficultpecisio!), however, concerns penalizing the sprinter who has initiated the start after the sound
of the gun, but not ih reactionto it. In other words, how fast is too fast?
~wnerousexperimenters have studied how to optimize RT under various experimental constraints.
SirrtpleHTstypically rc:inge from about 130 ms to 200 ms or so, depending on the intensity of the
stimulus, the.effector used in the response, and so on. In the case of the sprint start, however, the
t.irne measured is from the presentation Of the stimulus until the first change in force against. the
~t~rting block, and additional timeis required to actually start the body.in motion so that it is vis-
ible to the starting official. Also, consider that the laboratory RT values have been achieved under
verycontroUed experimental conditions with responses involving very simple movements (such as
finger lifts). The sprint start entails a much more complex coordination of effectors and much larger
forces. Consequently, an even longerRT than is typical..,--not a shorter one-might be expected in
the sprintsituation.
Still, one could argue thatttie RTs of world-class sprinters might be expected to be faster than
those oftheaverage subjects in an RT experiment. However, though sprint training does improve
mahy speed-related factors, there is only so much· that can be shaved off the stages of processing
involyedinsimple RT. The minimum of 100 ms from stimulus to force change as an RT for a sprint
startis a conservative criterion that still leaves plenty of time for some anticipation (perhaps up to
100 ms). Anything faster is mostlikely due to anticipating and "early responding;' as discussed in
this chapter. When all of.the factors affecting sprint RT are combined, we must conclude thatChristie
was almost certainly guilty of a false start.
 Human Information Processing 83

(e.g., "Left ... ") to provide the soldiers with event of estimating the advantages and disadvantages
predictability and then later, at a very predictable of anticipating called a cost-benefit analysis.
time, give another command-"Face!"-which In the study by Posner and associates (1978),
is the stimulus to perform the action (a 90° left the subject fixated on the center of a screen and
turn); a good unit will respond as a single person, received one of three precues. One second after
right on command. Anticipation is also very the precue, a signal would come on at one of
important in defensive driving skills. Young and two locations on the screen (which could be seen
Stanton (2007) provide a good review of various without an eye movement), and the subject's task
individual, vehicle, and situational factors influ- was to lift a finger from a key as rapidly as pos-
encing brake response times in driving. sible after stimulus onset. Only one response was
These concepts are also evident in American ever required (lifting a single finger from a key)
football, in which the quarterback provides a set regardless of the stimulus location. One of the
of signals for his teammates just before the ball is precues was a plus sign, presented on one-third
snapped. The basic idea is for the quarterback to of the trials, indicating an equal probability that
enable his team to anticipate and begin moving either of the two signal locations could be used
before the opposition can do so. According to (these were called "neutral-precue trials"). On
the principles discussed in the previous sec- the remaining two-thirds of the trials, however,
tions, the signal count should be predictable, the precue was an arrow pointing to the left or
allowing temporal anticipation. For example, to the right, meaning that the signal would be
the quarterback could bark out distinctive, sharp presented on the side of the screen to which the
sounds (such as "two, two, two, two, ... ")such arrow pointed; however, this precue was correct
that the timing between the sounds is regular on only 80% of the trials. On the remaining 20%
and predictable. Earlier (in the huddle), the team of the trials, the subject was "tricked"; the signal
will have been informed that the second "two" is would arrive on the side of the screen opposite the
the signal to start the play. This allows all of the one indicated by the arrow. The trials in which the
offensive team to have nearly perfect temporal signal arrived on the side indicated by the arrow
and spatial (event) anticipation because they were called valid-precue trials; those in which the
know what is to be done and when to do it. The arrow pointed away from the eventual signal
opposition does not have this capability, which location were called invalid-precue trials.
forces them to be delayed by at least one RT Posner and colleagues (1978) found that the
before responding. The signal caller should be
average RT on the neutral-precue trials was 265
careful to avoid the aging-foreperiod effect, as
ms. On the valid-precue trials, the average RT
an alert defense could predict the temporal onset
was 235 ms, revealing a 30 ms benefit to perfor-
of the "two" signal with increasing certainty as
mance for a correct anticipation, relative to the
the count "ages," since the rules do not allow
neutral-precue condition. However, when the
for "catch trials."
signal was presented in the location opposite that
indicated by the arrow (an invalid-precue trial),
Benefits Versus "Costs"
the average RT was 304 ms, revealing a 39 ms cost
of Anticipating of anticipating the direction incorrectly, relative
The previous sections have described benefits to to the neutral-precue condition.
performers when they correctly anticipate tem- Notice that the cost in this study involved only
porally, spatially, or both. However, the adage the detection of the signal, because the response
that "you don't get something for nothing" holds was always the same. But what happens when
just as well in motor control laboratories as in the task also includes different response alterna-
anything else, and there are necessarily "costs" tives? What are the costs involved in actually
of various kinds that result from anticipating. moving incorrectly? Schmidt and Gordon (1977)
What happens if we anticipate incorrectly, as used a two-choice RT task in which the subject
occurs when the batter anticipates a fastball but had to produce a correct amount of force on a
receives a curve ball instead, or the boxer expects lever in a direction indicated by a signal light. In
a blow from his opponent's left hand but receives one series of trials, the right and left signals were
one from the right hand? LaBerge (1973) and presented in random order, and subjects could
Posner, Nissen, and Ogden (1978) used a method not successfully anticipate the direction of the
84 Motor Control and Learning

upcoming response. But in another series of trials, Next, if the subject was anticipating left but the
the signals were presented in an alternating order right signal came on, and the subject did avoid
(right, left, right, left, ... ), and the subjects would making an error, then the RT was somewhat
develop strong spatial anticipation about the next longer (276 ms) than it would have been if the
response. In this alternating series, however, a person had not been anticipating at all (235 ms).
few signals were embedded that were opposite Thus, inhibiting an already planned (incorrect)
to the direction expected; for example, a series movement does require time (276 - 235 = 41 ms).
might involve the following responses: left, right, This can be thought of as the cost of anticipating
left, right, left, right, right-an alternating series incorrectly and is very similar to the 40 ms cost
in which the subject was expecting the last signal found by Posner and colleagues (1978). Interest-
to be a left, not a right. ingly, though, in the Schmidt and Gordon study
On those trials in which the subject was antici- (1977), there was an 83 ms benefit (i.e., 235 - 152) of
pating one direction but was presented with the anticipating correctly, which is somewhat larger
unexpected signal, there were errors on about than benefits found by Posner and colleagues.
64% of the trials. The nature of this effect is illus- However, more important is the finding that
trated in figure 3.17. These erroneous responses people actually moved in the incorrect direction on
were initiated with a rapid RT (144 ms, on the a majority of the trials (64%) in which they were
average); the subject started moving in the incor- falsely anticipating. Eventually, they did start
rect direction (i.e., left) for 144 ms, and only then moving in the correct direction, but not until 288
reversed direction to begin to move in the correct ms (144 + 144 ms; see figure 3.17) had elapsed.
direction. Falsely anticipating appears to be a This was compounded by the fact that the person
major cause of this kind of error. now had farther to go to reach the correct target,

Right

c:
~ Center
·u;
0
D..

Left

152 ms --1
RT for correct moves to
_ _ _ _ _~ correctly anticipated stimuli
,_,...

276 ms "- I
RT for correct moves to
,_,...,,,__ _ _ _ _ _ _ _ __,,__ incorrectly anticipated signals

~---2_3_5_m_s_ _ _~ RT for correct moves

with no anticipation

FIGURE 3.17 Movement error caused by false anticipation.

Adapted from R.A. Schmidt and G.B. Gordon, 1977, "Errors in motor responding, "rapid" corrections, and false anticipations;' Journal of Motor Behavior 9: 107, adapted by
permission of the publisher (Taylor & Francis, www.informaworld.com).
 Human Information Processing 111 85

so that arriving at the target was delayed even decisions on the basis of that information is often
longer. And, in having to move farther to the critical. Deciding whether or not to perform some
target, the errant performer was less accurate in action (e.g., "Should I tum left at an intersection or
hitting the target, because the error in hitting a wait?"), and whether or not something occurred
target is roughly proportional to the movement in the environment (e.g., "Did I see a warning
distance (chapter 7). When we put together all signal or not?"), are examples of this kind of deci-
these negative aspects of making an error, it is sion making. We discuss these issues further to
clear that the cost due to an incorrect anticipation illustrate a style of decision-making analysis that
can be quite high. has been frequently referred to as signal-detection
Many theoretical questions arise from these theory (Green & Swets, 1966; Swets, 1964).
data on error production, but a major importance Signal-detection theory assumes that we make
of these findings is in their application to various decisions on the basis of what we perceived, and
common activities. The pitch count in baseball that such perceptual processes are subject to vari-
is used by both batters (Gray, 2002, 2009a) and ability and error. As such, decisions are not based
umpires (MacMahon & Starkes, 2008) to make on what is actually happening, but on what we
predictions about the nature of the next pitch. perceive to be happening, which could be different.
A tennis player anticipating that a shot will be Consider the example in which a production-line
hit to her left moves to the left, only to experi- employee's job is to inspect glassware, search-
ence that sinking feeling when the ball is hit to ing for flaws or defects in color or shape. The
her right. The defensive lineman in American employee examines each piece and, based on
football expects to be blocked to his left so leans what she sees, accepts the good glasses and rejects
or moves to his right, only to find that he is now the flawed ones, which are then destroyed. Table
being blocked to his right, making the task a very 3.4 presents the four possible outcomes of making
easy one for his opponent who was going to block a decision about the glass. Here, the two possible
him in that direction anyway. Effective coaching decisions to be made by the employee are com-
techniques employ the notion that anticipating bined with two truths" regarding whether or not
/1

has certain benefits and costs-and that one can the glass was actually flawed. In two situations
determine whether or not to anticipate in a certain the employee makes a correct decision: when
situation by weighing the probable gains against she judges a flawed glass to be flawed (called a
potential losses. In many situations, the benefit of hit), and when she judges an acceptable glass as
correctly anticipating might be very small com- acceptable (termed a correct rejection). 4 However,
pared to the cost of a false anticipation (e.g., when in two other cases, the employee makes an error.
a driver anticipates what another driver may do). In one case, she judges an actually acceptable
In other cases, the reverse is true. Obviously, these glass as flawed (called a false alarm); in the other,
factors will depend on the particular activity, as she fails to detect the flaw in the glass and judges
well as on the particular situation (e.g., the score it as acceptable (called a miss).
of the game, position on the field, etc.). Signal-detection theory assumes that the
"strength" of the perceptual information about
which a decision is to be made varies along a
Signal-Detection Theory continuum, and has a normal distribution (i.e.,
has a dispersion of scores with known statistical
One of the most important components of motor properties) as seen in figure 3.18; by convention,
performance occurs prior to any action at all. the area under these normal curves is taken to be
Taking in sensory information and making yes I no 1.0. There are two curves here, one representing
86 Motor Control and Learning

the situation with a (truly) flawed glass (curve when a criterion (a kind of cutoff point) is set
on the right in figure 3.18) and one indicating along the x-axis, as illustrated in figure 3.18. The
an acceptable glass (curve on the left in figure location of the criterion is usually described by a
3.18); of course, for a given glass, only one of measure called 13 (beta; see Green & Swets, 1966,
these curves can exist, depending on whether for a formal definition). Here we have the curves
or not it is actually flawed, so only one of these for the two glasses, but have added the criterion
two curves can actually exist at the same time. at 33 "strength units." This criterion is set by the
The x-axis represents a hypothetical measure of employee as a result of the task instructions, her
the "strength" of the sensory information point- experience, and various biases she might have
ing to the conclusion that a particular glass is about the task. The criterion is a decision rule:
flawed, with increased strength (moving to the Whenever the employee perceives the strength
right along the axis) indicating more (or "stron- of the signal to be greater than 33 strength units,
ger") information that a flaw exists. A particular she makes the decision that the glass is flawed;
glass has a particular strength for an observer, if the signal is less than or equal to 33 strength
with some dispersion about the mean because units, she decides that the glass is acceptable. This
the perceptual processes are variable or "noisy." provides a "rule" that the employee uses to make
The y-axis refers to the probability that a certain decisions as she examines the various glasses.
strength will be perceived on some occasion when The setting of the criterion allows the employee
the mean strength is at the center of the curve. The to adjust the sizes of the various kinds of errors
y-axis refers to the probability that a particular she is willing to make (in table 3.4). Suppose
perceptual strength will occur. In our example, for that the employee is biased by a perfectionist
a given glass, sometimes the glass appears to be employer so she feels that a flawed glass should
more flawed than it is and sometimes it appears never go undetected. In this case the employee
to be less flawed. And, for an acceptable glass (left would set the criterion very far to the left (say,
curve), sometimes the unflawed glass even appears around 20-far to the left of the flawed-glass
to be more flawed than an actually flawed one, curve in figure 3.18), so that essentially all of the
leading to errors in judgment by our employee. actually flawed glasses would be detected and
destroyed. The problem with this is that many
Setting the Criterion actually acceptable glasses would be destroyed
According to signal-detection theory, the actual as well (i.e., those glasses to the right of "20" on
process by which humans make decisions occurs the acceptable-glass curve in figure 3.18). On the

10 20 30 40 50
Strength of perceptual information
indicating a flaw (arbitrary units)

FIGURE 3.18 Basic components of signal-detection theory.

 Human Information Processing 87

other hand, if maximizing production volume is termed correct rejection, because the acceptable
the employer's goal and some flaws are tolerated, glass was correctly rejected as one being flawed
then the employee would set the criterion farther [i.e., p(correct rejection) = 0.85].
to the right (say, around 40), so that an acceptable
Reducing Errors
glass is almost never judged to be flawed. In
doing so, however, she will accept some flawed In this kind of analysis, the two types of errors
glasses. Thus, this procedure allows the employee in table 3.4 occur because of overlap in the dis-
to "trade off" the quality of the glassware for the tributions seen in figure 3.18. This overlap can
quantity of glassware produced-a classic issue be reduced in essentially two ways, as shown in
in manufacturing. Another classic example in figure 3.19. We can see one way by comparing
signal-detection analysis is the job of the radiolo- the top example in figure 3.19 with the middle
gist, which is to detect the presence or absence of example, in which the classes of objects being
a tumor in an X-ray (Swets, Dawes, & Monahan, evaluated are more physically different. Imagine
2000). There are many other practical examples, a situation in which the manufacturing process
especially in situations in which the individual produces glasses that either are essentially perfect
must make a decision based on sensory informa- or have a massive flaw (e.g., a severe discolor-
tion. ation). Now, the means of the "acceptable" and
"flawed" distributions are far apart, resulting
Misses and False Alarms in very little or no overlap and essentially no
misses or false alarms (middle example). Sec-
Now, with the addition of the criterion, we can ondly, overlap can be smaller if the variability
relate the curves in figure 3.18 to the types of of the two distributions is reduced, as seen in
decisions represented in table 3.4. Consider the the bottom example in figure 3.19, while the
rightmost curve, indicating that the glass being means are kept the same distance apart as in the
judged by the employee is actually flawed. Given top example. Ways to accomplish this might be
the employee's criterion at 33, the probability of to enhance training for the employees, to select
detecting that this glass is flawed is represented people who are very talented perceptually, or to
by the area to the right of the criterion (light gray provide more effective lighting or background
shaded area), which is about 60% of the area fall- conditions to facilitate visual detection. There
ing beneath the flawed-glasses curve. Note that are nearly countless real-world examples. One of
this is the probability of a hit-correctly detecting these concerns the effect of viewing conditions on
that the flawed glass is flawed [i.e., p(hit) = 0.60]. signal-detection errors when soccer officials make
On the other hand, the probability of judging this offside calls; the visual angle at which the play
glass acceptable is the remaining 40% of the area was viewed contributes greatly to the specific
under the flawed-glasses curve that lies to the types of errors made (Oudejans et al., 2000; see
left of the criterion (dark gray shading). This is also Baldo, Ranvaud, & Morya, 2002).
the probability of a miss-judging that a flawed
glass is acceptable [i.e., p(miss) = 0.40]. The Nature of Decision-Making Errors
Next, consider the case in which the employee One of the advantages of signal-detection meth-
is examining an actually acceptable glass, rep- ods has to do with evaluation of the nature of the
resented by the curve on the left in figure 3.18. errors that subjects make, as well as evaluation of
Here, the probability of judging this glass as the effects of various experimental variables (e.g.,
flawed is represented by the area under the left lighting conditions) on detection. Benefits in accu-
curve lying to the right of the criterion (white racy can come in essentially two different ways.
section), which represents about 15% of the area First, consider our example with glassware-
under the acceptable-glasses curve. This decision flaw detection under different lighting conditions.
is a false alarm, because the acceptable glass was From figure 3.20 (p. 89), assume that the employee
judged to be flawed [i.e., p(jalse alarm) = 0.15]. chooses to enhance the lighting in her workplace,
Finally, the probability of judging the good glass and that this enhanced lighting produces a shift in
to be acceptable is given as the area under the left the employee's criterion (i.e., a shift in 13), say to
curve to the left of the criterion (light gray shaded the right from 25 to 35 units. This shift in 13 results
area), which represents the remaining 85% of the in a reduction in false alarms, but at the same
area under the acceptable-glasses curve. This was time produces an increase in misses. These kinds
88 mi Motor Control and Learning

FIGURE 3.19 Effects of reducing variability in signal-detection judgments.

of changes, and the shift in the types of errors The reader would then know that the improved
that result, reflect the subject's voluntary choice lighting conditions increased the fundamental
of criterion (13), perhaps due to lighting choices, detection sensitivity (d'), but also had an effect
or perhaps because of a new directive from the on the subject's choice of criterion (13).
company hierarchy, new ownership, or the like.
Second, the effects of the lighting change
might reduce the variability in her detections, Memory
as in figure 3.19 (bottom example). This kind of
shift reflects a fundamental change in detection So far, our focus has been on what happens to
sensitivity and is independent of a shift in criterion information as it enters the system and is pro-
(13). This kind of shift in sensitivity is reflected in cessed, eventually leading to a response. Informa-
a measure called d' (" d-prime"), which represents tion must be retained (some would use the word
the amount of separation between the means of "stored") for future use, perhaps in a way analo-
the two distributions. 5 Note that when this hap- gous to how information is stored in a computer
pens, the number of both misses and false alarms or library. The persistence of this information that
can be reduced because the subject has become a is stored for future processing is called memory.
more sensitive "detector" of the flaws. Memory is one of the most controversial and
Finally, experimenters examining decision highly debated topics of information process-
making usually report both 13 and d' in their ing. Researchers tend to conceptualize ideas and
experimental write-ups. In our example, we experimental methods according to particular
might have said that the lighting conditions theoretical frameworks, and these vary greatly
produced a shift in 13, but also an increase ind'. among psychologists. We briefly present one of
 Human Information Processing !Jill 89

10 20 30 40 50
Strength of perceptual information
indicating a flaw (arbitrary units)

FIGURE 3.20 Effects of shifting the criterion on signal-detection errors.

these frameworks. However, the interested reader Memory has a direct influence when one makes
should consult the many books on this topic for a deliberate attempt to recollect past experiences
more complete discussions (e.g., Roediger, 2008a). for the purpose of facilitating current informa-
In this viewpoint, memory is considered to be tion processing (e.g., Richardson-Klavehn &
a consequence of information processing. When Bjork, 1988). For example, one of life's awkward
previously processed information influences moments occurs when you meet an acquaintance
current information processing, we assume that on the street and cannot recall the person's name.
memory is the reason. Indeed, when viewed In this case, there is a failure to recall a specific
this way, it is obvious that everything we do is memory as a purposeful and explicit means to
influenced by memory. The current state of our solve a problem (addressing the person by name).
skills and knowledge reflects previous informa- Scientists use terms such as memory search and
tion processing. Thus, there seems to be no doubt retrieval to describe deliberate attempts to use
that memories exist. Rather, the debate among memory in a direct way.
psychologists concerns (1) how memory affects Memory can also have an indirect effect on
the performance of our daily activities and (2) information processing. For example, in the act
what form these memories might take. of typing, the production of the letters on the
computer screen is determined by coordinated
Direct Versus Indirect Influences actions of the two hands. In this case, the memory
for past experience (i.e., learning) in typing is
of Memory
having an indirect impact on the ability to carry
Memory seems to influence our daily activi- out this activity. You do not need to recollect
ties in two rather distinct and separate ways. specifically when you last typed, or even when
90 Ill Motor Control and Learning

you learned to type, in order for the memory of (1960). Sperling presented a matrix of three rows
the skill to influence your performance. One does of four letters each on a tachistoscope, a device
not have to be conscious that memory is being for presenting visual information very briefly
used in order for it to influence performance. and under controlled conditions. The matrix was
Motor skill can be described mainly in terms of presented for 50 ms so that the subjective impres-
an indirect memory influence. Since this book is sion was a bright flash of the letters. One of three
about the learning and control of motor skills, tones was also presented, indicating which row
our interest in memory is in how it affects cur- of four letters the subject was to recall. The tone
rent information processing indirectly. However, could be presented 100 ms before the letter matrix
much of what is known about memory comes was flashed on, simultaneously with the letter
from experiments in which subjects are asked to matrix, or 150, 300, or 1,000 ms after the matrix
recall or recognize information in a direct manner appeared.
(i.e., consciously). The number of letters recalled in Sperling's
In the remainder of this chapter, we describe experiment is plotted as a function of the tem-
a rather traditional distinction that psychologists poral location of the tone in figure 3.21. When
have made among various memory systems, cus- the tone was presented before the letters, the
tomarily labeled short-term sensory store (STSS), recall was about 3.3 letters (out of 4). When the
short-term memory (STM), and long-term_ memory tone was presented 150 ms after the letters, recall
(LTM). In the last chapter of the book (chapter was only about 2.3 letters. When the tone was
14, "Retention and Transfer"), we focus on the presented after a full second, recall was only
indirect influences that memory for motor skill about 1.5 letters.
has on our daily activities. The concept revealed in Sperling' s experiment
Psychologists sometimes refer to these ideas is that all of the letters are delivered by the flash
as "box theories" of memory, because the storage to STSS, where they are stored briefly. However,
systems are often discussed as if they were places the subject does not know which of the rows to
in which information is stored, which is a common attend to until the tone is presented. If the tone is
metaphor for memory (e.g., like a computer presented immediately, the letters are still avail-
hard drive or a university library). However, one
able and the subject can recall them. But if the
should note that this is just one way to concep-
tualize memory-other views discuss memory
simply as the persistence of the products of infor-
mation processing, without taking a stand about
memory's location. This reflects one of the many
controversies that make the study of memory so
interesting (see Roediger, 2008b, for a compelling
historical review of the memory literature).

Short-Term Sensory Store

The most peripheral memory is thought to hold
massive amounts of information, but only for
brief periods of time. When information is pre-
sented to the system, STSS accepts it without
much recoding and then loses it rather quickly
as new information is added. Just as the redness
of a burner on an electric stove fades when the
burner is turned off, the information in the STSS -100 0 150 300 1000
is thought to fade or decay with the passage of Delay of tone (ms)
time. Such a system can be proposed for each of
the stimulus modalities-vision, touch, audition, FIGURE 3.21 Number of items correctly recalled as
a function of the delay of the tone indicating which row
kinesthesis, and so on.
should be recalled.
Some of the earliest and strongest evidence Reprinted from G. Sperling, 1960, "The information available in brief visual
about STSS came from the work of Sperling presentations;' Psychological Monographs 74: 498.
 Human Information Processing Iii 91

tone is delayed, the letters have begun to decay during the retention interval. Thus, all the subject
from STSS and the subject cannot report as many had to do was remember the trigram after count-
of them. This evidence suggests that (a) STSS is ing backward for up to 18 s.
capable of holding all the information presented Peterson and Peterson's results are shown in
to it (because the subject could report any of figure 3.22, where the probability of successfully
the letters in the row if the tone was presented recalling the trigram is graphed as a function of
immediately) and (b) STSS loses information very the length of the retention interval. When the
rapidly with time. recall was nearly immediate, the probability of
On the basis of later experiments, the infor- recall was about 0.90; but when the retention
mation in STSS is thought to have a maximum interval was increased only by a few seconds,
duration of about 1 s, with a more practical limit there was a marked decrease in the recall. This
of about 250 ms. Also, it involves rather literal persisted until, at 18 s, almost no trigrams could
storage of information, in that the stimulus is be recalled. The evidence suggests the existence
recorded in the same way it came into the system of a memory system that loses information rap-
in terms of both spatial location and form; this is idly (in about 30 to 60 s) unless the information
analogous perhaps to how film records images is rehearsed in some way.
that enter the lens of the camera (see table 3.5). Short-Term Memory Versus Short-Term
Contents of STSS are then available for further Sensory Store
processing.
A major difference between STSS and STM (table
Short-Term Memory 3.5) relates to capacity. Previously we mentioned
that the capacity of STSS is very large-practically
Short-term memory is thought to be a storage limitless. However, on the basis of experiments
system for information delivered either from in which subjects have been asked to remember
STSS or from LTM. It has a limited capacity and as many members of a list of items as they can,
a relatively short duration. Originally, STM was evidence suggests that STM has a capacity (often
thought of as a kind of "work space" for pro- called the span of memory) of only about seven
cessing, and one where control processes such as (plus or minus two) items (Miller, 1956). This
decision making, rehearsal, coding, and so on are conclusion, however, depends on the definition of
performed (Atkinson & Shiffrin, 1971). Peterson an "item." Sometimes subjects organize separate
and Peterson (1959) and Brown (1958) provided items into larger groups, so that each group may
evidence for this kind of system that was to have contain five "items" of its own; this process has
a strong influence on research in memory for the been termed chunking (Miller, 1956). The idea is
next two decades. In the paradigm developed by that if there are 55 letters to remember (e.g., the
Peterson and Peterson, subjects were provided number of letters in the first sentence in this para-
with a single trigram (three unrelated letters, e.g., graph), it would be difficult to remember them as
XBF) for a period of study time; then the letters completely separate items without chunking. By
were removed, and the subjects had to count chunking the letters into larger, more meaning-
backward by threes from a three-digit number ful groups (words or sentences), one can recall
until recall of the trigram was requested, from 0 the items more easily. In this sense, the capacity
to 18 slater. The backward counting was intended of STM is thought to be seven chunks. Ericsson,
to prevent the subject from rehearsing the trigram Chase, and Faloon (1980) showed that after 175

7 ± 2 items
92 Motor Control and Learning

1.0

0.8

iU
"~ 0.6
0
>-
~
:cIll
.c 0.4
0
....
Cl..

0.2

0 3 6 9 12 15 18
Retention interval (s)

FIGURE 3.22 Probability of correct recall of a single trigram as a function of the retention interval.
Reprinted from L.R. Peterson and M.J. Peterson, 1959, "Short-term retention of individual verbal items,'' Journal of Experimental Psychology 58: 198.

days of practice with a technique for chunking limited-capacity conscious processing (controlled
effectively, a subject was able to increase the processing, discussed further in chapter 4) can be
capacity of STM from 7 to 79 items by creating performed. Working memory is thought to con-
larger and larger chunks. Therefore, there were tain items in a high level of activation for rapid
probably only about seven chunks or groups of access, but these are lost relatively quickly unless
items held in storage at any one time. they are attended to, rehearsed, or allocated some
Another distinction between STM and STSS of the limited capacity (Anderson, 1990).
(table 3.5) is in the nature of the coding processes. For motor behavior, the most important idea
In STM, coding is considered to be abstract. For is that information processing entailing the
example, stimuli are given names, and the sepa- choice of actions is involved in working memory,
rate stimuli are often combined in various ways which suggests that working memory is closely
to produce chunks that can reduce the number of related to the response-selection stage mentioned
separate items in STM. Although many theories earlier in this chapter. Also, during the response-
of STM do not directly say so, the implication is programming stage, information in the form of
that STM is related to consciousness; those things stored motor programs-described more fully
in STM are essentially things of which we are in chapter 6-or other well-learned information
consciously aware. about a task can be retrieved from LTM. In work-
ing memory, presumably, environmental infor-
Working Memory
mation from STSS (such as the locations of seen
Atkinson and Shiffrin's (1971) earlier notion was objects, speed of a flying ball) can be integrated
that STM comprised a kind of limited capacity with information from LTM, creating what some
"work space" where various operations were term an "action plan" that takes into account
performed. More recently, this view of STM one's stored capabilities and the specific environ-
has been expanded and incorporated into the mental demands. The prepared movements are
notion of working memory (e.g., Baddeley, 2003). then triggered from working memory to generate
Although there are differing viewpoints, the muscle contractions and actions.
essential idea is that working memory is a part
of STM where (a) information from STSS can be Motor Short-Term Memory
stored for processing, (b) information from LTM In a motor analog of verbal STM, Adams and
can be retrieved for processing and integrated Dijkstra (1966; Posner & Konick, 1966) asked
with information from STSS, and (c) effortful and blindfolded subjects to move a slide along a
 Human Information Processing oo 93

trackway until it struck a fixed stop that defined An example is learning a new phone number. The
a criterion target position. Then the subject first time you hear it, you are likely to forget it
moved back to the starting position to wait for quickly if you do not rehearse it. Practice results
the remainder of the retention interval (from 10 to in the transfer of the number to more permanent
120 s ), after which the subject attempted to move storage. In some cases, this storage is indeed
the slide to the criterion position with the stop permanent. Can you remember what your home
removed. The absolute error in recalling the posi- phone number was when you were a child? We
tion increased sharply as the retention interval can.
increased from 10 to 60 s and changed very little Long-term memory, of course, provides the
thereafter (figure 3.23). These findings closely capability for making movements that have
paralleled the early findings of Brown (1958) and been practiced before. Some of the variables that
Peterson and Peterson (1959) with verbal materi- appear to determine retention of well-learned acts
als, in that nearly all the forgetting of the position are discussed in chapter 14. For now we can say
occurred within the first 60 s, which is interpreted that practice leads to the development of "better"
as the approximate upper limit for retention in or "stronger" (or both) LTM for movement and
short-term verbal memory. Chapter 14 presents that these memories are often present after many
more about short-term motor memory studies. years, even without intervening use of that stored
Long ago, James (1890) proposed the distinction information. Riding a bicycle is the most often
between "primary memory" (what is today called cited example, as people appear to be able to
STM) and "secondary memory" (today's LTM), ride acceptably well after 40 years or more with
described next. no intervening practice.
Another major distinction between LTM and
Long-Term Memory STM relates to the amount of information that
When items are practiced (or rehearsed), which of can be held (table 3.5). Most argue that STM has
course requires information-processing activities, a functional capacity of about seven chunks,
they are in some way transferred from short-term whereas LTM must have a very large capacity
storage to long-term storage, where they can be indeed. In the motor realm, the analog of well-
held more permanently and protected from loss. learned facts and principles is well-learned motor
skills. Thus, the functional capacity of motor LTM
must also be very large if it is capable of retaining
all of the movements that humans can typically
30 perform on demand.

26 Summary
eE
':' 22 We can understand a great deal about the way
.
~
QI people move by considering the human as an
0~
information-processing system that takes in
18
information from the environment, processes it,
~
m and then outputs information to the environment
c:
81 14 in the form of movements. Using the concepts
:E of subtractive logic first advanced by Donders
10
(1868/1969), stages of processing can be defined
and studied through the use of RT methods. The
first stage, called stimulus identification, concerns
20 40 60 80 100 120 the reception of a stimulus, preliminary (precon-
Retention interval (s)
scious) analyses of features, and extraction of
patterns from the stimulus array. Variables like
FIGURE 3.23 Absolute error in positioning recall as a stimulus clarity and stimulus intensity affect the
function of the retention interval length.
Reprinted, by permission, from J.A. Adams and S. Dijkstra, 1966, "Short-term
duration of processing in this stage. A second stage,
memory for motor responses," Journal of Experimental Psychology 71: 317. called response selection, concerns the translation
94 Ill Motor Control and Learning

or decision mechanisms that lead to the choice of a. Discuss an everyday-life example of

response. The duration of this stage is sensitive to uncertainty that illustrates the concept
variables such as the number of S-R alternatives of information theory.
and S-R compatibility (the extent to which the b. Choose any sport to illustrate the con-
stimulus and response are "naturally" linked). tributions of each of the three stages
The final stage, called response programming, is of information processing.
associated with changing the abstract idea of
c. Using the same sport as for question
a response into muscular action. The duration
1b, describe examples of the various
of this stage is related to variables affecting
kinds of anticipation.
the response, such as response complexity and
response duration. Some of the processing in 2. Find a research article that uses signal-
some of these stages can apparently be bypassed detection theory to describe processes
by anticipation. Providing both spatial and involved in decision making.
temporal predictability allows so-called early
responding, whereas providing other information Web Resources
leads to increased readiness and faster response.
Another aspect of information processing that
A fun choice-RT game with variable foreperiod
considers factors related to making decisions has
delays:
been termed signal-detection theory and involves
two separate types of information-processing www.bbc.co.uk/science/humanbody/sleep/
activities: setting the criterion for making a deci- sheep/
sion, and the sensitivity of the system that is used
to make the decision. Factors such as the relative A number of experiments related to concepts in
importance of the type of error that would result chapter 3 can be run from this Web site:
from a wrong decision are major influences on www.artsci.wustl.edu/-rabrams/psychlab/
setting the criterion, whereas the skill level of the index.htm
participant and the nature of the environmental
information are major factors in the sensitivity of
the information upon which the decision is based. Notes
Parallel to the ideas about stages of processing
are concepts about information storage systems, 1 It is fitting that today, F.C. Dondersstraat (which

or memory. Motor skills use memory in an indi- would be translated as "F.C. Donders Street" in English)
rect way, although much of what we know about runs adjacent to Donders' original lab in Utrecht, Holland.
2 Note that our description of Henry and Rogers' (1960)
memory involves directed remembering. In a
common framework, memory can be thought of most complicated task (C) is slightly different than in
earlier versions of this book, due to recent revelations
as consisting of three compartments: a short-term
about a small error in the describing the apparatus in the
sensory store (STSS), capable of storing a large original article (Fischman et al., 2008).
amount of literally coded information for perhaps 3 Titls finding is closely related to research on voluntary
a second; a short-term memory (STM, sometimes motor timing, which is discussed in considerable detail in
discussed as working memory), capable of storing chapter 7. Essentially, the research concludes that produc-
only about seven (plus or minus two) abstractly ing a string of regular time intervals (e.g., as in drumming)
coded items for perhaps 30 s, as well as providing is more variable as the duration of the intervals lengthens
a kind of "work space" for conscious information (e.g., the variability in producing 1 s intervals is greater
processing; and a long-term memory (LTM) capable than the variability in producing 500 ms intervals). The
foreperiod duration effect is similar-anticipating the end
of storing very large amounts of abstractly coded
of a regular foreperiod is more variable as the foreperiod
information for long periods of time. duration lengthens.
4Perhaps the use of the labels "hit" and "correct rejec-

Student Assignments tion" in this example is confusing. Consider that the

employee is searching for a flawed glass and, when she
finds one, this is a hit. Accepting a "good" glass is then a
1. Answer the following questions and bring correct rejection (i.e., a "good" glass is rejected as being
the information to class for discussion: a flawed glass).
 Human Information Processing 111 95

5 Students often find this confusing because the sepa- because the variation of the distributions is smaller. The
ration (in centimeters) between the means in the top and second panel has a larger d' than the top panel because
bottom panels in figure 3.19 is the same. But these are actu- the separation is greater with similar standard deviations;
ally measured in standard deviation (Z) units based on the it is larger in the lower panel than the top panel because
standard deviation of the distributions, and these units of the reduction of standard deviation with a constant
are smaller in the lower panel than in the upper panel mean difference.
 CHAPTER 4

ttention has always been a topic of major to coordinate our limbs in time and space as
interest to psychologists and motor behav- skill develops. Still other types of attention are
ior researchers. Early research and theorizing required to offset the mental drowsiness that
began in the 19th century (Cattell, 1886; Welch, accompanies long periods of driving without
1898), and interest in the topic remains high today. rest, or to talk on a cell phone. So, as we read
Much of the early work involved introspection; in the quote from William James' description
for example, William James (1890), one of the of attention, a number of features of the phe-
most renowned experimental psychologists, nomenon are considered important and reflect
wrote: various ways to think about the different types
Everyone knows what attention is. It is the taking of attention that may exist.
possession by the mind, in clear and vivid form,
of one out of what seem several simultaneously
possible objects or trains of thought. Focalization, Types of Attention
concentration, of consciousness are of its essence.
It implies withdrawal from some things in order to There are many different ways to view the
deal effectively with others. (pp. 403-404) concept of attention. One of these is the notion
But does everyone agree on what attention that attention is limited: We can attend to only
is? Many theorists, such as Norman (1976) and one thing at a time, or think only one thought
Moray (1970), suggested that different defini- at a time. In terms of motor behavior, we seem
tions of attention exist, and people use the term strongly limited in the number of things we can
in a variety of ways. Consider the task of driving do at a given time, as if the limits to some maxi-
a car. Drivers must be aware of a preplanned mum" capacity" would be exceeded if too much
route, as well as where they currently are in rela- activity were attempted. Another important
tion to the route, in order to make appropriate feature is that attention is selective: We can con-
turns at the right times. The driver must also centrate on one thing or on something else, and
be aware of other traffic (e.g., cars, pedestrians, can freely shift attention back and forth among
and bicyclists) and be capable of responding numerous things. Here we discuss a few of the
to sudden changes. The control of movement types of attention that are particularly relevant
requires another type of attention, although we in the control of motor skills. Note, however,
spend less time and energy thinking about how that the topic of attention entails a very broad

97
98 111 Motor Control and Learning

research area-much more than can be covered explained as situations in which an unconscious
in this chapter. The interested reader is encour- or automatic action has not been successfully
aged to seek out some of the many excellent inhibited or counteracted by conscious, controlled
recent reviews for more in-depth discussion of processing (Hay & Jacoby, 1996; Reason, 1990;
attention (e.g., Baddeley & Weiskrantz, 1993; Reason & Mycielska, 1982).
Folk & Gibson, 2001; Neumann & Sanders,
1996; Pashler, 1999; Shapiro, 2001; Wickens & Attention as Effort or Arousal
McCarley, 2008). Another way to operationalize the notion of
attention is based on the idea that when people
Attention and Consciousness perform attention-demanding tasks such as
Early in the history of research on human per- balancing a checkbook or diving in competitive
formance, as implied by James' (1890) state- swimming, they are expending mental effort that
ment, attention was linked to the notion of is revealed in various physiological measures. For
consciousness, which is defined loosely as "what example, Kahneman (1973) and Beatty (Beatty &
we are aware of at any given time." The term Wagoner, 1978) have used pupil diameter, mea-
"conscious," and in particular the concept of sured by special techniques that do not interfere
unconscious behavior, fell out of favor during the with eye movement, as an indirect measure of
growth of behaviorism after the tum of the 20th attention. When subjects are asked to perform
century. The measurement of consciousness was various memory tasks, pupil diameter increases
troublesome because at the time the only way to when they are under pressure to provide an
understand what was "in" subjects' conscious- answer; the increase is larger for more "difficult"
ness was to ask them to introspect, or "search tasks. Similarly, it is useful to consider attention
their own minds," and this was far too subjective as reflected by various physiological measures
for the behaviorists' approach to accumulating of arousal, a dimension indicating the extent to
data and theorizing. which the subject is activated or excited. Kahne-
Toward the end of the 20th century, however, man (1973) used physiological measures of skin
the concept of consciousness saw a resurgence in resistance (resistance to a weak current passed
popularity among cognitive neuroscientists, and between two electrodes on the skin decreases
that interest continues to grow (Cohen & Schooler, with increased arousal) and heart rate as indirect
1996; Posner & Petersen, 1990). Examining brain measures of the attention demand of various
function using methods such as fMRI (functional tasks.
magnetic resonance imaging) and TMS (transcra-
nial magnetic stimulation) (see chapter 2) has Attention as a Capacity
allowed scientists to measure patterns of brain or Resource
activity, revealing much more objective types
of information than had previously been avail- Another view of attention, and one that is an im-
able through methods of introspection (Baars, portant component of the information-processing
1997; Chalmers, 1995; Crick & Koch, 2003; Hag- concepts discussed in chapter 3, suggests that
gard, Clark, & Kalogeras, 2002). Consciousness humans possess a limitation in the capacity (or
has also been linked to the concept of controlled resources) available to handle information from
versus automatic processing (discussed later in the environment. The idea for the concept of a
this chapter). Performance on various memory limited capacity of attention is illustrated in the
tests (Roediger & McDermott, 1993), and the use ability to perform two tasks simultaneously. If
of process-dissociation measures (Jacoby, Ste- one activity (A) requires attention, then some (or
Marie, & Toth, 1993), suggests an independence perhaps all) of the "pool" of limited capacity of
between conscious and unconscious influences on attention must be allocated to its performance.
behavior. For example, automatic (unconscious) Because the amount of this capacity is thought
processing appears to be preserved well in older to be limited, some other activity (B) that also
adults, whereas controlled (conscious) processing requires a certain amount of this capacity
is quite susceptible to decline with aging (Craik & will compete with A for these limited atten-
Jacoby, 1996; Hasher & Zacks, 1988). Performance tional resources. When the combined need for
errors such as action slips (Norman, 1981) are often resources exceeds the total amount of attentional
 Attention and Performance 111 99

capacity available, then B will interfere with the not necessarily due to limitations in attentional
performance of A, and vice versa. Interference capacity. On the other hand, when one can rea-
could be demonstrated in many ways: (a) B sonably rule out the possibility that structural
could suffer in performance speed or quality interference between two tasks is occurring,
while A was relatively unaffected; (b) B could then a capacity interference-or a decrement in
be unaffected while A suffered; (c) both A and performance due to some limitation in central
B could suffer; or (d) B could be prevented from capacity (i.e., attention)-is inferred.
occurring altogether while A was in progress. The concerns about "distracted driving" pro-
These patterns of "interference," or competition vide a good example of the difference between
for attentional resources, could presumably tell structural and capacity interference. Many laws
us something about the nature of the limitations dealing with distracted driving ban the use of
in capacity. handheld communication devices, the argument
being that the perceptual and motor requirements
Interference as a Measure of Attention
for communicating with the device create struc-
If two tasks can be performed as well simul- tural interference with the hand and eye move-
taneously as each can be performed individu- ments required for driving. The more important
ally, then at least one of them does not require problem, however, is that the task of communica-
attention, or a portion of the limited capacity. tion is not simply a structural interference issue-
We would say that at least one of the tasks is hands-free communication devices also interfere
"automatic." On the other hand, if one task is with the task of driving because of the attention
performed less well when it is combined with (capacity interference) required to use them.
some secondary task, then both tasks are thought
to require some of the limited capacity. In this Selective Attention
instance, both tasks are attention demanding. Over
the past few decades, this interference criterion Very closely related to the limited-capacity view
became the critical test of whether or not a is the concept that we can direct (or allocate)
certain task "required attention." Although this attention to different inputs or tasks. Selective
test for attention achieved popularity during attention can be either intentional or incidental,
the cognitive revolution, it is not really a new depending on how a specific allocation has
research method (Welch, 1898). been achieved (Eimer, Nattkemper, Schrager, &
Prinz, 1996). Intentional selection occurs when
Structural Interference we purposefully choose to attend to one source
and Capacity Interference of information (e.g., listening to the radio) while
The simultaneous performance of two tasks avoiding or inhibiting attention to other sources
can result in interference between them for a (e.g., the television or someone talking to us). An
variety of reasons, only some of which would involuntary capture of attention usually occurs as
be interpretable as interference due to limita- a response to an external stimulus-for example,
tions in some central capacity (attention). To when you suddenly pay attention to a loud or
confront this problem, researchers have defined pertinent sound (e.g., the sound of two cars col-
two kinds of interference: structural and capac- liding). Theorists sometimes refer to intentional
ity. Structural interference results when physical selection as "top-down" processing, and invol-
(or neurological) structures are the source of the untary selection as "bottom-up" processing, to
decrement. For example, the hand can be in only indicate that the orienting of attention is concep-
one place at a time, and interference between tually versus perceptually driven.
handwriting and pressing the buttons to make Selective attention is readily observed in the
a phone call with the same hand would be due, patterns of interference already mentioned in
at least in part, to this kind of limitation and dual-task situations. Directing attention toward
not necessarily to a limitation in some central activity A may reveal deficits in the performance
capacity. Also, the eyes can focus at only one of task B, although no performance deficit is
signal source at a time, and thus detecting two observed for A. However, by shifting the attention
simultaneous visual signals presented in widely to activity B, you may observe that activity A is
different locations could suffer in processing now the one that suffers and that performance of
speed because of a structural limitation, again B is very proficient.
100 m Motor Control and Learning

Theories of Attention at a time. A major question for these attention

theorists was where (e.g., early or late) such a
bottleneck, or "filter," was located. These filter
If attention is defined as, or measured by, the
theories differed in terms of the kinds of infor-
degree of interference between two tasks, then mation processing that required attention and,
which kinds of tasks do and do not interfere with therefore, the location of the bottleneck during
each other, and under what conditions might the stages of information processing.
these patterns of interference be expected to
occur? Most of the everyday tasks we perform can Early-Filter Theories
be thought of as collections of processes involving Welford's (1952) theory assumed that all processes
stimulus input and encoding, response selection require attention; in other words, the human
and choice, and motor programming and move- could be regarded as a single information chan-
ment control. The fact that two complex tasks nel that could be occupied by one and only one
interfere with each other (or do not) might not be stimulus-response operation (or "channel") at a
very meaningful by itself, because it would not time. As illustrated in figure 4.1, line 1, Welford's
be clear what the cause of the interference was theory located the bottleneck at the earliest stage
or where in the information-processing activities of information processing. Of course, if only one
the interference occurred (Jonides, Naveh-Benja- operation can be done at a time, then any task
min, & Palmer, 1985). Did the two tasks require attempted at the same time as another task will
response-selection activities at the same time, or interfere (or even be blocked altogether). Thus,
did they require movement programming at the this single-channel idea viewed a secondary task
same time? As a result, simpler laboratory tasks as a rather severe source of interference. For this
are used often in this research so that the various reason, processing in the single channel is defined
processing stages can be more specifically identi- as attention demanding on the basis of the inter-
fied and studied. The following theories of atten- ference criterion described earlier.
tion attempt to explain the patterns of interference Weaker versions of the single-channel theory
found in performing these types of tasks, using denied that all the stages of processing require
various hypothetical structures and processes. attention, especially the earliest stages, and there-
fore the location of the bottleneck was moved to
Single-Channel, Filter Theories a later stage in the sequence of processing. Thus,
Some of the first few theories of attention (e.g., these other theories (Broadbent, 1958; Deutsch &
Broadbent, 1958; Deutsch & Deutsch, 1963; Keele, Deutsch, 1963; Keele, 1973; B. Kerr, 1973; Norman,
1973; Norman, 1969; Treisman, 1969; Welford, 1969; Treisman, 1969) presumed that early stages
1952; see also B. Kerr, 1973), while different in of processing occurred without attention but
detail, had some important features in common. that attention was required at the later stage(s)
They all assumed that attention was a fixed of processing. Processing without attention
capacity for processing information and that per- implies parallel processing, such that a number
formance would deteriorate if this capacity was of separate signals can be processed simultane-
approached or exceeded by the task requirements. ously without interfering with one another. For
These were single-channel theories of undifferenti- example, processes that translate sound waves
ated capacity, in that attention was thought of as a into neurological impulses in the ear, and those
single resource that could be directed at any one that change mechanical stimuli into neurological
of a number of processing operations. Theorists activity in the movement receptors in the limbs,
often described single-channel theories as bottle- can occur together, presumably without interfer-
neck theories of attention, because the "neck" ence. In other words, these theories assumed that
limits the information that can be later processed peripheral (mainly sensory) information process-
and its location determines when interference ing occurs simultaneously and without interfer-
is likely to occur. Here, as shown in figure 4.1, ence, but the theories differed with respect to the
information flovy is left to right, and many opera- exact information-processing stages at which the
tions can occur at the same time in the large part interference occurs.
of the bottle; but, at some point, a "bottleneck" Broadbent (1958) and Deutsch and Deutsch
is encountered, which allows only one operation (1963) theorized that a kind of filter is located
 Attention and Performance im 101

Stimulus identification

FIGURE 4.1 Utilization of attention in various stages of processing, according to various theories. (Line 1 represents
the original single-channel theory [Welford, 1952]; line 2 represents Broadbent's [1958] filter theory; line 3 represents the
Deutsch and Deutsch [1963] and Norman [1969] theories; and line 4 represents Keele's [1973] theory.)

somewhere along the series of stages of infor- physical stimuli into neurological signals. The
mation processing (see figure 4.1, lines 2 and perceptual analysis stage involves the process
3). According to these theories, many stimuli that abstracts some preliminary, simple meaning
can be processed in parallel and do not require from the stimuli (e.g., perception of right angles,
attention prior to reaching the filter. When the or verticality). (Notice that the stages labeled sen-
filter is reached, however, only one stimulus at sory storage and perceptual analysis in the language
a time is processed through it (the others being of these theorists can be readily combined to yield
"filtered out"), so that the information processing the stimulus-identification stage discussed in
from then on is sequential, requiring attention chapter 3.) Broadbent viewed perceptual analy-
in the single channel. The decision regarding sis and later stages as requiring attention, while
which stimuli are filtered out and which one is Deutsch and Deutsch, Treisman, and Norman
processed further into the single channel presum- saw perceptual analysis as automatic (i.e., not
ably depends on which signal arrives first to the requiring attention), with later stages requiring
filter, the nature of the activity in which stimuli attention. Thus, these theories are similar, but
are expected, and which stimuli are relevant to they differ with respect to where the proposed
the task in question. filter is located in the chain of processes.
Lines 2 and 3 in figure 4.1 show the locations
of the proposed filter for these two theories. The Keele's Late-Filter Theory
sensory storage stage is considered the most Keele's (1973) theory of attention places the bot-
"peripheral," involving the translation of the tleneck even later in the sequence of stages than
102 Motor Control and Learning

the Deutsch and Deutsch theory. According to the relative importance of the tasks, their relative
Keele's concept, information is processed in par- difficulty, and other factors. Trade-offs between
allel and thus is attention free through the stimu- proficiency in two simultaneous tasks have been
lus-identification and response-selection stages. discussed by Norman and Bobrow (1975), Posner
At this point, memory contact is made, in which and Snyder (1975), and Navan and Gopher (1979).
certain associates of the stimuli are activated, such
as items in related categories, items closely asso- Multiple-Resource Theories
ciated with the stimulus, or even certain aspects
of early preparation for a movement that is to be Some researchers have argued that attention
triggered by the stimulus. In this view, because should not be conceptualized as a single resource
all such stimuli "contact memory" at about the but rather as multiple pools of resources, each with
same time, selective attention must determine its own capacity and each designed to handle
which of these memory contacts are to receive certain kinds of information processing. In this
further processing. These subsequent operations, view, for example, separate resources would be
such as memory searches, rehearsal, recoding, responsible for selecting the finger to make a
or readying a movement for production, are the movement and for selecting the movement of
ones that are attention demanding, according the jaw to say a word. Hence, these two opera-
to Keele. If two such processes are required at tions could coincide without interference (e.g.,
the same time, decrements in performance will McLeod, 1977; Navan & Gopher, 1979; Wickens,
occur. Thus, Keele's view is represented as line 4 1976, 1980; Wickens & Hollands, 2000).
in figure 4.1, indicating that processing can be in Similarly, Shaffer (1971) and Allport, Antonis,
parallel and without interference even through and Reynolds (1972) argued that attention can be
response selection, with subsequent operations devoted to separate stages of processing at the
requiring attention. same time. Such a position is inconsistent with
fixed-capacity theories, in which the process-
Flexible Allocation of Capacity ing was thought to be confined to a single stage
although it might be possible to perform one
In contrast to filter theories, an argument pre- or more separate operations in parallel. These
sented by Kahneman (1973) was that the capacity views help to explain skill in complex tasks such
of attention could change as the task requirements as typing, simultaneous translations, and sight-
change. For example, as the "difficulty" of two reading of music, for which attention is thought
simultaneous tasks increases, more capacity is to be devoted to input (sight-reading) and output
used in processing the information. Eventually, (muscular activation and control) stages at the
when task requirements for processing two same time.
streams of information begin to exceed maximum
capacity, decrements occur in one or more of the Action-Selection Views of Attention
simultaneously presented tasks; that is, interfer-
ence occurs. Kahneman' s theory also differed In all of the viewpoints about attention that have
from the earlier views by suggesting that parallel been described, the basic assumption is that
processing could occur in all of the processing information-processing activities require some
stages, but with some demand on attention at the kind of capacity (or "fuel") in order for behavior
same time. Kahneman's view that the amount of (skill) to occur. Decrements in performance result
allocated attention is not fixed creates a number of when various activities compete for this capacity.
difficulties for certain secondary-task techniques Some researchers however, question the validity
for measuring spare capacity (see chapter 2), of this fundamental assumption. Scientists such
which assume that capacity is fixed. as Neumann (1987, 1996; see also Allport, 1987,
Other theories of attention have focused on 1993) have criticized the various resource theo-
issues of flexibility in information processing. ries and presented a view that is considerably
For example, rather than assuming that processes different. Neumann argued that when an animal
requiring attention can deal with only one stimu- or human has a certain momentary intention to
lus at a time, other theories suggest that these obtain some goal (e.g., to run, to drink), many
resources can be shared by parallel processing. stimuli received at this time are processed in par-
How they are shared is presumably a function of allel in the early stages-the final product of this
 Attention and Performance m1 103

processing being the selection of a certain action. information is processed regardless of our inten-
Then, as a result of this selection, certain other pro- tions. Three specific research areas have arisen in
cesses are prevented from occurring, or can occur the literature to demonstrate quite clearly that
only with great difficulty. Thus in Neumann's selective attention can be both intentional and
view, interference between two simultaneous incidental. The research areas discussed in the next
tasks occurs not because attention (as a resource) sections include the Stroop effect, the cocktail
/1

is needed in order to perform various processes; party phenomenon," and inattention blindness.
rather, it occurs because an action has already been
selected, and these other processes are completely The Stroop Effect
or partially blocked. Thus, selection is the most A very interesting and powerful effect that helps
basic and fundamental process of attention, not us understand intentional and incidental infor-
resources or capacity (see earlier discussion of mation processing is the so-called Stroop effect,
Keele's late-filter theory). named after the psychologist who first identified
This theory has an interesting ecological the phenomenon (Stroop, 1935).1 Many hundreds
aspect, in that if a particular action is important of research articles on the effect have been pub-
(e.g., escape from a predator) and is selected, lished since this classic article appeared, and the
then it would seem critical that other possible Stroop effect is now commonplace in numerous
actions be prevented, at least for a while, until applications (MacLeod, 1991). The typical experi-
the original action has run its course. The selected ment involves at least two important conditions.
action requires certain processes or structures for In both, the subject is asked to watch a display
its completion, and preventing some other action and perform a simple task: to name the ink color
from using them (and thus interfering with them) in which a word is printed. In one condition,
would ensure that the selected action would, in subjects are presented with a neutral word and
fact, have a good chance of being completed and asked to name the color in which it is printed (e.g.,
of having its goal fulfilled. As we will see in the for the word "house" printed in blue ink, the sub-
next sections, this theory is consistent with the ject's response would be "blue"). In the second
general finding that very little interference occurs condition, the word name is itself a color (e.g., for
in processes related to stimulus processing, with the word "blue" printed in blue ink, the subject's
most of the interference between tasks occurring response would be "blue"). Typically, the time to
. in stages related to the planning or production of respond by naming the color is very fast when the
movements. word and target color are the same, or congruent.
However, when the word name and the color in
Competition for Attention which it is printed are not the same, or incongru-
ent, the subject's response is typically delayed in
time and quite prone to error (e.g., for the word
It is beyond the scope of this text to do a thorough
"green" printed in blue ink, the subject's response
analysis of the extensive literature documenting should be "blue," but it is often incorrectly given
the patterns of interference among tasks. For our as green" or is delayed in time so that the error
/1

purposes, however, some generalizations about

can be suppressed). This finding-namely that
the nature of these patterns, as well as about the
the color meaning of the word interferes with
situations in which the most and least interference
the naming of the ink color in which the word
seems to be produced, will enable us to have a
appears (if the two are incongruent)-represents
reasonable insight into the nature of attention,
the basic Stroop effect (many variations exist; see
at least as it relates to the selection and control
MacLeod, 1991, and MacLeod & MacDonald,
of movement.
2000, for examples).
Processing Unexpected Why should an irrelevant dimension of the
display (the name of the word) interfere with
Information the intentional act of naming the color of the ink?
'I
In this section, we examine the competition that Why cannot the subject simply focus on the color
exists for capturing attention-sometimes we of the ink and ignore the word that the letters
are able to block or bypass things to which we spell? One possibility is that since the processing
do not want to attend, only to find that certain speed for identifying the name of a color is slower
104 111 Motor Control and Learning

than for naming a word (e.g., Fraise, 1969), the discussion in one of these unattended conversa-
name of the word is more quickly readied as a tions cannot be ignored, such as when your name
response than is the name of the color. However, is spoken. In this case, some feature of the ignored
a clever experiment by Dunbar and MacLeod message gets through" to conscious process-
/1

(1984) produced evidence that argued against ing-it captures our attention.
this interpretation. In their studies, Dunbar and The findings from the dichotic-listening para-
MacLeod presented the word names as you might digm led to the suggestion that all of the auditory
see them in a mirror (reversed). Under these stimuli are processed through stimulus identifica-
types of conditions, identifying the name of the tion in parallel and without attention, and that
word was much slower than naming the color. some mechanism operates to prevent attention
Nevertheless, the Stroop effect was still present from being drawn to unwanted sources of sound.
when these mirror-oriented word names were When the sound is particularly relevant or perti-
presented in incongruent colors. Thus, the simple nent to us (e.g., our name or a danger alert, such as
"speed of processing" explanation is insufficient the sound of a car's horn), the stimulus is allowed
to explain the Stroop effect. The resistance of to "pass through" for additional processing and
the Stroop effect to conform to these and other attention. Perhaps stimuli from the environment
theoretical predictions likely explains why psy- had entered the system simultaneously and had
chologists have maintained a fascination with the been processed to some superficial level of analy-
nature of this attentional interference. A popular sis, with only those relevant (or pertinent) to the
view is that the subject's inability to ignore the individual being processed further. Of course, this
irrelevant message is evidence that the irrelevant further processing will usually require attention,
signal (color name) and relevant signal (ink color) implying that two such activities cannot be done
are processed in parallel, perhaps without any together without interference. Like the evidence
interference in the early stages of processing. from the Stroop effect, the evidence from these
observations argues against the early-selection
The "Cocktail Party Problem" models of attention presented in figure 4.1.
Another important illustration of the intentional
and incidental effects of sensory-information pro- Inattention Blindness
cessing is provided by the dichotic-listening para- and Change Blindness
digm. In a typical dichotic-listening experiment, The cocktail party phenomenon" is a nice illus-
/1

the individual is presented (via headphones) with tration of how intentional and incidental human
a different message in each ear. The subject's task information processing coexist in daily activities.
is to ignore one of the messages and to concen- We can selectively filter out unwanted informa-
trate on (and later report about) the other. Nor- tion yet be receptive to pertinent information
mally, subjects are very skilled at concentrating when it arises. Another example of the competi-
on the intended message (termed "shadowing" tion for sensory-information processing illustrates
the message) and ignoring the message being a potentially dangerous consequence of selective
presented to the other ear. However, when a attention-when intentional processing inhibits or
message cannot be ignored, the implication is prevents the processing of a critical sensory event.
that it is being processed through the stimulus- This failure to process incidental information is
identification stage whether the individual tries to a frequent complaint in vehicle-bicyclist traffic
ignore it or not, perhaps without attention being accidents: The driver is looking for an available
required for that processing. parking space or a specific street sign and fails
The dichotic-listening paradigm is a formal to see the cyclist who is directly in front of the
way to study the "cocktail party problem" car. These are called "looked-but-failed-to-see"
described by Cherry (1953). At large, noisy par- accidents (Hills, 1980) because the drivers often
ties, attending to one conversation while ignor- claim that they had been looking in the general
ing the many other conversations is often quite direction of the accident victim but did not see
difficult. But, with effort, the various potentially the bicyclist at all (Herslund & J0rgensen, 2003;
"interfering" conversations can be tuned out, Koustana'i, Boloix, Van Elslande, & Bastien, 2008).
just as they can in dichotic-listening experiments. One of the common reasons given for looked-
Situations occur, however, in which the ongoing but-failed-to-see accidents relates to the findings
 Attention and Performance II!! 105

from studies using selective-looking paradigms. Subjects in the experiment often miss or fail to
Essentially, these methods use the visual analog see the change (see Rensink, 2002; Simons &
of the dichotic-listening paradigm (Neisser Rensink, 2005, for reviews). In one dramatic live
& Becklen, 1975) to investigate the cause of demonstration of change blindness, an actor stops
inattention blindness-the failure to see certain a person on the campus of a university to ask
visual stimuli when focusing on other stimuli. directions (see figure 4.2). While the subject (the
Typically, subjects in a selective-looking para- person who is providing directions) explains the
digm are asked to focus on (and report on) the route to the actor, two people holding a large door
activities occurring in a visual display, but later walk between them. The actor is then replaced by
are asked to report on other (normally obvious) one of the door carriers, who assumes the role
visual events that occurred at the same time. of the lost actor after the door passes. Although
For example, subjects in a study by Simons the two actors look somewhat alike, they are
and Chabris (1999) watched a video showing clearly two different people. Yet nearly half of
a basketball being passed among three players the subjects in the study failed to recognize the
dressed in white shirts and another ball being change. The change-blindness phenomenon
passed among three players dressed in black reflects a strong role of memory, and expectation,
shirts. The subject's task was to count the passes as to what we attend to and what we become
made by the team dressed in white and ignore aware of (consciously). Similar to what occurs
the activities of the other team. What made this with the inattention-blindness effect, top-down
experiment particularly interesting was that an processes play a very important role in attention
unusual event occurred about halfway through and awareness.
the video-a person dressed in a gorilla costume Together, the results of experiments on the
walked through the middle of the two teams, Stroop effect, the "cocktail party phenomenon,"
stopped and faced the camera, then continued and inattention or change blindness illustrate
walking off screen. The "gorilla" was in perfect that humans are capable of searching selectively
focus and was superimposed on the images of and attending to specific information in the envi-
the ball game players-as if two separate movies ronment. Sometimes this intentional processing
were being projected onto a screen at the same facilitates performance by filtering out unwanted
time. After the trial, subjects were asked how information. At other times the process can have
many passes were made, then questioned as adverse consequences if the filtered informa-
to whether anything unusual had happened tion becomes important for task performance.
during the trial. Astonishingly, about half of the Awareness appears to be a kind of confluence of
subjects tested were completely unaware of the top-down and bottom-up processes (Most, Scholl,
"gorilla." An important additional finding was Clifford, & Simons, 2005).
that if subjects were not engaged in a primary
task (e.g., counting the passes), then nearly all of Automatic and Controlled
them reported the unusual event. The intentional Processing
task of processing specific information selectively
in the visual-search task (the ball game informa- Whereas a number of "early" processes (stimu-
tion) made the subjects unaware of other events lus encoding, feature detection, and so on) can
that were occurring in the same visual field, even apparently be conducted in parallel and without
if the event was unusual (for other examples see attention, it is clear that other processes prior to
Chabris & Simons, 2010; Driver, Davis, Russell, choosing an action cannot. For example, if you
Turatto, & Freeman, 2001; Hyman, Boss, Wise, are asked to detect whether the name of your
McKenzie, & Caggiano, 2010; Memmert & Furley, home town has a t in it, various mental opera-
2007; Simons, 2000). tions are required in order for you to come to
A variant of this paradigm, illustrating what the answer, and common experience tells you
is called change blindness, reveals a similar effect. that performing them would be detrimental
Subjects in a change-blindness study are typi- to a number of other tasks that might be called
. cally shown a series of still photos (or a video) of for at the same time (e.g., remembering your
a scene in which a portion of the visual display friend's phone number). This kind of process-
disappears or changes significantly over time. ing is what Schneider and Shiffrin (1977) have
106 Motor Control and Learning

FIGURE 4.2 Failure to detect changes to people in a real-world interaction.

Reprinted from D.J. Simons and D.T. Levin, 1998, "Failure to detect changes to people in a real-world interaction,'' Psychonomic Bulletin & Review 5: 644-649. By permission of
D.J. Simons and D.T. Levin.

called controlled processing. This type of process- set size of two). After the presentation of the
ing is (a) slow; (b) attention demanding, in that memory set, they received a frame" of letters 11

other similar tasks interfere with it; (c) serial in that could also be composed of one, two, or four
nature; and (d) strongly volitional," in that it
11
letters (e.g., B KM J represents a frame size of
can be easily stopped or avoided altogether. four). If either of the two letters in the memory
But Schneider and his colleagues (Schneider, set was presented in the frame (as J is in this
Dumais, & Shiffrin, 1984; Schneider & Fisk, 1983; example), the subject was to respond by press-
Schneider & Shiffrin, 1977) have also argued for ing a yes button as quickly as possible; if none
11 11

another class of information processing: automatic of the letters from the memory set was in the
processing. This form of information processing is frame, a no" button was to be pressed. Subjects
11

qualitatively different from controlled process- practiced under one of two conditions. In varied-
ing: Automatic processing (a) is fast; (b) is not mapping conditions, on successive blocks of trials
attention demanding, in that other operations do the memory set would be changed. This meant
not interfere with it; (c) is parallel in nature, with that a given letter in the frame sometimes would
various operations occurring together; and (d) is be a target, and on other blocks of trials it would
not "volitional," in that it is often unavoidable not be a target. Thus, seeing a Jin the frame, for
(Underwood & Everatt, 1996). example, would in one block of trials lead to a
Schneider and his colleagues studied these yes" response and in another block lead to a
11

processes using a variety of visual-search tasks. no." On the other hand, with consistent-mapping
11

In one example, Schneider and Shiffrin (1977, conditions, a given letter in the frame was either
experiment 2) gave subjects two types of com- always a target or never a target, leading to a
parison stimuli. First, subjects were presented consistent response when it was detected (i.e.,
with a memory set that consisted of one, two, or either yes" or no," but never mixed responses
11 11

four target letters (e.g., JD represents a memory in different blocks of trials).

 Attention and Performance IN 107

Figure 4.3 illustrates the results for the "yes" her usual schoolwork (reading about history) after
responses from these conditions after consider- being in the experiment; an E (a target letter in the
able practice for a memory set size of four. The experiment) would unavoidably "jump out" of the
varied-mapping condition (open circles) shows a page and distract her.
strong effect of the frame size, with reaction time For various reasons, however, the concept of
(RT) increasing approximately 77 ms per item in automaticity is not perfectly clear. One problem
the frame. On the other hand, there was virtually is that if a process is to be truly automatic, then
no effect of frame size on RT in the consistent- performance of any other simultaneous task
mapping condition. Both early in practice and should be possible without interference. Neu-
later in practice, for the varied-mapping conditions, mann (1987), in a review of this topic, argues that
search of the frame for a target letter seemed to be no information-processing activity has ever been
slow, serial, and strongly influenced by the number shown to be interference free across all secondary
of items to be searched; these features typify con- tasks. Whether or not interference is seen-as well
trolled processing. But after much practice with as the amount of interference-seems to depend
consistent-mapping conditions, the processing on the nature of, or the relationship between, the
was much faster, appeared to be done in parallel, two tasks (McLeod, 1977; Schmidt, 1987). Second,
and was not affected by the number of items to be the findings of Schneider and colleagues show-
searched; this is typical of automatic processing. ing no "resource cost" demands (i.e., no effects
In other experiments using very similar tasks, Sch- of frame size in figure 4.3), in fact, sometimes
neider and Fisk (1983) reported that after consider- do show a very small, positive slope for the
able practice with consistent-mapping conditions, consistent-mapping conditions, suggesting that
subjects could do these detections simultaneously these tasks are not completely interference free.
with other secondary tasks without interference. Recent thinking suggests that the controlled/
However, sometimes these detections became automatic dimension may better be thought of as
unavoidable, as if they were triggered off without a continuum (MacLeod & Dunbar, 1988; Moors &
much control. One subject told of interference with De Houwer, 2006; Underwood & Everatt, 1996).

800

700

600

"iii
g
Ii: 500

400

2 3 4
Frame size

FIGURE 4.3 Reaction time (RT) to detect target letters in a letter matrix of one, two, or four letters (frame size) for varied-
and consistent-mapping conditions; more letters to be searched led to no increase in RT for the consistent-mapping condi-
tion, suggesting automatic processing.
Reprinted, by permission, from W. Schneider and R. Shiffrin, 1977, "Controlled and automatic human information processing: I, detection, search, and attention," Psychological
Review 84: 20. Copyright© 1977 by the American Psychological Association.
108 !!I Motor Control and Learning

Certain information-processing activities are done in parallel and without much interference
attention free with respect to certain kinds of sec- from other tasks, the situation appears to be dis-
ondary tasks, and the problem is then to define tinctly different with respect to the organization
which kinds of tasks will and will not interfere and initiation of movements. Research studies
with what other kinds of tasks (Neumann, 1987). from various sources point independently to the
Various theoretical models have been developed view that only one movement can be initiated at a
that expand this view of automatic processing as a time. We turn next to the various lines of evidence
continuum, from memory-retrieval (Logan, 1988) for single-channel processing during these "late"
and parallel-distributed processing perspectives stages of information processing.
(Cohen, Dunbar, & McClelland, 1990), for example.
Psychological Refractory Period
The study of automatic processing, and of the
ways in which it is developed with practice, has Probably the most important evidence for single-
strong implications for understanding control channel processing emerges from the double-
in skills. Many processes in skills, whether they stimulation paradigm, in which the subject must
be the detection of individual letters or words as respond to two closely spaced stimuli that require
you read this sentence, or the recognition of pat- different responses. An example of this paradigm
terns of automobiles in traffic, can, with extensive is shown in figure 4.4. The subject is presented
practice, become faster and more efficient. Such with a sound (stimulus 1, or 51) that requires a
gains in skilled situations are of course extremely response by the right hand (response 1, or RJ
important, in that information-processing loads After a very brief interval, a light is presented (52)
are reduced so that the performer can concentrate that requires a response by the left hand (Ri). The
on other aspects of the situation (e.g., the meaning two stimuli are presented in different modali-
of a poem, navigation in driving), processing is ties and require responses by different hands (to
much faster, and many processes can be done in minimize structural interference). The stimuli
parallel. On the other hand, such automatic pro- are usually separated by at least 50 ms, and they
cessing can occur only with some cost," which is
/1
may be separated by as much as 500 ms or even
often seen when attention is drawn to the wrong more; that is, the second stimulus could, in some
place (distraction) or when an inappropriate situations, come well after the response to the first
movement is triggered (e.g., you respond to your stimulus is completed. The separation between
opponent's "fake" in tennis). the onsets of the two stimuli is called the stimulus
onset asynchrony (SOA). Further, the arrival of the
Interference and Movement signal onsets is usually randomly ordered, so that
the subject cannot predict the occurrence or timing
Production of a given stimulus (either 51 or 52) on a given trial.
Whereas the general findings are that many Thus, both stimuli must enter the information-
early" stages of information processing can be
11 processing system and be processed separately.

FIGURE 4.4 The double-stimulation paradigm (SOA = stimulus onset asynchrony, IRI =interresponse interval).
 Attention and Performance 111 109

Experimenters have been interested especially of 51 and R1 caused a marked increase in RT2 •
in the RT to the second of the two stimuli (RT2) In other studies, the amount of increase in 52
because it provides an indication about how caused by 51 and its processing can be 300 ms
the processing of the second stimulus has been or more (Creamer, 1963; Karlin & Kestenbaum,
affected by the ongoing processing of the first. 1968), making the RT for the second stimulus
The critical comparison is between RT2 when very slow-around 500 ms!
preceded by a response to 51versus RT2 when 51 Another important result in studies of refrac-
is not presented at all; that is, the" control RT is t toriness is the effect of the length of the 50A
a measure of RT2 when the subject does not have on R2 • Figure 4.6, also containing data from
51 presented at all. the Davis (1959) study, plots the values of RT2
Using this method, experimenters have shown for various values of the 50A, which ranged
repeatedly that the processing of 51 (through to from 50 ms to 500 ms. Notice that, because the
R1) is generally uninfluenced by the presence of RT1 was about 160 ms in this data set, all the
52 • However, the response to the second of the 50As greater than or equal to 200 ms occurred
two closely spaced stimuli (i.e., 52) is influenced when the second stimulus was presented after
by the presence of 51. Here, the general finding the subject had responded to the first (see also
is that RT2 with 51present is considerably longer figure 4.5). In figure 4.6, we see that as the 50A
than RT2 in the control condition (where 51is not was increased from 50 ms to 300 ms, RT2 was
present). Apparently the processing of 51 and R1 shortened systematically, until there was no
causes a great deal of interference with the pro- delay at all (relative to the control RT2 shown
cessing of 52 and~- This important phenomenon in the figure) with the longer 50As. The most
was discovered by Telford (1931), who named important points from figure 4.6 are that (a) the
it the psychological refractory period (PRP). 2 The delay in RT2 decreased as the 50Aincreased and
findings from Davis (1959), presented in figure (b) there was considerable delay even though Ri
4.5, are typical of the PRP effect. In this example, had already been produced (i.e., at 50As of 200
52 followed 51 by an 50A of 50 ms. The control ms or more).
RT2 (RT for 52 in separate trials in which 51 was One major exception to this generalization
not presented) was 124 ms. However, when 52 about the effect of the 50A should be mentioned.
followed 51, the RT2 was 258 ms-about twice If the second signal follows the first one very
as long as the control RT2 • Thus, the presence quickly, with an 50A as short as, say, 10 ms,

FIGURE 4.5 A demonstration of psychological refractoriness.

Adapted from Davis 1988.
110 Motor Control and Learning

300

250

Ii) 200
.§.
N
l-
a:
150

100

50 150 250 350 450

SOA (ms)

FIGURE 4.6 Refractoriness decreases as the stimulus onset asynchrony (SOA) increases.
Adapted from Davis 1988.

then the two signals are apparently dealt with as processing model, because sl and s2 could not be
a single, more complex stimulus (called a grouping processed together without interference.
effect). The two signals elicit the two responses at How does this theory explain psychological
about the same time but with a slightly greater RT refractoriness such as that seen in figures 4.5
for both than if only one of the responses had to be and 4.6? Referring back to figure 4.5, S2 was pre-
made to a single stimulus (see Welford, 1968). To sented 50 ms after S11 and RT1 was160 ms. Thus,
explain this phenomenon, some writers (see Wel- 160 - 50 = 110 ms remained before the completion
ford, 1968) have suggested the idea of a gate" that
/1
of RT 1 when S2 was presented. According to the
"slams closed" about 50 ms after the presentation single-channel hypothesis, processing of s2 must
of S1, presumably to prevent a second signal from be delayed until the channel is cleared. Thus the
entering the information-processing mechanisms predicted RT 2 will be the control RT 2 plus the
and interfering with the response to the first signal. 110 ms delay, according to the single-channel
If a second signal comes before this point, the first view. Note from the data that the control RT2 was
and second signals are processed together as a 124 ms, which makes the estimate of RT2 in the
unit, and the two responses are grouped. double-stimulation situation 124 + 110 = 234 ms.
If you look up the actual value of RT2 (258 ms)
The Single-Channel Hypothesis
given in figure 4.6, you will see that the predicted
A major contribution to the understanding of RT2 (234 ms) is fairly close. Thus, the duration of
human information processing in motor tasks RT2 was thought to be the control RT2 plus the
was made in 1952 when Welford proposed the amount that RT 2 overlapped with RT1 or
single-channel hypothesis to account for the well-
RT2 =Control RT2 + (RT1 - SOA) (4.1)
known findings about psychological refractori-
ness (see line 1 in figure 4.1). In his version of the We can see in figure 4.5 and equation 4.1 that, as
theory, Welford hypothesized that if S1 entered the size of the SOA increases, there is less overlap
the single channel and was being processed, then between the two RTs, and the predicted value of
processing of s2 had to be delayed until the single RT 2 decreases. This accounts for the finding that
channel was cleared-that is, until the response the RT 2 decreases as the SOA increases, as shown
to S1 had been started. This was a strict serial- in figure 4.6 and in other research (Welford, 1968).
 Attention and Performance 111

Evidence Against the Single-Channel or response complexity, and stimulus-response

Hypothesis (S-R) compatibility (see also Pashler, 1993, 1994).
While the original single-channel hypothesis Effects of Practice Practice has marked effects
accounts for some of the data, a considerable on the exact nature of the delay in RT2 in this
amount of evidence has suggested that the theory paradigm. Gottsdanker and Stelmach (1971)
is not correct in its details. According to the single- used 87 sessions of practice, 25 min each, in
channel hypothesis, RT 2 lengthens as a direct the double-stimulation paradigm for a single
function of the amount of overlap between RT 1 subject. They found that the amount of delay
and RT 2, as can be seen in equation 4.1. The first in RT2 steadily diminished from about 75 ms to
concern was that even when there was no overlap 25 ms over this period. But the delay was never
between RT1 and RT 2 (that is, when S2 occurred quite eliminated, suggesting that refractoriness
after the subject had already produced a response might have a permanent, structural basis in the
to S1), there was still some delay in RT 2 • Look at information-processing system.
figure 4.6. When the SOA was 200 ms, so that Complexity of Stimulus 1 Karlin and Keste-
S2 occurred 40 ms after R11 there was still some nbaum (1968) found that the delay in RT2 was
delay in RT 2 (about 50 ms) that did not disappear strongly affected by the number of choices
completely until the SOAhad been lengthened to involved in RT 1 . When S1 was a simple RT (one
300 ms. How can there be refractoriness, accord- stimulus, one response), with an SOA(until S2 ) of
ing to the single-channel view, when the RT1 and 90 ms, the amount of delay in RT 2 was approxi-
RT2 do not overlap at all? mately 100 ms. However, when the first response
Welford (1968) suggested that after R1 is pro- was a two-choice RT, the delay was approxi-
duced, the subject directs attention to the move- mately doubled, so that RT 2 was over 500 ms.
ment, perhaps to feedback from R1 to confirm A five-choice S1 produced an even larger delay,
that the movement was in fact produced correctly increasing RT 2 to about 630 ms (Keele, 1986).
before processing S2 • Thus, according to Welford' s Because increasing the complexity of S1 affects its
view, attention was directed to feedback from processing time according to Hick's law (chapter
R1 after the response, which delayed RT2, just as 3), the magnitude of the delay in RT2 apparently
the attention produced during the time from sl depends on the duration of processing for sl.
to R1 did.
But this explanation could not solve other Stimulus-Response Compatibility An impor-
problems. In the double-stimulation paradigm, tant moderating variable is S-R compatibility, or
as the SOA decreases, say, from 150 to 50 ms (in the relationship between the stimuli and responses
figure 4.5), the overlap between the two stimuli to be made (see chapter 3). Greenwald and Schul-
increases by exactly 100 ms; the single-channel man (1973) varied the compatibility of the first
model assumes that the delay in RT2 is a direct response (S1-R1). In a compatible condition, S1 was
function of this overlap, so the model predicts an arrow pointing to the left or right, and R1 was
that RT2 should be increased by exactly 100 ms a hand movement in the indicated direction. In a
in this example. Generally, the increase in RT 2 less compatible (or less "direct") condition, S1 was
in these situations has been much smaller than the visually presented word "left" or "right," and
expected on the basis of the model (Davis, 1959; R1 was again the manual response in the indicated
Kahneman, 1973; see Keele, 1986, for a review). direction. Thus, reading the word "left" or "right"
These effects are probably attributable to the fact required transformations from the stimulus to the
that some processing of S2 was being completed response that were more complicated and less
while sl and its response were being processed, "natural" than seeing the arrows (see chapter 3). In
which, strictly speaking, is contrary to the single- both conditions, S2-Rz was unchanged; S2 was the
channel hypothesis. number 1 or 2 presented auditorily, and Rz was the
vocal response "one" or "two." Figure 4.7 gives the
Moderating Variables in Psychological results; C on the horizontal axis refers to the con-
Refractoriness trol RT2 with no S1 or R1 required. When S1-R1 was
Various factors act to change, or modify, the not compatible (open circles), RT2 was lengthened
effects just seen in double-stimulation situations. considerably and refractoriness was increased as
Some of these variables are practice, stimulus the SOA decreased, as we have seen before (e.g., in
112 Motor Control and Learning

650

550

450

FIGURE 4.7 Reaction time (RT) to the second of two closely spaced stimuli at various stimulus onset asynchronies
(SOAs) as a function of the stimulus-response (S-R) compatibility of the first reaction; the compatible arrow stimuli produced
no refractoriness for the second reaction.
Adapted, by permission, from A.G. Greenwald and H.G. Schulman, 1973, "On doing two things at once: Elimination of the psychological refractory period effect,'' Journal of
Experimental Psychology 101: 74. Copyright© 1973 by the American Psychological Association.

figure 4.6). But when S1-R1 was compatible (filled (see Creamer, 1963; Karlin & Kestenbaum, 1968),
circles), there was no lengthening of RT2 at any of which is a very long time in fast games such as
the SOAs. If the compatibility among the stimuli basketball and hockey and daily activities such
and responses is very high, the usually devastating as driving in high-speed traffic. But this is not the
effects of the first signal and its response can be only problem. If the player has "taken the fake,"
reduced or even completely eliminated. (i.e., responded fully to S1), then he not only will
suffer a delay in the RT to the "real" movement
Implications of Refractoriness (S 2), but also must overcome the momentum that
for Practical Situations the first movement has produced-plus make up
We have seen that the second of two closely any distance that he may have been traveled in
spaced reactions can suffer considerably in pro- the wrong direction (see also figure 3.17 on p. 84).
cessing speed, and this fact can have important Based on findings from the PRP literature, a
practical implications. For example, consider basic principle of faking is that the actual move
many fast-game situations, which often involve should follow the fake by enough time that the
offensive and defensive players. Here, we often second move is treated separately rather than
see the offensive player "fake" an opponent by grouped with the first one. Thus, the SOA should
displaying the initial parts of one action (e.g., a probably be around 50 ms or longer. Also, the
slight movement to the right) followed quickly by second move must not follow the fake by so long
a different action (e.g., a movement to the left) that that the refractory effects of responding to the fake
is actually carried to completion (such as what have dissipated-probably not more than 250 ms
the hockey player is trying to do to the goalie (see figure 4.6). It would be interesting to study
in figure 4.8). If the defensive player responds effective fakes in sport to discover whether the
to the first move, a full RT (about 150-220 ms) most effective SOAs correspond with estimates
will be required, plus the added delay caused by from experimentation. Such intervals may repre-
refractoriness, before the defensive player can sent a part of natural defensive actions in other
even begin to respond to the second move. Thus, species. Watch a rabbit being chased by a dog. The
RT to the second move could be as long as 500 ms rabbit runs with unpredictable directional changes,
 Attention and Performance 111 113

FIGURE 4.8 Classic example of a psychological refractory period (PRP): The shooter is trying to draw the goalie out of
position by faking the delivery of the shot.

probably using intervals between directional and that responses must therefore be separated
changes that are highly effective in confusing the considerably in time. How closely in time can two
dog. Refractoriness might be an important survival responses be produced, provided that they are not
mechanism in animals other than humans. grouped and produced simultaneously? Kahne-
man (1973) examined this separation-called
Separation Between Responses the interresponse interval (!RI) in figure 4.4-as it
Evidence about psychological refractoriness is affected by bringing the stimuli closer to each
suggests that the perceptual-motor system has other in time. In figure 4.9, we have plotted some
difficulty responding to closely spaced stimuli, of the data from Smith's (1969) study, as Kahneman

500

400

Ui'
s 300
!!;

200

50 150 300 400 500

SOA(ms)

FIGURE 4.9 The relation between the interresponse interval (IRI) and the stimulus onset asynchrony (SOA).
Data from Smith 1969; adapted from Kahneman, 1973.
114 Motor Control and Learning

(1973) did. The SOAs were 50, 150, 300, and 500 ms, conflicting signals. Consequently, rather than
and the separations between responses (i.e., the seeing refractoriness only as a "problem" to be
IRis) that resulted from these intervals are plotted. overcome by the motor system in producing
As the SOA decreased (moving leftward on the rapid-fire actions, we can also view refractoriness
graph), the interval between the two responses as protective. It tends to ensure that responses to
decreased, but only to a certain point. It appears important stimuli are appropriate and complete.
that no matter how small the interval between This is consistent with Neumann's (1987, 1996)
stimuli, provided they are not grouped so that view of attention as discussed earlier, according
their responses are emitted simultaneously, an to which interference can be the result of an action
approximately 200 ms separation occurs between having been selected.
sequential responses. This is a most important
result and suggests that if a signal gets into"
/1
Inhibition of Return
the information-processing stages up to a certain Another finding in the literature that reveals an
point, a response to it is generated. If another interference "after-effect" following the orienting
stimulus is presented soon afterward, indicating of attention has been termed inhibition of return
that the system should do some other action, the (Posner & Cohen, 1984). The typical sequence of
second action must wait for at least 200 ms before events is illustrated in figure 4.10. A trial begins
it can be initiated. with the presentation of an orienting cue-a type
Given this general principle, why is it that of warning cue that requires no overt response
skilled piano players can produce many move- but simply alerts the subject that a stimulus will
ments of the fingers in which the separation follow shortly, either at the same location or at a
between movements is far less than 200 ms? different location. An imperative signal occurs
Surely this observation contradicts the notion after a variable SOA either at the originally
that a response can be emitted only every 200 oriented cue location (top of figure 4.10) or at
ms. We will discuss this in detail in chapter 6, an uncued location (bottom of figure 4.10). The
but for now, the issue mainly concerns the defi- subject's goal is simply to make a response as fast
nition of a "response." The system can prepare as possible to the imperative signal. In compari-
a "response" to a given stimulus that in itself is son to the situation with the double-stimulation
complex and involves many movements in rapid paradigm (figure 4.4), no response is required to
succession. The "response" that is planned in the orienting cue. But, similar to what we see with
the response-programming stage is still just one the PRP effect, the length of the SOA appears be
response, but it may have many parts that are the critical determinant of how quickly the subject
not called up separately. This can be thought of will respond to the imperative signal.
as output chunking, whereby many subelements The results from the original Posner and Cohen
are collected into a single unit, controlled by what studies (1984), which have been replicated often,
is called a motor program (Keele, 1973; Schmidt, are shown in figure 4.11 (p. 116). Two important
1976a). Also, according to this general view, these effects emerge as a function of the SOA. First, if
programmed outputs occur in discrete "bursts" the imperative stimulus occurs at the same loca-
separated by at least 200 ms. These discrete ele- tion as the orienting cue and with a short SOA (less
ments are difficult to view directly, however, than 200 ms), then RT is facilitated relative to an
because the muscles and limbs smooth out the imperative stimulus location that is different than
transitions between them, giving the impression the orienting cue (the open symbols have less RT
that we respond continuously. In essence, then, than the filled symbols in figure 4.11). This is the
many different movements can be separated by less beneficial effect of orienting the subject's attention
than 200 ms if these movements are all contained toward the location at which the imperative cue
in one "response." occurred, even though no overt response was
It is fortunate that S2 cannot get into the required. More important, however, is what
system to disrupt the preparation of a response happens when the imperative signal occurs 300
to S1 . When preparing a response to a danger- ms or more after the orienting cue. In this case,
ous stimulus, subjects will be successful only if completing the orienting response to the precue
they can process the information and produce delays a response to the orienting (cued) location,
the movement without interference from other compared to the RT in response to the cue that
 Attention and Performance im 115

Variable SOA duration (0-500 ms)

FIGURE 4.10 Events of the typical inhibition-of-return (IOR) paradigm. Top: The imperative signal occurs in the same loca-
tion as the orienting cue. Bottom: The imperative signal occurs in a location other than the orienting cue location.
Adapted, by permission, from M.I. Posner and Y. Cohen, 1984, Components in visual orienting. In Attention and performance X: Control of language processes, edited by H.
Bouma and D.G. Bouwhuis (Hillsdale, NJ: Erlbaum), 531-556.

was not at the orienting location (the open sym- Once that location has proven not to be the target
bols now have more RT than the filled symbols of the search, its previous importance is inhibited,
in figure 4.11). thereby allowing other potential locations to rise
The "inhibition of return" seems to be a protec- in salience. Another suggestion is that the inhibi-
tive mechanism that prevents returning attention tion has a motor basis-having made a saccadic
to a previous orienting cue that captured attention eye movement to the precue makes a response to
but did not require a response. Hypotheses differ the same cue slower than to a new target location
regarding the mechanisms involved in produc- (Fischer, Pratt, & Neggers, 2003). Klein has pro-
ing the inhibition-of-return effect. Some suggest vided reviews of this extensive literature (Klein,
that inhibition of return is caused by an atten- 2000, 2004; Taylor & Klein, 1998).
tion shift, whereby the previously attended-to
cue loses salience (importance) after being fully Revised Single-Channel View
processed (e.g., Fecteau & Munoz, 2003). Such a In sum, a single channel does seem to exist,
view would explain how one searches a map for contrary to the conclusion presented earlier in
a particular location, or searches a page for Waldo the section about the (original) single-channel
in a "Where's Waldo" scene. The start of the hypothesis. But the single channel does not
map search combines both top-down (directed appear to apply to all stages of processing as
search) and bottom-up (something that appears to Welford's (1952) version of the theory stated.
"jump" off the map) features of orienting salience. Rather, it appears that parallel processing can
116 Motor Control and Learning

here) represents the bottleneck where single-chan-

nel processing occurs. According to the model in
425 figure 4.12, the processing of S2 in response selec-
tion and programming must wait (as indicated by
the horizontal line) until sl has cleared this stage.
Ui' 400
_§_ In other words, only one response at a time can
Cll be selected and programmed. Further processing
.§ of the second of two closely spaced stimuli is put
c: 375
.!2 on hold until the response selection and program-
0co
ming for the first stimulus are complete. This leads
ti. 350
us to ask whether the response-selection stage or
the response-programming stage (in our terms
325 in chapter 3; different from Pashler's), or both of
the stages, might be the precise location of the
bottleneck in processing.
0 100 200 300 400 500
SOA (ms)
Attention During Movement
FIGURE 4.11 Effects of the relation between the ori-
enting cue location and stimulus location as a function of To this point, we have examined attentional
stimulus onset asynchrony (SOA): the inhibition-of-return processes present during the RT period, which,
(IOR) effect
of course, takes place prior to movement. But
Adapted, by permission, from M.I. Posner and Y. Cohen, 1984, Components
in visual orienting. In AttenNon and performance X: Control of language one can conceptualize a "stage" of processing,
processes, edited by H. Bouma and D.G. Bouwhuis (Hillsdale, NJ: Erlbaum),
531-556.
following response programming, in which the
individual carries out the movement and keeps
it under control. Strictly speaking, it is not a stage
occur during the early stages of information in the sense of the other stages, as it does not
processing and that a single channel is properly occur during RT (see chapter 3). Nevertheless, it
placed during the stages in which decisions are is important to consider the attentional charac-
made about the response. Pashler (1993, 1994) has teristics of these processes.
presented persuasive evidence suggesting that In thinking about our own skilled movements,
the response-selection stage (which, for Pashler, we are left with the impression that skills require
includes response programming as we define it conscious awareness for their performance. We

Time

FIGURE 4.12 Pashler's bottleneck theory. Response selection and programming for the second stimulus must wait until
the selection and programming of the first response are complete.
Adapted, by permission, from H. Pashler, 1993, "Doing two things at the same time," American Scientist 81 (1): 52.
 Attention and Performance Iii! 117

must be careful with these analyses, though, task technique, the subject performs a primary
because we cannot be sure whether it is the task for which the attention demand is of interest.
movement itself that requires awareness or the But the subject is also occasionally presented with
programming and initiation of future movements another task. These secondary tasks are of two
(e.g., in a dive, when to "open up" to enter the types: continuous and discrete.
water). We are in a weak position for determin-
Discrete Secondary Tasks
ing the extent to which movements themselves
require consciousness when we use only the In the discrete secondary task, a stimulus, called
methods of introspection. a probe, is presented at various times or places in
On the other hand, sometimes it seems that a the performance of the primary task, and a quick
particular movement is performed without aware- response to the probe is made with an effector not
ness (some consider such movements" automatic"), involved in the main task. The probes are often
especially when it is a part of a well-learned auditory, and the response may be either manual
sequence (e.g., a part of a dance routine). People or vocal. This secondary-task method has been
who routinely drive a car with a standard transmis- called the "probe technique" and is the subject of
sion sometimes find, when driving an automatic a large literature within the study of movement
transmission car, that they attempt to depress the control (Abernethy, 1988, 2001). We discuss some
"clutch" at a stop sign even though the car has no details of this method later in the chapter.
clutch pedal. Do you typically realize that you have Continuous Secondary Tasks
produced this movement only after having done
Continuous secondary tasks are those that are
so? Have you ever found, in buttoning your shirt
performed together with and throughout the dura-
and thinking sleepily about the upcoming day's
tion of the primary task. Experiments reported by
activities, that you are performing the motions
Welch (1898), for example, assessed maximum
even where a particular button is missing? Have
handgrip strength during the simultaneous perfor-
you ever left the house and wondered if you turned
mance of various tasks, such as reading, writing,
off the burner on the stove? These examples, plus
arithmetic calculations, and various visual and
many others that come to mind, suggest that not all
auditory perception tasks. In all cases the maxi-
movements require attention for their performance
mum force output during the control trials (with
while some seem to require considerable attention. no secondary task) was greater than when the
Some of these "mindless" aspects of performance
task was performed simultaneously with another
are described by Langer (1989).
task. Importantly, Welch (1898) was the first to
When a movement requires a great deal show that different dual-task situations created
of attention, what exactly is involved? As we
different "amounts" of interference. For instance,
mentioned earlier, attention could be devoted to
grip-strength performance was more profici~nt
the response-programming stage for future ele- when subjects added two numbers, compared with
ments in a long sequence. Attention also could be multiplying two numbers, leading to the conclu-
devoted to other aspects of the environment. In sion that the multiplication task demanded more
driving, for example, the movements themselves of the subject's attention than did the addition task.
might be carried out with minimal attention; As an example, one research area that fre-
instead, attention is directed to traffic patterns quently uses continuous secondary tasks has
and other relevant features of the environment. to do with the attention demands of gait and
Another possibility is that we pay attention to posture. Beginning with the research of Kerr,
our movements to fine-tune their control-to Condon, and McDonald (1985), numerous
carry out any necessary corrections, which can be researchers have assessed attention demands of
thought of as "responses" as well. We consider standing balance under various conditions (see
some of these questions in the following sections. review by Woollacott & Shumway-Cook, 2002).
Balance and gait control represent increasing
The Secondary-Task Technique challenges with advances in age, as falls often
An effective tool for evaluating the role of atten- result in major health concerns. Although there
tion in simple movements involved secondary are many contradictions among the findings of
tasks, introduced in chapter 2. In the secondary- these studies, an important association exists
118 Motor Control and Learning

between dual-task performance (and thus atten- In these examples, the two tasks are internally
tion demand) and incidents of falling among the generated and not" driven" by any obvious envi-
frail and elderly (Beauchet et al., 2009). Thus, ronmental signal. It would thus seem that not all
the use of the secondary-task method may have of the interference we observe among tasks has to
important clinical implications for early detection do with the processing of environmental stimuli
of at-risk individuals. leading to a response. Rather, an additional source
In some experiments, the experimenter is inter- of interference might have to do with the control
ested in discovering how the subject coordinates of the limbs per se. Although it is tempting to
the execution of two (or more) simultaneous attribute an attentional "overload" explanation
movements. In this case, it is not the attention to these and related effects, the issue is much
demands of one or the other that are of interest more complex than this. In fact, later in the book
per se; rather, the primary concern is how the we devote an entire chapter to the problems of
two tasks become regulated in space and time. simultaneously coordinating the actions of two
You are probably well aware of the difficulty in or more effectors (chapter 8).
rubbing your stomach with one hand while pat-
ting your head with the other. Of course, doing The Probe Technique
either of these tasks by itself is easy for most of This secondary-task method assumes a fixed (non-
us; but when we attempt them together, the two changing) attentional capacity. The "pie chart,"
hands are particularly difficult to control, and a illustrated in figure 4.13, is an effective way to
great deal of mental effort seems to be required. In illustrate this fixed-capacity model. The "whole
another example (Klapp et al., 1985), it is particu- pie" represents the total attentional capacity avail-
larly difficult to tap a regular rhythm with your able. If the capacity required for the primary task is
left (nonpreferred) hand while tapping as quickly low (about 75% of the fixed capacity, as illustrated
as you can with your right (preferred) hand. Such in the diagram on the left in figure 4.13), then
effects are not limited to these clever little demon- the capacity remaining for processing the probe
strations. In many piano performances, the two stimulus will be relatively large, which should
hands must perform with different rhythms for lead to a probe RT that will be fast. However, if the
a short while, and most pianists claim that this primary task demands a much larger proportion
is one of the most difficult tasks in these pieces. of the fixed capacity (say, about 90%, as illustrated
Rubato involves the gradual speeding or slow- on the right side of figure 4.13), then much less of
ing of one hand with respect to the other, also the spare capacity remains for the probe stimulus,
considered a very advanced technique in piano and its processing will be slower. Thus, the early
playing (Peters, 1985). work of this kind assumed that the duration of the

Rapid probe RT Slow probe RT

Attention
to probe

Fixed capacity Fixed capacity

FIGURE 4.13 Assumptions of the probe reaction-time (RT) task: With fixed total capacity, attention to the probe decreases
as the complexity of the primary task increases.
Reprinted from M.I. Posner and S.W. Keele, 1969, Attentional demands of movement. In Proceedings of the 16" Congress of applied physiology (Amsterdam, Amsterdam: Swets
and Zeitlinger). By permission of M.I. Posner.
 Attention and Performance I!! 119

probe RT could give an indication of the attention perhaps due to positioning the pointer accurately
requirements of the main task. in the target zone. Finally, note that the probe RT
In an early study by Posner and Keele (1969), increased as the target size decreased, especially
subjects made 700 ms wrist-twist movements of near the end of the movement, suggesting that
a handle through a range of 150°, attempting to the movement became more attention demand-
move a pointer to either a large or a small target ing as end-point precision increased (see also
area. The experimenters presented probe signals Ells, 1973; Salmoni, Sullivan, & Starkes, 1976).
at various points during the movement-at the In another of Posner and Keele's (1969) experi-
start or at 15°, 45°, 75°, 105°, or 135° of handle ments, blindfolded subjects made a movement
movement. The probe RTs are plotted in figure to a stop, which required essentially no target
4.14 as a function of their location within the accuracy; the movement was analogous to push-
movement. The horizontal line represents the ing a door closed. In this example, there was no
no-movement, control probe RT (i.e., RT in the increased probe RT beginning at any of the later
absence of a primary task). Several features of points during the movement, implying that this
this figure are important to note. First, the find- movement used no attention once it was initiated.
ing that the probe RTs were always larger than The probe technique has been used in a number
the corresponding values of the control probe of ways in attempts to assess the attentional
RT was taken as evidence that performing the demand involved in performing various jobs
primary task (making the pointing movements) (Ogden, Levine, & Eisner, 1979); in the control
generated interference-that is, they required of posture and gait (Abernethy, Hanna, & Plooy,
attention. Next, probe RT for both tasks showed 2002; Ojha, Kern, Lin, & Winstein, 2009; Woollacott
a U-shaped function; there was a marked slowing & Shumway-Cook, 2002); and in various sport-
of the probe RT at the beginning of the movement, related activities such as catching a ball (Populin,
with somewhat less slowing at the end. This Rose, & Heath, 1990; Starkes, 1987), shooting a
finding suggests that attention to the movement pistol (Rose & Christina, 1990), receiving tennis
is strongest at the beginning, that a "relatively" and volleyball serves (Castiello & Umilta, 1988),
attention-free portion occurs near the middle, and skating and stick-handling skills in hockey
and that attention is used near the end again, (Leavitt, 1979). A very important and recent

500

400

UI
§.
Ii: 300

200

0 15 45 75 105 135
Position in movement (degrees)

FIGURE 4.14 Probe reaction time (RT) elicited at various points in movements to large and small targets.
Reprinted from M.I. Posner and S.W. Keele, 1969, Attentional demands of movement. In Proceedings of the 16'" Congress of applied physiology (Amsterdam, Amsterdam: Swets
and Zeitlinger). By permission of M.I. Posner.
120 Motor Control and Learning

addition to this literature involves the attention ated capacity is or is not used. Conversely, find-
demands of time sharing during driving, which ings that the secondary-task performance is not
we consider in more detail later in this chapter. elevated by the primary task do not necessarily
mean that the primary task is automatic; rather,
Problems With Secondary Tasks they indicate that the particular processes needed
A number of potential problems with the second- in the secondary task are not also involved in that
ary-task paradigm require that interpretation of primary task. Of course, other secondary tasks
these experiments be made with caution. First, the could probably be found that would interfere
method assumes a fixed, unitary capacity, which with the primary task (Neumann, 1987); if so, this
is, of course, contrary to Kahneman' s (1973) views would force us to conclude that the primary task
and contrary to the idea of pools of resources is not automatic after all.
discussed earlier. Second, McLeod (1977, 1980) What has usually been meant by the conclu-
has identified a number of other difficulties. One sion that a task can be performed" automatically"
of these was shown in his finding that a probe is that it can be performed without interference
RT with a manual response interfered with a from other mental tasks involving (conscious)
simultaneous tracking task, whereas a probe RT information-processing activities. This fits with
with a vocal response hardly interfered at all. If our subjective experiences about "automatic"
the probe RT actually measures some general, movements that are performed without con-
"spare," central capacity (attention), why was sciousness. But, as has been said previously
more of this capacity available when the only (Neumann, 1987; Schmidt, 1987), it is probably
change was in the nature of the movement to be best to think of automaticity with respect to some
made in response to the probe? other simultaneous secondary task(s).
Interpretation of results from continuous dual-
task experiments involves a similar issue. A study Action-Centered Interference
by Yardley, Gardner, Leadbetter, and Lavie (1999) A different way to examine interference effects
illustrates this point well. In this experiment, during movement was introduced by Tipper,
subjects attempted to stand perfectly still on a Lortie, and Baylis (1992). This experimental
platform while performing various secondary approach relates to the everyday problems of
tasks. Of importance was the difference observed interference that one encounters when reaching
in balance when subjects were asked to count for an object in a cluttered environment, such
backward by 7s from a three-digit number. When as reaching for a paper on a desk that has other
subjects performed the counting task silently, the papers scattered on it, or reaching for an apple in a
investigators observed no increase in postural basket containing various fruits (Castiello, 1996).
sway compared to the single-task control condi- The questions of interest relate to how potentially
tion. However, when subjects counted backward distracting objects in the environment produce
out loud, postural sway increased dramatically, interference effects in either the preparation or
but no more so than when they simply repeated the execution of movement.
a three-digit number over and over. Yardley and The task for subjects in studies by Tipper and
colleagues concluded that it was not the attention colleagues (1992) was to respond to the illumi-
demanded by the secondary task that interfered nation of a light by pushing one of the buttons
with performance; rather, increases in postural located on a table directly in front of them.
sway were more likely due to the extraneous As illustrated in figure 4.15, the buttons were
postural movements caused by vocalizing aloud, arranged in three rows of three buttons each; near
or the attention demanded by the act of speak- each button were two lights-a red light indicated
ing, or both. that the button should be pushed, and a yellow
Whatever the exact nature of the interference light indicated a distracter (nontarget) button that
might be, clear interference is found when various was simply there and was to be ignored. Response
secondary tasks are paired with a variety of dif- times to move from the home button to the target
ferent primary tasks. These experiments seem to location were very fast in conditions in which
tell us that various movement tasks require many none of the distracter lights were illuminated
of the same processes that auditory-manual when a target light appeared. However, on some
probe tasks do, and not that some undifferenti- trials, a yellow light was illuminated at the same
 Attention and Performance 121

l
660

640

620
Ci)
.§.
!!: 600

580

560

Front to back Back to front

Direction of movement

FIGURE 4.15 Action-centered interference in various arrangements of targets and distracters.

Adapted, by permission, from S.P Tipper, C. Lortie, and G.C. Baylis, 1992, "Selective reaching: evidence for action-centered attention," Journal of Experimental Psychology:
Human Perception and Performance 18: 893, 896. Copyright 1992 by the American Psychological Association.

time that a red target light appeared. The critical ject (arrangement at upper right of figure 4.15).
issue addressed by Tipper and colleagues was In this configuration, compared to no-distracter
whether the distracters (the yellow lights) would conditions, targets in the middle row were now
affect response time to reach the target button, unaffected by distracters in the front row, but
and if so, which distracter locations would have were severely interfered with by distracters in
the largest effects. the back row (back and front as viewed from the
Two sets of results are presented in figure 4.15. subject's perspective). Therefore, in both spatial
On the left are the findings for a layout in which arrangements, the interference was largest when
the subject's hand was moving away from the the distracter was located between the start and
front of the table (which was the "start position") end points of the movement, but not when it was
to contact a target in the middle row (i.e., one located "beyond" the target.
of the filled boxes). In this condition, only the The findings of Tipper and colleagues (1992)
distracters in the front row (FR) interfered with have been replicated and extended in a number
response time compared to a no-distracter (ND), of experiments showing that selective visual
control condition. When distracters appeared in interference during movement is highly depen-
the back row (BR), the response time was almost dent on the nature of the action (e.g., Howard
the same as in the no-distracter, control condi- & Tipper, 1997; Meegan & Tipper, 1998; Pratt &
tion-hence, there was essentially no interference. Abrams, 1994). The "amount" of interference
Tipper and colleagues were able to reverse this produced by a particular distracter appears to
pattern of findings by having movements start at depend on its functional significance to the inten-
the back of the table and move toward the sub- tions of the main action (Weir et al., 2003; Welsh
122 Motor Control and Learning

& Pratt, 2008). For example, when distracters phone in the other, and who had what appeared
are located close to a target, they tend to have to have been a newspaper balanced on the steer-
the largest interference effect on the kinematics ing wheel-all at approximately 70 miles per
of the grasping hand (Jervis, Bennett, Thomas, hour" (Ambrose, 1997).
Lim, & Castiello, 1999; Mon-Williams, Tresilian, With increasing frequency, one hears anecdotal
Coppard, & Carson, 2001). In contrast, objects reports of traffic accidents that occurred while a
located along the path of the movement tend driver appeared to be using some type of hand-
to have larger effects on the kinematics of the held communication or entertainment device. The
reaching action (Mon-Williams et al., 2001). This controversy was fueled by an analysis of acci-
view suggests a rather radical departure from dent reports and cell phone call records, which
interference effects during movement as implied revealed a fourfold increase in accident risk when
by various capacity views of attention. Rather, the the driver had been using a cell phone at the time
findings support Neumann's (1996) view that the of the incident (Redelmeier & Tibshirani, 1997).
nature of the selected action drives attention and The researchers compared the elevated risks of
determines which potential factors will or will not distracted driving to the dangers of driving under
cause interference. the influence of intoxicants (see also Strayer,
Drews, & Crouch, 2006).
Distracted Driving Lawmakers in many countries around the
"Several years ago I observed what I believe to be world have debated how to deal with the issue
the all-time record for distracted drivers. Driving of distracted driving, and researchers have
north on I-95 into Baltimore, I was passed by a responded with a wealth of data to inform them.
male driver who was holding a Styrofoam cup Although this is a current "hot" topic, investiga-
and a cigarette in one hand, and a cellular tele- tions on cell phone use were initiated by Brown,

l.D. BROWN ON DISTRACTED DRIVING

Research on talking on a phone during driving, though a topic of considerable recent research
activity, was investigated many years ago by Brown and colleagues (1969). As evident in the
following quotes, their theoretical orientation was clearly concerned with attention. Although
single~channel theory (according to which switching between tasks is necessary in order to
carry out more than one task) dominated their thinking, their hypotheses and findings remain
as relevant today as they were in 1969. It is unfortunate that their plea for additional research
on specific outstanding issues was not taken up in earnest for more than two decades.
"Having to use a hand microphone and having to manipulate push buttons to make or take
a call will be inconvenient and may impair steering, gear changing, or other control skills. This
is a pr6blem which may be solved by engineering advances and is not the concern of the pres-
ent paper. A more important and lasting problem arises from the hypothesis that man can be
considered to act asa single communication channel of limited capacity. The prediction from this
hypothesis is that the driver will often be able to telephone only by switching attention between
the informational demands of the two tasks. Telephoning could thus interfere with driving by
disrupting visual scanning, since visual and auditory information would have to be transmitted
successively. It could also interfere by overloading short-term memory and impairing judgment
of relative velocity... :· (p. 419).
"The general conclusion must be that some mutual interference between the concurrent tasks
is inevitable under conditions of telephoning while driving on the road. The results suggest that,
although more automatized control skills may be affected minimally by this division of attention,
some perception and decision skills may be critically impaired. The extent to which this impairment
is a function of the driving task, the informational content of the telephone message, and the
individual characteristics of the driver must remain a subject for further research" (pp. 423-424).
 Attention and Performance 123

Tickner, and Simmons (1969) over 40 years ago. gest that the distraction remains about the same
Their subjects made safety margin judgments (Cooper & Strayer, 2008). However, there is no
(whether or not the car could fit through gaps) by evidence that the distraction from cell phones is
either driving through or around markers on the eliminated with practice. Also, conversations that
road). On some trials the subjects also performed require a larger number of reasoning operations
a reasoning task via a hands-free telephone. impair performance more than tasks involving
Performing the reasoning task while driving fewer operations (Briem & Hedman, 1995).
elevated the frequency of gap judgment errors by In sum, the impact of the driving environment,
almost 13%. (See the authors' concerns about this driver experience, and the nature of the con-
increase in ''I.D. Brown on Distracted Driving.") versation is consistent with general attentional
Sending and receiving text messages is an obvi- capacity limitations in performance discussed
ous source of structural interference during driv- earlier in this chapter. However, more complex
ing-one cannot fully scan the driving environ- driving environments usually impose greater
ment while looking at a communication device, visual demands on the driver, and some have
and the texting hand is not available for other argued (e.g., Strayer et al., 2003) that secondary-
subtasks. But what are the dangers in simply task performance interferes with visual attention
talking on a cell phone during driving? Are hand- processes, causing an effect like inattentional blind-
held cell phones more dangerous than hands-free ness (see our earlier discussion in this chapter;
units, and is the whole issue of interference from Yantis, 1993). Cell phone conversation reduces
cell phones related only to structural interference the capacity to perceive changes in the visual
(i.e., looking at the screen, entering keystrokes)? environment such as traffic patterns (Trbovich
Is talking on cell phone more dangerous than & Harbluk, 2003).
carrying on a conversation with a passenger
Cell Phone Use Versus Other Driver
in the car? We consider these and other issues
in the next sections (see also reviews by Caird, Distractions
Willness, Steel, & Scialfa, 2008; Collet, Guillot, & Cell phone use is only one of many potential
Petit, 2010a, 2010b; Haigney, & Westerman, 2001; distractions in driving (Cohen, 1997). Conver-
Horrey & Wickens, 2006; McCartt, Hellinga, & sations with other passengers, listening to and
Bratiman; 2006). adjusting audio equipment, and reading instru-
ment panels and mirrors are just a few of the
Does Cell Phone Use Affect Driving? ways in which attention might be distracted.
The degree to which driving is compromised by The research is not entirely clear, however, on
cell phone conversation appears to depend on how these other sources of distraction influ-
at least three factors: the driving environment, the ence driving (Consiglio, Driscoll, Witte, & Berg,
characteristics of the driver, and the nature of the 2003; Irwin, Fitzgerald, & Berg, 2000; Strayer &
conversation in which the driver is engaged. The Johnson, 2001). Tasks that require the driver to
decrement in driving performance during talking divert visual attention from the outside to the
on a cell phone has been found to increase with inside environment (e.g., tuning a radio) appear
increased traffic demand (Lee, Caven, Haake, to be detrimental to performance (McKnight &
& Brown, 2001; Strayer, Drews, & Johnston, McKnight, 1993), presumably because hazards
2003) and environmental complexity (Strayer to driving come mainly from the outside. The
& Johnston, 2001). Also, a greater performance intensity and complexity of the cognitive/visual
decrement is observed under dual-task driv- processes are important contributors to interfer-
ing conditions for older drivers compared to ence, and not simply the manual handling of the
younger drivers (Alm & Nilsson, 1995; Hancock, device as many have believed.
Lesch, & Simmons, 2003; McKnight & McKnight, One of the arguments frequently raised against
1993), although there is some discrepancy in the legislation on cell phone use during driving
literature regarding the role of practice. Some suggests that the distraction is no greater than
studies suggest that practice lessens driving dis- talking with a passenger (which is both common
traction (Brookhuis, de Vries, & de Waard, 1991; and legal). In an epidemiological study, McEvoy,
Chisholm, Caird, & Lockhart, 2008; Shinar, Trac- Stevenson, and Woodward (2007) found that the
tinsky, & Compton, 2005), whereas others sug- accident-risk odds rose by a factor of 1.6 when
124 Motor Control and Learning

one passenger was in the vehicle and by a factor we specifically direct our attention to differ-
of 2.2 with two or more passengers. However, ent sources of information, such as the internal
those increases were much less than the accident sources of the feedback that naturally arise during
risk associated with cell phone use, which rose movement or an object in the environment that
by a factor of 4.1. These findings are supported represents the outcome goal of an action? In
by experimental evidence (Drews, Pasupathi, & other words, what is the effect when we vary the
Strayer, 2008) suggesting that in-vehicle conver- "focus" or object of our attention?
sations among passengers tend to be consider-
ably less hazardous for drivers than cell phone Internal Versus External
conversations. Another argument is that the Attentional Focus
passenger can adjust the conversation (or stop
it altogether) in demanding driving situations; Views about how one can direct attention opti-
obviously, this is not possible for the person on mally during action have been around since at
the other end of the cell phone conversations. least the late 19th century (see "Cattell and Bern-
Also, some argue that the passenger can provide stein on Focus of Attention"), although empirical
comments that benefit safety ("Look out for the studies on the topic are relatively recent. How-
bicyclist!"), even during conversations with the ever, the more recent research in this area has
driver. shown that the suspicions of Cattell and Bernstein
were rather accurate-that experts benefit from
Handheld Versus Hands-Free Cell an "external" focus of attention, whereas less
Phones skilled performers benefit from an "internal"
Despite the preponderance of legislation that focus of attention.
bans handheld cell phone use during driving In most investigations, the definition of "inter-
but permits the use of hands-free units, almost nal" involves instructions to the subject to focus
all experimental studies on the topic have shown on some aspect of the motor skill, such as how
that both are about equally detrimental to driving a movement is being executed or the sensory
performance (Consiglio, Driscoll, Witte, & Berg, consequences of the movement. For an "exter-
2003; Lamble, Kauranen, Laakso, & Summala, nal" focus of attention, experimenters often ask
1999; Strayer & Johnston, 2001; Tornros & Bol- subjects to focus on an object in the environment
ling, 2005). In reviewing the evidence regard- or on some expected outcome of the action, such
ing distraction with the use of hands-free and as where a golf ball went after a putt (see Wulf,
handheld phones, Ishigami and Klein (2009) 2007, for a review). One obvious limitation of
conclude simply that "the epidemiological and this type of research is the lack of any hard evi-
the experimental studies show a similar pattern: dence that the subject has accurately followed the
talking on the phone while driving impairs driv- experimenter's instructions. Therefore, it is very
ing performance for both HH (handheld) and HF important that instructions be clear, concise, and
(hands-free) phones" (p. 163, emphasis added). repeated often to encourage subjects to focus their
Moreover, given the evidence that drivers tend attention as directed.
to compensate more for the potential distraction In studies by Beilock and her colleagues, golf-
of a handheld unit (e.g., by driving more slowly), ers in an internal-focus condition made putts
Ishigami and Klein argued that driving and talk- while concentrating on the motion of the putter
ing with a hands-free unit might even be more and saying the word "straight" aloud when the
dangerous than with a handheld phone because ball was contacted (Beilock, Bertenthal, McCoy, &
of the driver's mistaken belief that a conversa- Carr, 2004; Beilock, Carr, MacMahon, & Starkes,
tion with a hands-free device is not attention 2002). In another condition, the golfers performed
demanding. their putts while monitoring simultaneously in
order to detect when (or if) a tone was presented,
which presumably generated an external focus.
Focus of Attention The results were dramatic. Expert golfers showed
more skill in the external than the internal condi-
In this section we turn the discussion around tion, whereas the opposite was the case for novice
and ask, What is the effect on movement when golfers.
 Attention and Performance 125

CATTELL AND BERNSTEIN ON Focus OF ATTENTION

James Cattell's work and writing on attention predated and anticipated many of the research
issues that would later become prominent in mainstream experimental psychology (after the
demise of the behaviorist tradition). Bernstein's book (On Dexterity and Its Development) was
written in the 1940s in Russian. Discovered and translated into English many years later, this text
is written in a style geared toward a wide audience (Feigenberg & Latash, 1996) and contains
many practical examples and illustrations of motor control and learning in sport and daily life.
In the following quotes, Cattell and Bernstein address the issue of attentional focus. Note that
both advocate what is referred to today as an "external" focus of attention, but only for skills
that are well learned. In addition, their theoretical rationale, that focusing "internally" would be
disruptive to automatic movements, shares many similarities with current thinking.
"In the practiced automatic movements of daily life attention is directed to the sense impres-
sion and not to the movement. So, in piano playing, the beginner may attend to his fingers but
the practiced player attends only to the notes or to the melody. In speaking, writing and reading
aloud, and in games and manual work, attention is always directed to the goal, never to the
movement. In fact, as soon as attention is directed to the movement, this becomes less auto-
matic and less dependable" (Cattell, 1893).
"Consciously watching the movements of a teacher and intent attention toward one's own
movements make sense only at the beginning of the process of skill development, when the motor
composition of the skill is being defined. After the automations have already been elaborated
and switched out of consciousness, it is useless and even detrimental to chase them behind
the movement curtain. One needs to trust the level of the muscular-articular links; most of the
time, our confidence is justified.
"So, what should be in the focus of our attention during the final phases of skill development?
There is a very definite answer. The attention should be focused on a level where consciousness
resides and which takes the responsibility for movement success in its major and most impor-
tant components. Therefore, one should concentrate on the desire to solve a motor problem as
accurately and expediently as possible. This desire will lead to basic, meaningful corrections for
the whole movement. For example, the attention of a person who has learned to ride a bicycle
should be fixed not on his legs or arms but on the road in front of the bicycle; the attention of
a tennis player should be directed at the ball, the top edge of the net, the movements of the
opponent, but certainly not at his own legs or on the racket. Such concentration on the problem
maximally mobilizes the leading level with all its abilities" (Bernstein, 1996, p. 203).

Perkins-Ceccato, Passmore, and Lee (2003) had skilled golfers benefited from internal focusing
subjects perform pitch shots (in golf) under condi- instructions whereas the high-skilled golfers were
tions in which they were asked to focus either on more skillful in the external-focus conditions. Simi-
the force required to swing the club (internal focus) lar results have been obtained for soccer dribbling
or on the intended target (external focus). As an (Ford, Hodges, & Williams, 2005) and baseball
additional encouragement to focus as directed, batting tasks (Castenada & Gray, 2007; Gray, 2004).
after each shot the golfers were asked to make a However, Flegal and Anderson (2008) reported
performance estimate regarding the amount of the that an internal focus disrupted golf putting per-
force just produced (internal focus) or the accuracy formance only for higher-skilled golfers; the inter-
of the shot relative to the target (external focus). vention had no effect on the lower-skilled golfers'
Similar to what Beilock and colleagues (2002, 2004) performance. Overall, in most studies, an external
observed, high- and low-skilled golfers performed focus tends to enhance high-skilled golfers' per-
differently as a function of the instructed focus of formance, and some studies show that an internal
attention. As illustrated in figure 4.16, the low- focus benefits low-skilled golfers' performance.
126 Motor Control and Learning

For skilled performers, these findings suggest therefore, may have an interference effect on the
that directing attention toward production of the movement control processes that are relevant for
movement itself has a detrimental impact on advanced performers. In essence, when the expert
performance (although it may be equally correct "breaks into" the preprogrammed action with
to say that directing attention away from produc- consciously based information processing, aban-
ing the movement itself has a beneficial impact doning the elegant programs and routines she has
on performance-these influences have not been learned over many years, this may produce the
separated). Researchers agree that this effect jerky, slow, and hesitant actions characteristic of
is primarily related to attentional mechanisms a novice. These findings also add an important
that change as skill develops, although there is component to various considerations about how
disagreement about specific theoretical issues as to conduct practice sessions and to provide aug-
to how this happens. Beneficial effects of external mented feedback, as discussed further in chapters
focus of attention instructions have been demon- 11 and 12, respectively.
strated in a number of learning experiments by
Wulf and her colleagues, using an assortment Choking and Attentional Focus
of laboratory (e.g., stabilometer; see chapter 2)
Most of us have personal experiences with per-
and sport tasks (reviewed in Wulf, 2007; Wulf,
forming under pressure-situations in which we
McNiven, & Shea, 2001; Wulf & Prinz, 2001).
perceive a special expectation to perform well.
In many studies the subjects were novices at
The feeling of pressure may occur because of the
the beginning of the study and developed skill
importance of the performance (e.g., a concert
over several sessions of practice. As discussed
recital) or because of various other factors related
in the later chapters of this book, motor learning
to the event (e.g., presence of an audience; per-
is frequently thought to involve the cognitive
formance-outcome incentives, etc.). Sometimes
(or verbalizable) attention to movement during
the performer lives up to these expectations; but
the early stages of learning, and these processes
when performance fails to do so, the unfortunate
diminish or drop out completely as skill devel-
label of "choking" is often applied. Of interest in
ops. Directing attention back to the movement,
the present context is a common tendency for per-
formers to refocus their attention inward when
put in these situations, often leading to "choking
3000 under pressure."
In one of the first investigations conducted in
this area, Baumeister (1984) found a 25% drop in
2600
performance when subjects were approached at
a video arcade and asked simply to achieve their
~ 2200 best score possible in a one-trial performance test.
E
.£, Interestingly, these findings were attributed to the
...
e~ 1soo shift toward an "internal performance process"
<P (Baumeister, 1984), anticipating much of the
:c attentional focus literature to follow. "Reinvest-
·~ 1400
m
> ment theory" presents a similar view-motor
1000
skills that have been learned to an advanced level
are disrupted when individuals revert to a form
of conscious movement control (Masters, 1992;
600 Masters & Maxwell, 2008). Reinvestment theory
Internal External
has much in common with a view of learning
Focus of attention discussed in greater detail in chapter 13 called
"progression-regression" processes (Fuchs, 1962),
FIGURE 4.16 Effects of focus of attention in low- and
which theorizes that learners regress to previous
high-skilled golfers. level of control processes after a long retention
Adapted, by permission, from N.P. Perkins-Ceccato, S.R. Passmore, and T.D. Lee, interval or when faced with pressure situations.
2003, "Effects of focus of attention depend on golfers' skill," Journal of Sports
Sciences 21: 593-600, by permission of the publisher (Taylor & Francis Ltd,
Reinvestment may be seen as a reversion to an
http://www.informaworld.com). internal focus (Perkins-Ceccato et al., 2003) or
 Attention and Performance 1m 127

controlled-processing style of movement control mance related to arousal and anxiety and their
(Shiffrin & Schneider, 1977). Beilock and Carr :elationship to ideas about attention. Anxiety
(2001) proposed a view that has many similarities 1s common, as in an important game or match
to reinvestment theory, although they suggest or when we are threatened with harm in some
that choking in the performance of motor skills way. How do high-anxiety conditions affect the
may be caused by different mechanisms than in processing of information necessary for success-
the performance of cognitive skills (Beilock, 2008, ful performance?
2010; Beilock, Kulp, Holt, & Carr, 2004). Consider this true story. An airline pilot with
over 20,000 h of flying experience returned to
Ironic Effects of Attentional Focus San Francisco after an all-night flight from Hong
Another counterproductive effect of focused Kong. He drove home to Oakland (20 miles
attention is called the ironic effect. An anecdotal away), slept for a few hours, and then drove to
example of the effect is the golfer who stands at the local airport to check out his private plane
the tee, faced with out-of-bounds looming to the for a flight. With his family of three aboard, he
right. If the golfer mentally concentrates on the left for a destination a few hours away, watched
thought "Don't hit it to the right," the result that an automobile race there, and late at night began
seems to occur often is that the performer does the return flight. At about 2:00 a.m. he radioed
that his heater was not working and that he was
what is not desired. Wegner (1994, 2009) has made
a more thorough analysis of such situations. above a layer of clouds over the Oakland airport.
A pilot with his experience would, in this situa-
An experiment that illustrates ironic effects
tion, be expected to perform a relatively lengthy
was reported by Wegner, Ansfield, and Pilloff
(20 to 30 min) instrument approach through the
(1998) using a handheld pendulum. With the arm
clouds to the airfield, but instead he radioed
extended outward from the body, subjects held
that he was looking for a "hole" in the clouds,
a nylon line, at the end of which was a weighted
presumably to avoid the instrument approach.
object 2 cm above a glass plate. On the glass plate
The plane crashed a few minutes later, killing all
were grid lines with the x- and y-axes clearly vis-
aboard, after a wing broke off.
ible and with a green target in the middle. One
What happened? We might guess that the
(control) group of subjects was instructed to try
pilot was very fatigued from the overseas flight,
to hold the pendulum as steadily as possible over
from the two other flights that day, and from the
the green target for 30 s. The other (experimental)
car race; he was also cold and in a hurry to get
group was given the same instructions, but was
home. The fatigue and cold led to a bad decision
also told "You should not let it move in the direc-
to find a "hole." Also because of the fatigue, the
tion paralleling the horizontal line"-that is, in
pilot may have been handling the plane badly,
the left-to-right direction. The experimental group
perhaps becoming disoriented and diving too
produced more movements along the forbidden
steeply through the "hole," and the wing failed.
(horizontal) axis and slightly less movement along
Similar examples come from underwater diving, as
the ignored (vertical) axis than the control group
pointed out by Norman (1976) and Bachrach (1970):
(who were not given any "forbidden axis" instruc-
tions). These findings lend some empirical support A woman enrolled in a diving course but lacking
to the anecdotal feeling that trying to avoid a par- experience, was reported to have . . . drowned
ticular behavior paradoxically leads to that very while diving for golf balls in a twelve-foot [pool].
When her body was recovered, she was wearing her
behavior. Sport psychologists often spend many
weight belt, and, in addition, was still clutching a
hours with their clients to practice positive imag- heavy bag of golf balls. (p. 122)
ery techniques, perhaps with the goal of avoiding
the kinds of ironic behaviors noted by Wegner and Again, something went wrong. Perhaps fatigue
his colleagues (Janelle, 1999; Singer, 2000). and cold led to panic. What effects do anxiety and
panic have on information processing in these
situations, causing people to abandon highly
Attention and Anxiety practiced techniques and resort to the skill level
of an inexperienced beginner? We consider some
In this final section of the chapter we shift empha- possible answers to these questions in the next
sis to some important aspects of human perfor- few sections.
128 Motor Control and Learning

Arousal and anxiety, along with related terms

such as motivation and stress, generally refer to Good
states of activation that differ along a number
of dimensions. Someone who is "psyched" or
"pumped up," or otherwise in a highly energized
state of activation, may be referred to as highly
aroused or in a state of high anxiety. How does ~
c:
ca
arousal or anxiety, or the combination, influence ...E
0
performance? This relationship has been char-
acterized in a number of ways, each with some
ta.
supporting evidence. We review each of these
characterizations briefly in the next sections
(more in-depth analyses are presented in Gould &
Krane, 1992; Weinberg & Gould, 2011; Woodman Poor
& Hardy, 2001; Zaichkowsky & Baltzell, 2001).
Low Moderate High
The lnverted-U Principle Arousal level

One of the oldest and certainly one of the most

interesting aspects of arousal and performance FIGURE 4.17 The inverted-U relation between arousal
was discovered by Yerkes and Dodson (1908) in and performance.
studying discrimination learning in mice. They
found that increased intensity of electric shocks
males, were also told that this task was an
delivered to the mice reduced the number of trials
important indicator of success in most sport
required to acquire the discrimination criterion,
tasks involving throwing (which is not actually
but only up to a point. Beyond this point, further
correct, as we shall see in chapter 9), such (false)
increases in the intensity of the shock increased
the number of trials to reach criterion. If the statements about their level of performance were
shock was arousing, then there appeared to be an expected to be variably arousing. Indeed, arousal
optimal level of arousal; that is, the relationship data indicated that this was the case.
between shock intensity and performance fol- The throwing scores for the subjects in these
lowed an inverted U-shaped function. The prin- three stress conditions are illustrated in figure
ciple has been called the Yerkes-Dodson law, after 4.18. Increased arousal from the low-stress to the
its originators or, more commonly, the inverted-U moderate-stress condition produced strong gains
principle (Duffy, 1962). A general visual descrip- in performance; but further increases in arousal
tion of this principle is presented in figure 4.17. produced a decrease in throwing proficiency
Weinberg and Ragan (1978; see also Klavora, relative to that in the moderate-stress condi-
1977; Martens & Landers, 1970; Sonstroem & tion. Consistent with the inverted-U principle
Bernardo, 1982) have provided evidence for the (compare figures 4.17 and 4.18), there was a clear
inverted-U phenomenon in movement behavior. optimal level of arousal for the performance of
They asked college males to throw tennis balls at this motor task.
a 5 cm target located 6.1 m away. After 10 trials The inverted-U principle suggests that there
of initial practice, subjects were asked to fill out is no simple relation between the level of stress
a questionnaire, and they then received one of and the quality of performance. How often do
three statements. In the high-stress condition, coaches and sportscasters suggest that an athletic
the subject was told that 90% of the population performance was very good because the athlete or
of college males would have performed better team was "up" for the game (see Nideffer, 1976)?
than he did. In the moderate-stress condition, he The common, everyday belief seems to be that the
heard that 60% of the population would have had more "up" we are, the better we perform, which
better performances than his. In the low-stress is not usually the case. Although the inverted-U
condition, the subject was told that only 30% of principle has some explanatory power, the rela-
the population was better at throwing. Because tion between arousal, anxiety, and performance
the subjects, who were junior high school-aged is much more complex than this. 3
 Attention and Performance !!Ii 129

Considerable evidence is available to support

11 the notion of perceptual narrowing in a variety
of tasks (see Kahneman, 1973). In an interesting
practical setting, Weltman and Egstrom (1966)
studied novice scuba divers in air on a pool deck,
10 in a controlled swimming pool, and in the ocean.
CD
.... The divers performed two main tasks (arithmetic
0
()
IA and a dial-detection task). Responding to a light
Cl
·;:c: stimulus presented in peripheral vision in the
9 diving masks was measured with a secondary
-
0....
.c
c:
task. As the subjects were moved from the air to
Ill
CD the pool to the ocean conditions, large (300% to
::?:
400%) increases in time to detect these peripheral
8
signals were found; the performance in response
to light stimuli presented in central vision was
hardly affected. The focus of attention was clearly
narrowed with increased stress. Janelle, Singer,
Low Moderate High and Williams (1999) reported similar findings for
Arousal
a simulated driving task.
FIGURE 4.18 Throwing proficiency as a function of three Easterbrook's Cue-Utilization
stress conditions.
Adapted from Weinberg and Ragan 1978.
Hypothesis
Easterbrook's (1959) hypothesis uses a notion
similar to perceptual narrowing to account for
Perceptual Narrowing the inverted-U relation. This viewpoint assumes
The idea of perceptual narrowing has been that the individual takes in cues, from the envi-
described by Kahneman (1973) and others. With ronment or from his own movements, that sup-
this notion, increased arousal causes narrow- port future performances. There are presumably
ing of the attentional focus, with a progressive many relevant and irrelevant cues available in the
elimination of input from the more peripheral (or environment, and the most relevant for perfor-
unexpected) aspects of the environment. The term mance are located in central vision (where we are
peripheral need not refer to events that are actually looking). With low arousal, the selectivity for the
"in" the periphery (e.g., as in peripheral vision); it cues is poor, as the visual field is wide and many
refers to events that are relatively improbable. It relevant, and irrelevant, cues can be detected to
just so happens that, with vision, events that are generate action. With an increase in arousal to
expected are usually in our central vision (because moderate levels comes a reduction in the number
we direct our gaze toward them), and events that of cues used (because of the narrowing of focus),
are improbable occur in peripheral vision. , so that there is a shift to an area where relevant
In addition to the reduced range of cues that cues are more prevalent and irrelevant cues are
can be attended to as arousal increases, there is less prevalent. As the performer approaches an
an increase in the number of shifts in attention optimal level of arousal, the most effective combi-
to different input sources. Some researchers (e.g., nation of attention to relevant cues and minimal
Kahneman, 1973) have referred to this effect as inclusion of irrelevant cues occurs. However, with
increased distractibility. Thus, high levels of arousal a further increase in the arousal level comes a
are likely to cause the individual to direct attention further restriction in the range of cues used; many
to many different sources from moment to moment, relevant cues are now not included, and perfor-
with some of these sources providing irrelevant mance deteriorates. This may have occurred with
information and causing the relevant signals to be both the pilot and scuba diver discussed earlier.
missed. Apparently, the individual must discrimi-
nate between relevant and irrelevant cues during
Hypervigilance, or Panic
performance, and one effect of high arousal is When the stress conditions become more severe,
deterioration in the quality of such discrimination. the responses of human performers can be even
130 Motor Control and Learning

more severely disrupted. Many of these disrupted of capacity, each specialized for separate kinds of
responses occur in true emergency situations, processing. The successful performance of vari-
such as earthquakes, fires, battlefields, and other ous tasks together is presumably limited by these
life-threatening situations. As such, they are dif- capacities; therefore, attention demand is usually
ficult to study directly, and one must usually estimated indirectly by the extent to which tasks
rely on various kinds of postevent measures to interfere with each other. Processing of sensory
understand them. stimuli (or performing other processes early in
Such situations often stem from what Janis, the sequence) can apparently occur in parallel,
Defares, and Grossman (1983) call hypervigilance with little interference from other tasks. But
or what one might loosely term "panic." The con- processes associated with response selection or
tributing conditions seem to be (a) a sudden and with response programming or initiation interfere
intense stimulus, (b) a potentially life-threatening greatly with other activities.
outcome, and (c) a situation in which time to take Characteristics of attention play an impor-
an appropriate action is quickly running out. tant role in motor performance. Psychological
All of these seem to be present in most cases of refractoriness-the delay in responding to the
a kind of vehicle accident termed "unintended second of two closely spaced stimuli-provides
acceleration" (Schmidt, 1989b, 1993; see chapter evidence that some single channel, or bottleneck,
6, p. 203, "Pedal Misapplication Errors"), in which in processing exists in the response-selection or
the driver panics when she mistakenly presses response-programming stage, before which pro-
the accelerator rather than the brake, and the car cessing is parallel and after which processing is
accelerates unexpectedly (and with no intention serial. Other limitations in attention, such as the
on the driver's part). The individual appears Stroop effect, inattention blindness, and inhibi-
to freeze, perhaps because attention is so nar- tion of return, reveal that attending selectively to
rowed that no alternative actions are considered. some information may have consequences that
It could also be that the driver considers many are not intentional. Other evidence, based on
alternatives but, because of the increased distract- secondary-task techniques, suggests that atten-
ibility, none long enough to initiate a solution; tion demands are highest at both the initiation
she "dithers" until it is too late. Also, the person and termination stages of movements, particu-
tends to choose the action that is highest in the larly when the end of the action has an important
response hierarchy-the most common, "natu- precision component. The use of cell phones and
ral," or highly practiced action (e.g., pressing the other in-vehicle devices during driving provides
brake when the vehicle spins on ice). Moreover, an excellent example of attentional limits on daily
the individual tends to persevere, producing the activities. Some evidence suggests that directing
same response repeatedly, even though this action one's attention to movement or environmental
has not had the desired effect. These situations cues may differ according to one's skill level.
are not as rare as they might appear, as evidenced Performance is influenced by arousal, and sev-
by analyses of numerous vehicle accidents (Perel, eral hypotheses have been suggested to explain
1976). Further, they might also occur in highly this relationship. The mechanisms that appear
stressful sport events, industrial tasks, theater or to limit performance under stress are related to
night club fires, and so forth, with outcomes that the decrease in cue utilization and the failure of
are much less dramatic or not so easily analyzed. decision-making processes, especially under the
most stressful conditions called hypervigilance,
or panic.
Summary
Even though attention has had a long history Student Assignments
of thought in psychology, we are still unclear
about its nature and the principles of its opera- 1. Prepare to answer the following questions
tion-indeed, even its definition. Many theorists in class discussion:
think of attention as a single, undifferentiated, a. Using examples from a sport, illustrate
limited capacity to process information; others three different aspects of the concept
argue that attention is really a number of pools of attention.
 Attention and Performance 131

b. Provide examples of inattention blind- Notes

ness in driving.
c. Find two other types of Stroop effects 1 Actually, the basic phenomena underlying the

that have been studied in the research "Stroop effect" had been studied for about a half century
literature. before Stroop published his work in 1935, beginning with
Cattell (1886; see MacLeod, 1991, for a historical review).
2. Find a research article not covered in the
2 In a way, this label is unfortunate. The original idea
book on the effects of talking on a cellular was that the delay in the subject's response to the second
phone during driving. of two closely spaced stimuli is analogous to the delay
when a single nerve fiber is stimulated electrically twice
in rapid succession in physiological experiments. If the
Web Resources second stimulus is very close to the first (within about
5 ms), no response at all will result from the second
Stroop effect demonstration: stimulus. This effect has been termed the absolute refractory
period, meaning that the nerve is insensitive to additional
http://faculty.washington.edu/chudler/words. stimulation while it is refractory, and it "recovers" from
html the effects of the first stimulus. These neuronal processes
probably have little to do with the psychological refrac-
Original paper by J. Ridley Stroop: tory period, as the time course is much longer in the
behavioral work (e.g., 200 ms).
http://psychclassics.yorku.ca/Stroop/ 3 Note that even the original findings of Yerkes and

Dodson (1908) were actually more complex than is tradi-

Test your ability to multitask:
tionally depicted by this inverted-U function. They indeed
www.nytimes.com/interactive/2009/07119/ found an inverted-U relationship between increases in
technology/20090719-driving-game.html shock intensity and the number of trials required to reach
criterion discrimination performance. However, this was
Inattention blindness: true only for moderate and degraded lighting conditions.
When full lighting was provided, making the discrimina-
www.dothetest.co. uk/basketball.html tion relatively "easy" (i.e., choosing between a white and
black door), there was no inverted-U relationship. Rather,
Video showing a danger of texting while walking: the relationship was linear-the number of trials to reach
criterion decreased as shock intensity increased. Thus,
www.tinyurl.com/videodangers2 even Yerkes and Dodson's data were more complex than
this "simple" inverted-U relationship.
CHAPTER 5 Sensory Contributions to Motor Control
CHAPTER 6 Central Contributions to Motor Control
CHAPTER 7 Principles of Speed and Accuracy
CHAPTER 8 Coordination
CHAPTER 9 Individual Differences and Capabilities

he human motor system is a very complex whole with many interacting pieces,
processes, and mechanisms. To attempt to understand the entire system as a whole
would be extremely difficult. For this reason, scientists generally study various parts
of the system in isolation. These are often studied as independent modes of control-
fundamentally different ways in which the system's parts can work together. In chapter
5, the focus is on the role of sensory information; we consider the ways in which infor-
mation from the environment influences, or even determines, movement behavior. In
chapter 6, we examine the central control and representation of action in situations in
which sensory influences do not have a particularly strong role. Chapter 7 deals with
various laws and models regarding the principles of speed and accuracy, especially as
they pertain to the control of relatively simple" movements. This analysis is extended
/1

in chapter 8 to more complex tasks involving the coordination of more than one effec-
tor. Part II ends with a discussion in chapter 9 of factors that tend to make individuals
differ from each other in their skilled behaviors.

133
 CHAPTER 5

ne of the ways in which motor control is ways analogous to mechanical systems' control.
achieved relies heavily on the concept that Closed-loop systems are important in many situ-
we use sensory (or afferent) information to regulate ations, especially those that require a system to
our movements. This can be information that tells "control itself" for long periods of time (in con-
us about the state of the environment, about the trast, see chapter 6 for a discussion of open-loop
state of our own body, or about the state of our systems).Adiagram that illustrates how a simple
body with respect to the environment. A way to closed-loop system works is shown in figure 5.1.
think about how sensory information is used in the First, input about the system's goal is provided to
control of action is to consider the moving human a reference mechanism. In a home heating system,
as a type of closed-loop system. A closed-loop system the overall goal might be to achieve and maintain
depends heavily on the involvement of particular a certain temperature in the house. The refer-
types of sensory information as it executes its func- ence mechanism in this system is the specific
tion. Such sensory information, when discussed in temperature setting, say 68° F-the temperature
the context of closed-loop motor control, is often setting can be changed without affecting how the
termed movement-produced feedback, or simply feed- system operates. Next, the reference mechanism
back, implying that the sensory information to be samples the environment that it is attempting to
considered has arisen as the result of performed control to determine the current temperature.
actions. Of course there are many other forms of This information from the environment is usually
sensory information that are not associated with termed feedback. The reference mechanism then
the performed movements, and these are usually compares the value of the goal (e.g., 68°) to that
considered under the more general heading of sen- of the sample obtained from the environment
sation and perception. In this chapter we discuss (current temperature), and an error is computed,
the various kinds of sensory information that can representing the difference between the actual
be used in the control of movement. and desired states. The error is information that
is passed to an executive level, where decisions are
made about how to reduce the error. If the error
Closed-Loop Control Systems is large enough, instructions are sent from the
executive level to the effector level, and a mecha-
One way of attempting to understand motor con- nism that has some effect on the environment
trol has been to consider sensory contributions in is activated-in this case, the heater. The heater

135
136 Motor Control and Learning

System goal usually thought of as the muscles or as a program

of action that actually controls the muscles. Then,
the information obtained from the various muscle,
joint, and tendon receptors, as well as from the
eyes, ears, and so on, is sent back to the reference
mechanisms for analysis, and decisions about
future action are again made. Notice that, with this
feedback loop (and the stages of processing), the
implication is that all feedback is processed in this
way-requiring attention (and consciousness).
As we will learn later in this chapter, some (if not
most) feedback for action is processed reflexively,
"below" the level of consciousness (see Ghez &
Krakauer, 2000; Pearson & Gordon, 2000b ).
Feedback The various sources (or receptors) of sensory
information that are available during movement
are traditionally classified into three groups (after
Sherrington, 1906). Perhaps the least important
for our purposes is the class of receptors called
interoceptors. They tell us about the states of our
internal organs, as with "hunger pains," and have
questionable relevance for motor behavior. The
remaining two classes of receptors are divided
according to whether they provide information
about the movement of objects in the environ-
ment (exteroceptors) or information about our own
movements (proprioceptors). The roots extero and
FIGURE 5.1 Elements of the typical closed-loop con- proprio refer to events outside one's body and
trol system. events in one's own body, respectively.

raises the temperature of the room, and this raised Vision

temperature will be sampled periodically. When
the difference between the actual and desired Certainly the most critical receptor for supplying
temperatures is zero, the executive level shuts information about the movement of objects in
off the heater. In such a way, the heater cycles the outside world is the eye. Therefore, we begin
between being on and off as a function of the our discussion with the richest and most impor-
actual room temperature and thus maintains the tant feedback information used in the control of
system goal. Such systems are termed closed loop movement.
because the loop of control from the environment
to decisions to action and back to the environment How Vision Provides Information
again is completed, or closed.
How does this model of the closed-loop system Subjectively, we all know that darkness critically
relate to the human motor control system? For impairs the performance of many tasks, although
now, think of the reference of correctness and the we also know that vision is not essential for all
executive level as being contained in the stages motor performances. People who are blind learn
of information processing, so that the system can to move in the environment with remarkable
receive and process feedback information follow- facility, although they are at a large disadvantage
ing the stimulus-identification stage, and then in many situations. In terms of human perfor-
program instructions to the musculature to reduce mance, vision provides information about the
the error in the response-programming stage. movements of objects in the environment, as
Instructions are then given to the effector level, well as about the movement of one's body in that
 Sensory Contributions to Motor Control Iii 137

environment, and much of motor control involves from the environment by means of the retina
tailoring our motor behavior to meet these visu- to the primary visual cortex in the brain. From
ally presented environmental demands. there, however, the dorsal stream sends infor-
mation to the posterior parietal cortex, whereas
Dorsal and Ventral Visual Streams
the ventral stream projects to the inferotemporal
Concerning vision as an exteroceptive system (for cortex (Ungerleider & Mishkin, 1982). Of particu-
detecting events in the outside world), we have lar interest to researchers in motor behavior are
long known that central vision (involving visual the roles afforded by the two streams. Whereas
acuity, contrast, object identification, and con- ventral stream processing has been considered
sciousness) is a major sensory component of many responsible for providing cognitive information
types of motor behavior. However, evidence and about objects in the environment (e.g., object
theorizing argue for the existence of two distinct identification), processing by the dorsal stream
streams of visual processing. The lines of evidence is proposed to provide information specifically
to support these different streams come from a for the visual control of movement (Goodale &
wide range of levels of analysis, using various Milner, 1992; Goodale & Humphrey, 2001; Milner
techniques and subjects in the research. Indeed, & Goodale, 1993, 1995). Thus, a generalization
even the names for the two visual streams differ that has been the topic of hot debate is that
markedly, with such proposed dichotomies as cog- ventral stream processing is vision for perception
nitive versus motor vision, explicit versus implicit (identification) whereas dorsal stream processing
vision, object versus spatial vision, overt versus is vision for action Geannerod, 1997; Jeannerod &
covert vision (Bridgeman, 1996), and focal versus Jacob, 2005; Milner & Goodale, 2008).
ambient vision (Trevarthen, 1968). We will refer to The nature of the information processed
the two visual streams in terms of the anatomical appears to be a key component of the distinc-
distinction proposed by Ungerleider and Mish- tion. For the ventral stream, information that
kin (1982)-the dorsal and ventral visual stream is central to the identification and recognition
dichotomy (e.g., see the review and commentaries of objects is crucial-this includes information
in Norman, 2002; also Creem & Proffitt, 2001). picked up from the environm~nt and information
In one of these systems (the ventral system), stored in memory. In contrast, the dorsal stream
the primary input is limited to central vision. is particularly attuned to information about how
This information requires contrast, focus, and suf- we control our motor system to interact with an
ficient light. This system is specialized for object object. As Norman (2002) pointed out, however,
identification and (conscious) perception of the it is important to keep in mind that a complemen-
environment. Thus, the ventral system is sensitive tary relationship exists between visual processing
to what the person is looking at or focusing on in the two streams. For example, when looking
(or both). In the other system (the dorsal system), for a hammer we use ventral stream informa-
the visual input is full field (nearly 180°), does not tion to recognize the hammer and dorsal stream
require focus, and seems to operate effectively information to pick it up (Norman, 2002). In the
even in very low-light situations. We can think next sections, we review some of the evidence that
of .the retina of the eye as the recipient of rays of supports the existence of two streams of visual
light from all parts of the field of vision. For any information processing. 1
given position of the eye, a unique combination of
these rays allows specification of the eye in space. Evidence for Two Visual Streams
More importantly, when the eye moves through Classifying things is easy. Before a classification
space (as during walking), the angles of these is accepted, however, scientists usually demand
rays change in predictable ways. These changes strong evidence that it facilitates understanding.
in the light rays-termed "optical flow" because One of the strongest kinds of evidence is a disso-
the angles of these rays change continuously, or ciation, in which the effect of some independent
"flow"-specify the nature of the eye's movement variable on one hypothesized construct (e.g.,
(and thus the person's movement) in space. We'll ventral stream) is different than the effect of the
come back to optical flow and movement shortly. same variable on the other construct (e.g., dorsal
Anatomically, both the dorsal and ventral stream). Support for the separation of hypothe-
streams are thought to project visual information sized constructs is strong when clear dissociations
138 m11 Motor Control and Learning

are found. In the study of distinct streams of the target's position but did not bias the visual
visual processing, dissociations have been found information used for action. Another experiment
in two types of research investigations. used the reverse procedures: now, the target actu-
Perception-Action Dissociations in Brain- ally moved, and the background-frame moved in
Injured Patients Strong evidence that favors the opposite direction by such an amount that the
the dissociation between perception vision and subject perceived (consciously, but erroneously)
action vision has been found in patients who have that the target was stationary. When the target
specific damage to parts of the brain involved and background were extinguished, the pointing
in the processing of visual inputs. For example, tended to be to the actual location of the target,
patients who had a type of brain injury resulting not to the position that the subject perceived
in what is called optic ataxia were able to recognize consciously.
an object, but were unable to use this same visual This effect is actually a double dissociation: (1)
information to guide their hand accurately to the The moving background (or frame) altered the
object (Perenin & Vighetto, 1988). In contrast, conscious perception of the fixed target position
patients with a type of disorder called visual but not the pointing direction, and (2) the altered
agnosia were unable to recognize common objects, actual position of the target (whose location was
yet could use the visual information to grasp perceived as fixed because of the moving back-
the objects accurately (Goodale & Milner, 1992; ground) did not bias the pointing direction. Simi-
Goodale, Milner, Jakobson, & Carey, 1991). This lar double dissociations have also been shown
dissociation in the ability of patients to use vision for static images in which an off-centered frame
for one task but not the other represents a very biases the perceptual judgment of a particular
stimulus target but does not bias a subject's reach
important finding in support of visual stream
toward the target (Bridgeman, Gemmer, Forsman,
separation. Researchers have also relied both on
& Huemer, 2000).
experiments with animals (e.g., Schneider, 1969;
These important findings, in which a subject-
Trevarthen, 1968; Ungerleider & Mishkin, 1982)
viewed stimulus resulted in a perception vision
and on behavioral experiments involving healthy
bias but not an action vision bias, have been the
individuals to further study these visual streams.
basis for a large number of empirical investiga-
Perception-Action Dissociations in Behavioral tions involving visual illusions. Many of these
Studies An important early study in the behav- studies used a similar experimental paradigm
ioral identification of the two visual streams was in which subjects were asked to make perceptual
conducted by Bridgeman, Kirch, and Sperling judgments and actions involving various visual
(1981). They used an induced-motion paradigm, illusions as targets. Figure 5.2 illustrates four of
in which small movements of a surrounding these visual illusions.
background (a "frame") in a given direction In investigations of the Ebbinghaus-Titchener
made it appear that a fixed target element within illusion2 (figure 5.2d), subjects are asked to point
it moved in the opposite direction. Under these to a circle in the center of surrounding circles (a
conditions, subjects would say (i.e., they per- two-dimensional version of the task), or to grasp
ceived, consciously) that the (actually stationary) a cylindrical disc that is located in the middle of
target moved back and forth through about 2.5° surrounding discs (a three-dimensional version).
of visual angle. This is a measure of the bias in In this illusion, the Ebbinghaus-Titchener display
the ventral processing system. Bridgeman and creates a bias to perceptual vision that makes the
colleagues then suddenly extinguished the target center circle surrounded by a ring of small circles
and background, and the subject's task was to appear larger than the center circle surrounded by
move the hand to point to the last location of the a ring of large circles, even though the two center
target, providing a measure of the dorsal system's circles are identical in size. Most studies, however,
accuracy. These pointing movements (to the actual have shown a dissociation of the effect of the
position of the target) were largely unaffected illusion on perceptual versus action vision-the
by the movement of the background. In this illusion has no influence on action vision, evalu-
way, Bridgeman and coworkers found that the ated either as the movement time to point at the
movement of the background biased the visual circles or as the size of the grasp used to pick up
information used for (conscious) perception of the center object; sometimes, the bias is much
 Sensory Contributions to Motor Control 139

smaller than the effect on perceptual judgments Note, however, that much controversy exists
of these circles (Aglioti, DeSouza, & Goodale, regarding the interpretation of visual illusion
1995; Haffenden & Goodale, 1998; Haffenden, dissociations (e.g., Bruno, 2001; Carey, 2001).
Schiff, & Goodale, 2001). Similar perception Some of the behavioral dissociations are found
vision and action vision dissociations using the only under restricted and specialized experi-
other visual illusions have also been found for the mental conditions (e.g., Fischer, 2001; Franz,
Miiller-Lyer illusion (figure 5.2a) (Otto-de Haart, Biilthoff, & Fahle, 2003; van Donkelaar, 1999;
Carey, & Milne, 1999), the Ponzo illusion (figure Westwood, Dubrowski, Carnahan, & Roy, 2000;
5.2b) (Bartlett & Darling, 2002; Ellis, Flanagan, & Wraga, Creem, & Proffitt, 2000), and sometimes
Lederman, 1999; Ganel, Tanzer, & Goodale, 2008), are not found at all (e.g., Meegan et al., 2004).
and the horizontal-vertical illusion (figure 5.2c) Comprehensive reviews of this research area
(Servos, Carnahan, & Fedwick, 2000). conclude that visual illusions appear to have a

FIGURE 5.2 Various visual illusions used in aiming tasks. (a) Muller-Lyer illusion, (b) Ponzo illusion, (c) horizontal-verti-
cal illusion, (d) Ebbinghaus-Titchener illusion.
140 Motor Control and Learning

larger influence on perception than on action, information that we drive past (such as road
although both appear to be influenced in similar signs); the car that we are passing; the car that is
ways by the visual information (Bruno, Bernardis, coming rapidly toward us in the opposite lane;
& Gentilucci, 2008 ; Bruno & Franz, 2009). Con- and the information within our own car, which is
sequently, while some researchers feel that these traveling at the same speed as ourselves. All these
behavioral data support the distinction between sources of information represent visual cues that
separate streams of visual processing, others are we must perceive accurately in order to execute
less convinced. For example, Glover (2002, 2004; a safe pass.
Glover & Dixon, 2001) presented evidence that Considerable research in motor behavior has
a perceptual tilt illusion affected action during been directed at a subset of skills involved in the
the movement planning stage but not during the perception of dynamic visual information. Many
execution of the movement, leading to the view of these skills are used in ball games and require
that differences in the use of action vision are interceptive actions such as catching or striking.
dissociated between the planning and execution The moving object provides various types of
stages. In contrast, Smeets and Brenner (2001; information, such as its position in space and the
Smeets, Brenner, de Grave, & Cuijpers, 2002) way its position is changing over time (e.g., in the
argue that object orientation in the environment case of a pitched ball in baseball). However, prob-
affects visual processing strategies, and does not ably the most important and frequently studied
imply separate streams of vision. It seems that skill relates to the use of temporal information
further experimental evidence will be needed about when the object will arrive. This is called
before we have a clear understanding of what time-to-contact (T) information.
these behavioral studies suggest regarding ven- An important view regarding how Tc is per-
tral and dorsal visual stream processing. ceived was formulated by David Lee (1980,
1990; Lee & Young, 1985; see also Purdy, 1958),
Visual Information About Time and relates quite closely to the concept of visual
to Contact proprioception and to the work of James Gibson
Another important issue related to the general (1966, 1979; see "Visual Proprioception" on p.
topic of how visual information is provided has 142). Consider an object, such as a ball, that is
to do with the processing of dynamic informa- moving directly toward your eye, as shown in
tion-visual information that changes over time, figure 5.3. If the angles of the light rays from the
sometimes at a very rapid rate. For example, in edges of the ball (i.e., a 1 and a 2) are increasing
passing another automobile, we are bombarded at the same rate (with respect to straight ahead),
with various sources of visual inputs that change this information specifies that the ball is moving
at different rates: the stationary environmental directly toward your eye. Such information,

Eye

Motion

FIGURE 5.3 Diagram of a ball approaching an eye, showing the size of the image (A) it projects on the retina; at any
moment, the time until contact is proportional to tau, which is directly proportional to the retinal size A divided by A, the rate
of change of A (i.e., tau= k A I A).
 Sensory Contributions to Motor Control 111 141

called looming, usually elicits strong avoidance concept that a relatively simple transforma-
reactions that appear to be nearly "automatic," tion of optical flow allows the specification of
or at least difficult to inhibit, especially in infants. important temporal events in the environment.
But in addition, Lee's work shows that these It is a relatively simple theory in the sense that
changes in visual information tell the viewer perception does not depend on the "computa-
about the amount of time remaining until con- tion" of either distance or velocity cues per se. In
tact will be made (Lee, 1976, 1980, 1990, 2009; a later section we will see how ideas regarding
Lee & Young, 1985). In figure 5.3, the edges of Tc have been used to explain various features of
the ball cast a retinal image of size A, which motor control.
increases in size as the ball comes closer; also,
the more rapidly the ball approaches, the faster Time Required to Process Visual
the image will be expanding. Let's abbreviate Feedback
this rate of expansion occurring at some moment Consider the following situation: Someone is
(or the derivative of A with respect to time) as A, working at night in a toolshed, striking the head
where the dot above the letter means the "rate of a nail with a hammer. Now imagine that the
of change of." Lee defines "proportionate rate electrical power to the shed is lost and all the
of expansion" as simply the rate of expansion lights go out just as the hammer starts its down-
A divided by the size of the retinal image A (or ward swing to contact the nail. How does the
A!A), which is a measure of the rate of change at sudden loss of vision affect the accuracy? Would
some moment as a proportion of the size of the accuracy have been affected any more or any less
retinal image A. Lee (1976) showed mathemati- if the hammer had not yet started its downward
cally that the reciprocal of this ratio, or A/A, is swing, or if the size of the nail had been larger or
directly proportional to Tc. This is true regardless smaller, or if the swing had been faster or slower?
of the size of the object, the velocity at which it For over a century, experimenters have studied
travels, or its distance from the eye. That is, time research questions addressing situations such as
to contact (or T) is defined as these using various theoretical and experimental
T, =tau (or 'T) = k A IA (5.1) approaches.
where k is a proportionality (slope) constant. In Moving to Stationary Targets
other words, Tc is proportional to the reciprocal An initial estimate of the time required to pro-
of the proportionate rate of expansion of the cess visual feedback was provided more than a
retinal image. This makes common sense. At a century ago by Woodworth (1899). The research
given distance, an object has a retinal size of A; was motivated by his view that aiming move-
the rate of change in the object's retinal image (A) ments consist of a two-component process. In the
will be larger the faster it is traveling. This makes initial-impulse phase (also termed the ballistic- or
the denominator in equation 5.1 larger and hence programmed-movement pha$e by later researchers),
tau smaller, indicating that there will be less time Woodworth suggested that movements are initi-
until the object contacts the eye. ated by an impulse that drives the limb toward
More recently, Lee (1998) formulated a gen- the target. Once the movement is under way, the
eral theory of how tau is used in the control of second phase provides an opportunity to correct
movement. In this view, all goal-directed actions any spatial error that may have occurred in the
require the closing of a gap. For example, running intended path of the limb. Woodworth called
to catch a ball requires the closing of two gaps: this second component the current-control phase,
(1) the gap between the ball's initial position and during which the subject uses visual feedback
its point of interception and (2) the gap between to home in on the target, making fine corrective
the catcher's initial position and the point of adjustments as needed in order to land on the
interception. Lee has suggested that tau coupling target. According to Woodworth, the successful
is one component of a general tau theory-that completion of this second phase of the movement
the closure of the two gaps occurs at the same can occur only if there is sufficient time to process
relative rate (Lee, Georgopolous, Clark, Craig, & the error arising from the analysis of the visual
Port, 2001). This view follows closely the earlier feedback and to issue an efferent signal that will
theoretical formulations and retains the elegant alter the spatial course of the movement. Thus,
 ... r~p~~~h~~ch~r1gesinheadposition contrib-
··ef~G'~bsol"I, 1966,J979). The pattern of rays
•• µ~lqlle specification of the location of the
·t~~· eye is• moved from one place to another
Jro~~en["flbwspastus" as we move around .
•. w•specify distirlctkirlds of movements of the
...
e*1 1f•~he~11g1~bet\ll/een .the light rays from two
.difr~S:~flat;yc!.)oare nbtmoving wi.th respect to that
.)iqgKtheil yo1,1 ·are moving toward the object; if it
~\s~·;itthe.~pg!es.fromtwo.sides of an object (with
·. e ·sarrwr:r~te;fhe eye is.moving toward the center of
·:t~e;ci~~tel'p.f~j!>icture on the wall). In conditions
. tare\cbqngiQg in the same direction, if the rate of
·.·.. ··~id~of:tf!eobjectis grea~erthanthe rate of increase
'·;.y6~"\l\iillpa9s the object .so that it is on your right

,i·~~:e~J{fg~~entalso tells.you about the environment

•
j/Vete:statiol"}aryJ=or example, imagine looking out the

.•. } r·.:;··::~~~ely providing information about the

• h~.cabout our own movements. As well,
1t~~li~n$·tC>finforrning µs about the environment.
....... ~ .a~d reciprocally linked. Excellent discus-
,.j9/9) ari.d l:.ee (1980~ Lee & Young, 1985),
~. includin€J sporHelated motions and bird

142 II!
 Sensory Contributions to Motor Control 143

R~S. WOODWORTH ON MANUAL AIMf!iG

R.S. Woodworth became one of the most noted American experimental psychologists of his time.
The topic of hi~ doctoral dissertation.was motor control, and in 1899 he published the results of
a large number of experiments on the topic of "accuracy of voluntary movement." In this work,
Woodworth touches on a wide array of topics related to the control of movement: the use of
vision, handedness, biomechanical factors, fatigue, practice, and so on. One of his most lasting
contributions has been the two-component model of movement aiming (i.e., initial adjustment
vs. current control), and Woodworth's ideas remain at the forefront of theorizing today (Elliott et
al., 2010; Elliott, Helsen, & Chua, 2001; Newell & Vaillancourt, 2001 b). In the following quote from
Woodworth's dissertation (Woodworth, 1899), note how effortlessly he moves between topics
that would later become the center of much of our present understanding of these processes.
"If the reader desires a demonstration of the existence of the 'later adjustments' which consti-
tute the most evident part of the 'current control; let him watch the movements made in bringing
the point of his pencil to rest on a certain dot. He will notice that after the bulk of the movement
has brought the pencil near its goal, little extra movements are added, serving to bring the
point to its mark with any required degree of accuracy. Probably the bulk of the movement is
made by the arm as a whole, with little additions by the fingers. If now the reader will decrease
the time allowed for the whole movement, he will find it more difficult, and finally impossible, to
make the little additions. Rapid movements have to be made as wholes. If similar movements
are made with eyes closed, it is soon found that the little additions are of no value. They may
bring us further from the goal as likely as nearer" (p. 54).

Woodworth reasoned that aimed movements sures of error. Movements were performed with
with vision should be more accurate than those eyes open or closed, paced by a metronome at
made without vision, but only if there is sufficient speeds of 200 to 20 cycles per minute (i.e., move-
time to use the visual feedback. Movements that ment times [MTs] that ranged from 300 ms to 3 s
are completed before visual feedback can be per cycle, respectively). Aiming was done only in
processed would be expected to result in target one direction-the return movement was made
dispersions that are equally inaccurate for blind to a physically restricted stop. Thus, the values
and sighted aims. The critical question, then, is, of available time to process visual feedback from
What is the minimum amount of time required the stop to the intended target were about half of
to process visual feedback? A number of experi- the total cycle time, or between 150 ms and 1.5 s
ments using more or less the same experimental (Carlton, 1992).
method as Woodworth's have addressed this The average errors (similar to absolute error
question. Therefore, we will examine the Wood- [AE]; see chapter 2) in the eyes-open and eyes-
worth (1899) study in some detail (see also "RS. closed conditions are illustrated by the open and
Woodworth on Manual Aiming"). filled circles in figure 5.5. As seen on the far left
side of this figure (short MTs), the eyes-open
Woodworth (1899) To assess the minimal time and eyes-closed conditions produced equivalent
required to process visual feedback, Woodworth average error. As shown on the right side of the
asked his subjects to make back-and-forth move- figure (long MTs), the availability of vision in
ments of a pen. The goal for each action was to the eyes-open condition resulted in much lower
produce a movement equal in length to the pre- average error, relative to the eyes-closed condi-
ceding one. A strip of paper moved beneath the tion and relative to short MTs with the eyes open.
subject's pen at a constant velocity; thus a record The critical point in the graph occurs when the
of the previous movement could be used as the eyes-closed and eyes-open conditions begin to
target for the next and served also as a permanent depart in terms of average error. This departure
record that could be used later to calculate mea- point represents the time when vision of the target
144 $1 Motor Control and Learning

Smith, Kornblum, Abrams, & Wright, 1990,

5 describe other peculiarities).
In another well-known experiment, Keele and
Posner (1968) overcame these problems by using
E4
.§. a discrete-task version of Woodworth's research
... strategy. They trained their subjects to move a
e stylus to a small target about 15 cm away. Knowl-
lii 3
GI
Cl edge of results about MT was provided, and sub-
f!
GI jects were trained to move as closely as possible
~ 2
to assigned MT goals (150, 250, 350, and 450 ms).
On certain randomly determined test trials, the
experimenters turned off the room lights as soon
as the stylus left the starting position, so that the
entire movement was made in the dark.
200 600 1000 1400 Keele and Posner's results are illustrated in
Time to target {ms)
figure 5.6, revealing the probability of missing
this target as a function of the vision condition
FIGURE 5.5 Aiming errors in moving to targets with the
and the actual average MTs. In the 150 ms MT
eyes open and closed.
Adapted from Woodworth 1899.
condition (the actual MT was 190 ms), about
as many target misses were recorded when the
lights were on (68%) as when the lights were
significantly reduces average error compared to the off (69%). As the MTs increased, an advantage
average performance of the eyes-closed condition emerged when the room lights were on, consis-
(represented by the dotted line in figure 5.5). This tent with the view that vision can be used when
point of departure occurs between movements of sufficient time is available for detecting and
215 ms duration and movements of 250 ms dura- correcting errors.
tion. That is, movements with durations of 250
ms or longer could benefit from the availability Visual Feedback Uncertainty Despite the
of vision (as seen by reduced error), but move- changes in the design of their study, Keele and
ments with durations of 215 ms or less did not
benefit when vision was available. Notice that,
for this task, the estimate of the time to process
visual feedback seems to be about the same as 0.7
in responding to a visual stimulus in a choice
reaction-time (RT) task. Gi 0.6
El
Keele and Posner (1968) Although Wood- .fl
gi 0.5
worth's empirical and theoretical contributions "iii
U)
are considered landmarks in motor control .E
research (Elliott et al., 2001; Newell & Vaillan- - 0.4
0
court, 200lb ), there were some peculiarities. ~
The repetitive, back-and-forth nature of the task :E 0.3
Ill
..c
meant that subjects had to aim at the target and e
a. 0.2
prepare to reverse the movement (to start it) at
about the same time, making the role of vision
more complex than in an aimed movement that 0.1
simply terminates at the target, such as using a
computer mouse to move a cursor onto a screen
icon. As well, the nature of the task required Actual MT {ms)
subjects to try to match the previous movement
they had made; the target was more ambiguous FIGURE 5.6 Percentage of target misses with room
than, for example, hitting a nail with a hammer lights on and off during movements of various durations.
or striking a computer key with a finger (Meyer, From Keele and Posner 1968.
 Sensory Contributions to Motor Control 145

Posner's estimate of the time required to process

visual feedback information (between 190 and 260 5
ms) was very similar to the estimate suggested by
Woodworth (1899). But uncertainty about whether
or not visual feedback would be available on a 4
trial seems to have played an important role in the
experiment of Keele and Posner (1968). In their
methodology, the room lights went off unexpect-
e
§. 3
edly on some trials in a randomly determined ...
E
...
order. Perhaps if subjects knew in advance that Cl)

vision would or would not be available, their :e

0
2
I-
attention might have been focused more appro-
priately, and the estimate of the time to process
visual feedback might be considerably less than
in situations in which subjects are "surprised" by
the presence or absence of vision. If the subject
could never fully expect to have vision available, 75 150 225 300 375 450 525 600 675
then a strategy designed to employ vision might Actual MT (ms)

be avoided in favor of an alternative strategy. The

unpredictability of vision in Keele and Posner's FIGURE 5.7 Target accuracy (total variability, E) in a
stylus-aiming task as a function of movement time (MT)
study may have prevented subjects from using
and lighting conditions.
vision to its fullest when it was available. Thus, Adapted from H.N. Zelaznik, M. Hawkins, and L. Kisselburgh, 1983, "Rapid visual
visual processing time in the study may have feedback processing in single-aiming movement,'' Journal of Motor Behavior 15:
217-236.
been overestimated.
To address this issue, Zelaznik, Hawkins, and
Kisselburgh (1983) and Elliott and Allard (1985)
order. No advantage for the availability of visual
performed experiments similar to Keele and Pos-
feedback was found when MTs were about 75 ms;
ner's except that in some conditions the subjects
knew in advance when vision would or would however, clear differences were present for MTs of
not be available. Elliott and Allard (experiment about 150 ms and longer. These data suggest that
1) used a series of trials in which the room lights vision, when its presence can be expected, can be
were either left on or were turned off at random (as used in far less time than the minimum suggested
in Keele & Posner), and another series of trials in by Woodworth (1899) and by Keele and Posner
which room lights were manipulated in a blocked (1968), perhaps with visual processing times as
order (i.e., entire sets of trials were done either short as 100 ms. In other experiments, Zelaznik
with or without vision). For movements made as and colleagues showed that the advantage for
fast as possible (225 ms), the availability of vision the lights-on condition remained even if the pro-
increased target aiming accuracy only when trials cedures were reversed so that the lights were off
were blocked; no differences were found between initially but would come on suddenly.
vision and no-vision conditions when the trials Other studies support the contention that the
were randomized (i.e., when visual feedback visual processing values obtained by Wood-
availability could not be anticipated). These find- worth (1899) and Keele and Posner (1968) were
ings raised the possibility that unpredictability overestimates of the minimum time to process
and frequent switches between vision and no- visual information. Carlton (1981a) prevented
vision conditions reduced processing efficiency visual feedback during the initial 75% of the
(Cheng, Luis, & Tremblay, 2008). distance to the target; he found that vision could
Zelaznik and colleagues (1983) also found be used in as little as 135 ms (see also Carlton,
that visual feedback could be used more quickly 1979, 1981b; Spijkers & Lochner, 1994). Smith and
in predictable trials than when it was presented Bowen (1980) showed that distorted or delayed
randomly. The findings from their third experi- visual information caused disruptions in move-
ment are illustrated in figure 5.7. In this study, all ment accuracy when the MTs were only 150 ms,
trials were conducted in a blocked (predictable) also suggesting that visual processing time was
146 Motor Control and Learning

far faster than had been previously measured. a small bulb inside the ball could be lit for peri-
Similar results were reported by Elliott and Allard ods of 100, 150, 200, 250, or 300 ms after it hit
(1985, experiments 2 and 3). Suddenly perturbing the trampoline. With a ball flight time (from the
the location of a visible target (immediately after point when the ball left the trampoline until the
the movement to the old" target is begun) also
/1 catch was attempted) of 400 ms, these viewing
resulted in very fast movement corrections, with times left the subject without sight of the ball
latencies of about 100 ms (Paulignan, MacKenzie, for periods of 300, 250, 200, 150, or 100 ms prior
Marteniuk, & Jeannerod, 1991; Pelisson, Prablanc, to hand contact, respectively. The proportion of
Goodale, & Jeannerod, 1986), although perturbing balls caught in each of the vision conditions is
the target size necessitates considerably longer presented in figure 5.8. Although performance
processing times (Paulignan, Jeannerod, MacK- under the most favorable condition was not as
enzie, & Marteniuk, 1991). accurate as in a control condition in which all
When vision is available and MTs are sufficiently the room lights were on3 (represented by the
long to use that information, what is typically dotted line in figure 5.8), Whiting and colleagues
observed is a correction in the trajectory of an aimed found that catching performance improved as
movement to improve its end-point accuracy. Tech- the viewing time became longer-even when the
nological advances in movement analyses toward 150 and 100 ms conditions were compared! Thus,
the end of the past century permitted researchers one conclusion is that visual information of the
to examine the kinematics of these movements ball flight could be used in as little as 100 ms to
(see chapter 2) to make refined estimates of when improve catching performance (see Savelsbergh
subjects were using visual information. Reviews & Davids, 2002; Savelsbergh & Whiting, 1996;
of this research by Carlton (1992), Glencross and and Savelsbergh, Whiting, & Pijpers, 1992, for
Barrett (1992), Elliott and colleagues (2001), and discussion of related experiments).
Khan and colleagues (2006) include discussions of One potentially confounding factor in the
a number of related methodologies used to inves- study by Whiting and colleagues (1970) was that
tigate the time to process visual feedback. When the length of the viewing period covaried with
all this evidence is taken together, it suggests that the amount of time without vision; the greater
no single, absolute estimate of the time to process the time with vision, the less time without it, and
visual feedback is likely to be correct. The nature vice versa. An experimental approach that allows
of the task, the type of visual information available, examination of the independent contribution of
the predictability of this information, and so on all these factors uses stroboscopic conditions. The stro-
affect the speed and effectiveness with which we boscopic conditions are simulated with the use of
use visual information in aiming. goggles that can be alternately opened and closed
during ball flight to provide periodic snapshots"
/1

Intercepting Moving Targets of visual information. Using this method, Elliott,

Hitting a nail with a hammer is an example of just Zuberec, and Milgram (1994) examined the inde-
one type of action in which vision is used in motor pendent contributions of the length of time that
control; the object is stationary, and the goal of the the goggles were open and the length of time that
action is to make contact with the object (defined vision was not available (goggles closed). Interest-
as a "closed" skill in chapter 2). Now consider ingly, Elliott and colleagues found that the most
another class of skills in the situation in which the important contribution to catching performance
environment is changing during the action (called was not the amount of time vision was available,
"open" skills in chapter 2). Does processing visual but rather the time between the visual snapshots
information about moving objects differ from that of the ball (i.e., the length of time without vision).
for stationary objects (Pelisson et al., 1986)? Subjects could perform well with as few as 20 ms
An early study by Whiting, Gill, and Stephen- of available vision of the ball as long as no more
son (1970) suggests that the answer might be yes. than 80 ms intervened between "snapshots" (see
In this experiment, a ball was dropped from a 3 also Assaiante, Marchand, & Amblard, 1989).
m height to an angled trampoline-like spring, Apparently, longer periods without vision updat-
which bounced the ball another 3 m toward the ing can support performance during aiming at
subject, who attempted to catch it. Trials were stationary targets (Elliott, Chua, & Pollock, 1994).
conducted in a completely darkened room, and These findings do not constitute rejection of ear-
 Sensory Contributions to Motor Control 147

1.0

0.9

-°'
.c 0.8
:I
11)
t)
0.7
.!!J.
Cii
.c
0 0.6
c:
0
t: 0.5
0
0..
0
.....
ll. 0.4

0.3

300 250 200 150 100

Time before attempted catch
without vision (ms)

FIGURE 5.8 Number of balls caught in full room light and in conditions in which lights have been extinguished prior to
attempted catch.
Adapted from H.T.A. Whiting, E.B. Gill, and J.M. Stephenson, 1970, "Critical time intervals for taking in flight information in a ball-catching task'; Ergonomics 13: 269.

lier estimates of the minimum time to process 190 ms in McLeod, 1987; 150-190 ms in Carlton,
visual information. Rather, they serve to reject Carlton, & Kim, 1997). The longer delay times in
the idea that a minimum amount of continuous these latter two studies, however, may be attrib-
visual information is needed. These findings utable to the unexpectedness of the change in
suggest that visual information can be processed flight path and the larger objects that were being
intermittently (or quasi-continuously), using very manipulated (a cricket bat in McLeod, 1987; a
short durations of information (these ideas are tennis racket in Carlton et al., 1997).
reviewed in Elliott, Binsted, & Heath, 1999). Interpreting the Evidence
One can raise the argument that experimental
It seems that vision, under various conditions,
occlusion of vision is not a natural approach to
can be processed for motor control in consider-
examining the speed of visual processing. In other
ably less time than our usual estimates of RT to a
words, in our example of the toolshed, the lights visual stimulus would suggest. One interpretation
do not go out very often. Critics of this approach of these differences in visual processing time esti-
have used a number of alternative paradigms mates is that the various experimental conditions
in which the use of visual feedback informa- appear to introduce strategic differences in the
tion is inferred from the adjustment (homing ways subjects prepare for specific types of visual
in) actions that occur during the preparation for information. For example, Glencross and Barrett
object contact. For example, Bootsma and van (1992) suggest that in some situations, subjects
Wieringen (1990) found that expert table tennis may make deliberate errors (e.g., undershooting the
players could use visual information about the target) in the initial movement phase, perhaps in
moving ball to adjust their forehand drive in the order to reduce the uncertainty of the information
last 105 to 156 ms of ball flight (depending on the specified by the visual feedback (see also Barrett &
individual). Estimates ranging from 50 to 135 ms Glencross, 1989; Carlton, 198la). With use of this
were reported by Lee, Young, Reddish, Lough, strategy, the amount of information to be processed
and Clayton (1983) in a task that required subjects in the visual display is reduced because the direc-
to jump to hit a falling ball. Longer estimates for tion of the corrective action is highly predictable.
responding to changes in the path of a ball after a Another interpretation is related to the hypoth-
perturbed bounce have also been obtained (about esis of two visual systems proposed by Trevarthen
148 llll Motor Control and Learning

(1968) and Schneider (1969), discussed earlier in gests that the visual system specifies these types
the chapter (see also Aschersleben & Miisseler, of information in different ways.
1999). According to the dissociation (Goodale Temporal Information As mentioned previ-
& Milner, 1992), much of the visual information ously in the chapter, there now appears to be
for movement control is handled by the dorsal considerable evidence that Tc can be specified
system, which is thought to be nonconscious and by tau (Lee et al., 1983; but see also Abernethy &
spatially oriented (Bridgeman et al., 1981; Pelis- Burgess-Limerick, 1992). An experiment by Sav-
son et al., 1986). However, various experimental elsbergh, Whiting, and Bootsma (1991) illustrates
manipulations of visual feedback processing time how tau is used to specify when to carry out the
may require use of the ventral visual system. hand actions required for grasping a moving ball.
Thus, it may be that the motor control studies Subjects were positioned with their elbow on a
using stationary stimuli have forced the subjects table and their wrist locked into a position such
to use the (slower) ventral system, which is not that a ball that swung on a pendulum would
the system they would normally use in "real" come directly into contact with the subject's
environmental skill situations. What is clear is palm. The only task was to time the catch of the
that the time to use visual information changes as ball with a grasping action; there was no spatial
a function of the characteristics of the target and uncertainty. The experiment used 5.5 and 7.5 cm
the goals of the movement (Carlton, 1992). In the balls; a third ball was 5.5 cm but was covered with
next sections we describe other evidence related a balloon that could be inflated to appear to be the
to the role of vision in motor control. size of the 7.5 cm ball. Once the ball was released,
the balloon deflated from 7.5 to 5.5 cm over the
Vision and Anticipatory Actions period of the ball flight (which was about 1.7 s).
Situations in which the movement goal involves A plot of the relative size of these objects is
an interaction with a changing environment shown in figure 5.9a for a constant ball size (A)
include cases in which the individual, the envi- and with the deflating balloon (B). As the balls (A)
ronment (or an object in the environment), or approached the subject, the retinal image increased
both are moving. For example, in ball sports, in size about 1,200%, but the retinal image of the
there are times when the performer must stand deflating balloon (B) expanded only about 500%.
still and intercept a moving object (as in batting a Thus, the rate of expansion for the balls was con-
ball) or must move to intercept a ball in flight (as siderably faster than for the balloon. The average
in catching a forward pass in American football). apertures of the hand changed as subjects prepared
Activities such as running or driving a car involve to catch the moving objects during the final 200
situations in which the motion of the individual ms, as illustrated in figure 5.9b. Throughout the
results in a changing visual array. In all cases, final 200 ms of ball flight, the hand aperture was
information about the changing environment larger in anticipation of catching the large ball (L)
is used to prepare anticipatory actions that allow compared to the small ball (S). But, for the deflat-
people to complete the task (such as catching a ing balloon (B), at 200 ms before contact, the hand
ball or avoiding objects in their path). The issue had an aperture between the sizes of the large and
that we deal with next concerns how these actions the small ball; and as the balloon decreased to the
become specified by the changing environment. size of the small ball, so did the hand aperture.
Postexperiment interviews indicated that the
lnterceptive Control subjects were not aware of the deflating properties
Most of the research on the role of vision during of the balloon; the scaling of hand aperture to the
interceptive activities has involved a moving moment-to-moment size of the deflating balloon
object, as in catching or striking a ball. Accord- might have occurred without conscious awareness.
ing to Savelsbergh and colleagues (Savelsbergh One interpretation is that Tc is used by the
& Whiting, 1996; Savelsbergh et al., 1992), the subject via the expansion of the optical flow, and
primary use of vision is to specify information that the catching actions were scaled in anticipa-
about (1) when to intercept an object, in particular, tion of intercepting the object with a temporally
temporal information about Tc; and (2) where to precise grasp. However, as cautioned by Aber-
go to intercept it, information about the spatial nethy and Burgess-Limerick (1992), the fact that
characteristics of the ball flight. Research sug- this experiment and others (e.g., Lee, Lishman, &
 Sensory Contributions to Motor Control m 149

1200

1000
~

~ 800
c
0
"iii 600
c
Ill
Cl.
>< 400
w

200

-1000 -500 0 250

a Time before contact (ms)

6
'E
.2.
!!! 5
::s
t::
GI
Cl.
Ill 4
"C
c
Ill
J:
3

-180 -140 -100 -60 -20 0 20

b Time before contact (ms)

FIGURE 5.9 Temporal coordination of hand aperture with apparent ball size; zero point on horizontal axis indicates con-
tact with ball. (a) Apparent rate of expansion during ball flight (A = constant ball size, B = deflating balloon). (b) Hand aper-
ture changes prior to contact (L =large ball, B =deflating balloon, S =small ball).
Adapted, by permission, from G.J. P. Savelsbergh, H.T.A. Whiting, and R.J. Bootsma, 1991, "Grasping tau," Journal of Experimental Psychology: Human Perception and
Performance 17: 317 and 321. Copyright© 1991 by the American Psychological Association.

Thomson, 1982; Savelsbergh, Whiting, Pijpers, & McLeod and Dienes (1993, 1996), following
van Santvoord, 1993) provide evidence consistent earlier work by Chapman (1968; see also Michaels
with the use of tau does not rule out the possibil- & Oudejans, 1992; Todd, 1981), suggest that the
ity that Tc can be specified in other ways as well angle of elevation of gaze (ex) between the fielder
(see also Cavallo & Laurent, 1988; Rock & Harris, and the ball is used as the key information for
2006; Smeets, Brenner, Trebuchet, & Mestre, 1996; making decisions about where to go to intercept
Stewart, Cudworth, & Lishman, 1993; Tresilian, the ball. The sequence of illustrations in figure
1995, 1997, 1999; Wann, 1996). 5.10 describes McLeod and Dienes' (1996) analy-
Spatial Information A difficult task faced by sis. Figure 5.lOa shows how ex is computed in
an outfielder in baseball is judging where to go relation to a fielder in baseball or cricket. Given
to catch a batted ball, especially when the ball is the path of the ball in figure 5.lOa, the fielder
hit directly at the outfielder. It is not uncommon, determines that movement toward the ball (to
even among the most highly skilled players, for decrease x) is required before the ball hits the
an outfielder to run forward to catch a ball, only ground (as y approaches 0). The fielder's for-
to stop, reverse and run backward after realizing ward movement is depicted in figure 5.10, band
he has made a mistake. What information is used c (further decreasing x), as the ball falls toward
to make these decisions about where to go to catch the ground (further decreasing y). In moving
the ball? forward, the fielder must keep the angle of gaze
150 Motor Control and Learning

-., (a) between 0° and 90°. If a ever becomes 0° (i.e.,

......
w y = 0 and x > 0), then the ball will land in front of

t
y
the fielder (figure 5.lOd). Likewise, if a ever equals
90° (i.e., x = Oand y > 0), then the ball will go over
the fielder's head (figure 5.lOe). But how does the

i fielder use this information to make the catch?

CY

___._~
A considerable amount of research now sug-
x gests that the fielder will arrive at the correct place
a
to catch the ball if he adjusts running speed accord-
ing to visual information (Lenoir, Musch, Thiery, &
Savelsbergh, 2002; McLeod & Dienes, 1993, 1996;
McLeod, Reed, & Dienes, 2001; Michaels & Oude-
jans, 1992; Oudejans, Michaels, Bakker, & Davids,
1999). According to McLeod and colleagues, field-
ers accelerate and decelerate their running speed
based on successive samples of vertical gaze (a,
b 0° <ex< go 0
position of the height of the ball relative to eye
level) and horizontal gaze (o, position of the ball
relative to its origin M, home plate in the case of
catching a hit baseball). Specifically, where fielders
must run to the left or right (i.e., if the ball is not
hit directly at the fielder), they adjust their running
speed so that their horizontal gaze (o) increases
at a constant rate and the rate of increase in their
vertical gaze (a) decreases steadily, as illustrated
0° <ex< go 0 in figure 5.11 (McLeod, Reed, & Dienes, 2003, 2006;
c
82
B, ,..... ---0--.. .......... B3

''
''
site
'
'o·
\
\
\
\
\
d \
\
\ s.

ex= go 0

e FIGURE 5.11 Running to catch a fly ball. Running speed

of the fielder is adjusted in order to maintain a constant
rate of increase in the horizontal gaze (8), the gaze angle
FIGURE 5.10 How a fielder uses the angle of elevation relative to the origin of the hit) and to maintain a steadily
of gaze to decide whether to run forward or backward to decreasing rise in the rate of the vertical angle of gaze (ex,
catch a ball. height of the ball).
Reprinted, by permission, from P. Mcleod and Z. Dienes, 1996, "Do fielders know Reprinted, by permission, P. Mcleod, N. Reed, and Z. Dienes, 2006, "The
where to go to catch the ball or only how to get there?" Journal of Experimental generalized optic acceleration cancellation theory of catching," Journal of
Psychology: Human Perception and Performance 22: 538. Copyright© 1996 by Experimental Psychology: Human Perception and Performance 32: 139-148.
the American Psychological Association. Copyright© 2006 by the American Psychological Association.
 Sensory Contributions to Motor Control iii 151

Reed, McLeod, & Dienes, 2009). Of course, this common finding, as pointed out by Lee (1980),
strategy does not guarantee that the fielder will is that people who are blind are generally less
successfully make the catch, only that the fielder stable in posture than are sighted people; those
will be in the right position to try to do so. without sight sway more when they stand. Also,
sighted people sway more when they have their
Controlling Whole-Body Actions
eyes closed versus open. This suggests that vision
The work by McLeod and others represents just might have a role in balance (see also Shumway-
a small part of a much larger research field deal- Cook & Woollcott, 2007, chapter 9).
ing with the control of the body as a whole, as in But more convincing evidence for this asser-
balance and in locomotion. Early thinking about tion comes from some of Lee's experiments (Lee
postural control tended to focus on mechanisms & Aronson, 1974). Lee used a "moving room"
that are seemingly obvious contributors to the apparatus whereby a person stood on a station-
skills, such as the vestibular apparatus in the inner ary floor in a three-sided "room" with walls
ear. The receptors are sensitive to deviations from (suspended from a ceiling) that could be moved
the vertical, to the orientation of the body in space, backward and forward as a unit without move-
and to the accelerations applied to the head when ment of the floor. The experimenters studied the
the body is moved. All these aspects of balance effect of this wall movement on the posture and
possibly could be signaled by these receptors. sway of the subject (figure 5.12 shows the general
A second class of processes for the control of
balance includes the various receptors associated
with the joints and muscles. These processes are
discussed in more detail later in this chapter; but
for now, consider that in posture, the system is
organized to maintain particular angles (which
implies muscle lengths) for the joints associated
with a particular position of the body in space.
When the body begins to lose this posture, the
movement of the joints away from the goal
position can be sensed by the joint receptors, or
perhaps by the stretch of the muscles that control
the joint, or by both. Also, there could be tactile
sensation from the feet or toes indicating loss of
balance. Each of these receptors, alone or in some
combination, could conceivably provide the input
necessary in order for the person to sense a loss
of balance and could provide a basis for initiating
a correction. Numerous chapters in the volume
by Gandevia, Proske, and Stuart (2002) review
this evidence in considerable detail (see also Ros-
signol, Dubuc, & Gossard, 2006; Shumway-Cook
& Woollacott, 2007; Woollacott & Jensen, 1996).
Vision and Balance A third source of feedback
about balance was emphasized by David Lee (see
Lee, 1980, for a review). Earlier in this chapter
we discussed the concept of visual propriocep-
tion, noting that it tells us where our eyes (and
therefore our head and body) are in space and
how they are moving (if they are) via analysis of
the patterns of optical flow received from the sur-
FIGURE 5.12 Experimental apparatus and paradigm for the
rounding surfaces and elements in the environ-
"moving room:' Dr. Lee is second from left.
ment (seep. 142, "Visual Proprioception"). Could Reprinted, by permission, from D.N. Lee and E. Aronson, 1974, "Visual proprioceptive control of
vision be involved in the control of balance? One standing in human infants," Perception and Psychophysics 15: 530.
152 !II Motor Control and Learning

arrangement). With small children as subjects, Consider jogging along an irregular path. In
moving the wall a few centimeters toward the this case the placement of the feet with each step
subject caused loss of balance, resulting in a rather is much more critical than, say, on the smooth
ungraceful sitting response and great surprise on surface of a running track; to avoid injury, the
the part of the child. Moving the walls away from jogger's goal is to try to step on ground points
the child caused a drastic forward lean, which that indicate good footing and to avoid unstable
resulted in a stumble or a fall. When adult subjects footings (such as tree roots and loose debris). If
were studied, the effect was less dramatic; but the jogger is running at a constant velocity, how
increases in sway, in-phase with the direction of is stride length altered in order to step selectively
the wall movement, could be seen. on good ground footings?
How can this be explained? Remember, the floor Warren, Young, and Lee (1986) had two
of the room was fixed, so that the movements of experienced athletes run at a constant velocity
the walls could not have exerted mechanical influ- on a treadmill. At various times visual targets
ences on the position of the subjects. The mecha- appeared on the treadmill, and the runner's task
nisms associated with the joint angles and muscle was to try to step on these targets by altering stride
lengths, as well as the vestibular apparatus, also lengths as appropriate. The running kinematics
were not directly affected. The most reasonable revealed that almost all the variations in stride
explanation is that moving the wall toward the length were attributable to changes in the vertical
child changed the optical flow and the optical impulse applied at takeoff. An increase in the verti-
array. If the child was using this information as cal impulse propels the center of gravity farther
a source of feedback, she could have interpreted upward and thus lengthens the time the lead leg
the changed visual array as a loss of balance and spends in the air; thus, with forward velocity con-
consequently produced a compensation in the stant, the stride length will also increase. Accord-
opposite direction. The walls appearing to move ing to Warren and colleagues (1986), the vertical
closer to the eye would, if the room were "normal," impulse necessary to achieve a specific distance is
provide an optic array signaling that the person determined by the difference in the times to con-
was falling forward, and a compensation to move tact (T) of the takeoff and target points, also called
backward would be expected. This is just what Lee the tau gap (.:l 'T'). Since the runner is moving at a
found: Moving the wall toward the subjects caused constant velocity, the tau gap specifies the dura-
them to fall backward. Nashner and Berthoz (1978) tion, and hence the length, of the stride required
showed that responses to the changed optical flow for the foot to go from the first to the second target
could be generated in about 100 ms (from optical (see also Hollands, Marple-Horvat, Henkes, &
Rowan, 1995; Patla, 1989; Patla, Robinson, Sam-
change to change in postural muscles' electromyo-
ways, & Armstrong, 1989; Warren & Yaffe, 1989).
grams), which shows that these compensations
Of course, vision supplies much more than
are very fast and automatic and do not involve
Tc information about how to walk without col-
the stages of processing discussed in chapter 3.
liding with objects. Environmental cues provide
Vision and Locomotion The role of vision information about how we must accommodate
during walking and running also seems to be our actions. For example, in walking through
greatly influenced by the optical flow of infor- apertures such as doorways, there is a critical
mation; much of the thinking about how we use ratio between the aperture and shoulder width:
this information continues to be influenced by For any aperture less than about 1.3 times the
Gibson (1958, 1966, 1979; see also Warren, 1998, shoulder width, humans will generally rotate
for a review of Gibson's perspective and Wann & their upper body in order to increase the effective
Land, 2000, for a different perspective). We walk passage width (Warren & Whang, 1987). When
down busy streets and through crowded corri- we step over obstacles, a critical height value also
dors and rarely bump into things or people. We appears to be evident. There is even evidence that
walk on surfaces that have different textures and the leading leg uses a greater clearance difference
give them very little attention. But obviously we if the obstacle is fragile (Patla, Rietdyk, Martin, &
depend on vision to make our way through the Prentice, 1996). Many more examples have been
environment. Is vision in this context used in the investigated (e.g., Higuchi, Takada, Matsuura, &
same way as discussed in the previous sections? Imanaka, 2004). However, the main point here is
 Sensory Contributions to Motor Control 111 153

that vision is used in complex ways in order that 2009) suggested that animals (including humans)
we may locomote through the environment safely can determine Tc auditorily by using tau in exactly
and efficiently (Patla, 1997, 1998). the same way as described earlier-except that
tau is based on the acoustic-flow field rather than
the visual-flow field (see also Ashmead, Davis, &
Audition Northington, 1995; Jenison, 1997).

Another of the senses traditionally classified as

exteroceptive is hearing, or audition. Certainly, Proprioceptors
audition has a strongly exteroceptive role, inform-
ing us about the nature of movements in our envi- In the next section we review the set of sensors
ronment-the direction of approach of a bicyclist that provide proprioceptive information (or
in the dark, the sound of the starter's gun, and so kinesthesis) about the movement of the body.
on. But at the same time, like vision, audition can Although they do not seem to be as salient for
tell us a great deal about our own movements. movement control as vision is, these sensors are
As we all know, most of the movements we fundamental in their importance for closed-loop
make in the environment produce sounds, such control (Abbs & Winstein, 1990; Latash, 2008b). In
as the sound of footsteps when we are jogging the first section we outline briefly their anatomy
or the sound of our own speech. The nature of and function, and later we describe their role in
these sounds, then, provides us with a great deal motor control.
of information about our actions-for example,
crunching sounds tell us about the kind of terrain Vestibular System
we are jogging on, and beeps from the telephone
or automatic teller machine sometimes tell us Located in the inner ear are sensors that provide
whether or not we are using the keypad correctly. information about movements of the head. One
Sounds serve an important role in sporting events aspect of head movement that is critical for motor
as well. Judges use the sound of a diver as he enters control is its orientation with respect to gravity;
the water as one indication of performance. The that is, whether the head is upside down, tilted,
"crack of the bat" provides clues to the outfielder and so on. Such information is provided by the
regarding how solidly a ball has been struck in labyrinthine receptors such as the otolith organs-
baseball. Golfers can often tell what part of the club two small structures located in the inner ear (the
face has made contact with the ball on the basis saccule and utricle) that signal information about
of auditory cues. And yet, with all of this sensory the orientation of the head with respect to gravity.
feedback available to the performer, the role of If the head is spinning (e.g., in a somersault), they
audition and how it is used in motor performance provide information about the rate and direction
remains a largely understudied research area (but of spin. Located near the otolith organs are three
see Gray, 2009b; Roberts, Jones, Mansfield, & Roth- fluid-filled half circles, called the semicircular
berg, 2005, for some exceptions). canals. Because the canals are oriented in each
To some extent, audition and vision are very of the major planes of the body (frontal, sagittal,
similar, providing both exteroceptive and pro- horizontal), these structures are in a position to
prioceptive information. There are obvious differ- sense particular directions of movement, as well
ences, of course. In general, auditory information as rotation. All these vestibular structures contain
is processed faster than visual information, but thick fluid that moves when the head position
vision seems to provide more useful informa- changes. The movement of the liquid bends tiny
tion than audition. But there are some sugges- hairs that send information to the central nervous
tions that the two types of information as used system, informing about movements of the head.
in similar ways. For example, while flying in a As one might imagine, these structures are impor-
dark cave, bats use acoustic information to orient tant in balance, as well as in movements for which
themselves; sounds from objects (exteroceptive the person requires information about forces and
feedback) and from their own movements (expro- accelerations applied to the head (e.g., in flying a
prioceptive feedback) provide information for plane, doing a somersault; for further discussion
orienting them within the cave. Lee (1990, 1998, see Kelly, 1991; Lackner & DiZio, 2000).
154 I'll Motor Control and Learning

Muscle Receptors when the main muscle is stretched. The spindle

consists of three main components: small muscle
Two main types of receptors provide complemen-
fibers called intrafusal (muscle) fibers (meaning in
tary information about the state of the muscles.
the spindle) that are innervated by gamma(')') effer-
The muscle spindle is located in the fleshy part
ent (motor) neurons. The spindles can activate type
of the muscle body and is most active when
Ia and type II afferent (sensory) neurons. 4 The fibers
the muscle is stretched. The Golgi tendon organ is
in the main muscles are called extrafusal fibers
located in the junction between the muscle and
(meaning outside the spindle). The intrafusal fibers
tendon and is most active when the muscle is
are made up of two types, bag and chain fibers,
producingforce (Gordon & Ghez, 1991; Pearson
whose polar ends provide a tension on the central
& Gordon, 2000a).
region of the spindle, called the equatorial region.
Muscle Spindles The sensory receptors located here are sensitive to
Lying between the fibers of the main muscles of the length of the equatorial region, mainly when
the body are small spindle-shaped (cigar-shaped) the spindle is stretched (because the whole muscle
structures that are connected in parallel with the is stretched, including the spindle). The major
muscles in such a way that they are stretched neurological connection to this sensory region is

Skin------
Free nerve ending
(pain, temperature)

Meissner's
corpuscle (touch)

Joint
Pacinian corpuscle capsule
(pressure)----

Golgi
tendon
organ

Muscle spindle - - -

FIGURE 5.13 Muscle receptors: muscle spindle and Golgi tendon organ.
Reprinted, by permission, from J. Wilmore and D. Costill, 2007, Physiology of sport and exercise, 4'" ed. (Champaign, IL: Human Kinetics), 93.
 Sensory Contributions to Motor Control 155

the Ia afferent fiber, whose output is related to the information to the sensory cortex of the brain,
length of the equatorial region (thus signaling joint where other sensory events were thought to be
position information indirectly), as well as to the registered. This was suggested by Gelfan and
rate of change in length of this region (velocity Carter's (1967) research on humans undergo-
information). The spindle connects to the alpha ing operations involving wrist tendon repair
motor neurons for the same muscle, providing under local anesthetic only. When the tendons
excitation to the muscle when it is stretched. This is and connected muscles were passively stretched
the basis for the so-called "stretch reflex" discussed and the subjects were asked what they felt when
later in this chapter. Thus, the spindle appears to the tendon was pulled, they usually reported no
have a strong role in movement regulation (Vallbo, sensations or sensations that were inconsistent
1974; Pearson & Gordon, 2000b). with the direction of tendon pull. Primarily for
There has been a great deal of controversy these reasons, an older view held that the muscle
about what the spindle actually signals to the spindles were not important for the conscious
central nervous system (see Gandevia & Burke, perception of movement or position.
1992, and commentaries). A major conceptual Data from Goodwin, McCloskey, and Mat-
problem in the past was that the output via the Ia thews (1972) and others (Rogers, Bendrups, &
afferent that presumably signals stretch or veloc- Lewis, 1985; Sittig, Denier van der Gon, & Gielen,
ity is related to two separate factors. First, Ia affer- 1985a) have helped to change this point of view.
ent output is increased by the elongation of the In these studies, intact subjects had a rapid vibra-
overall muscle via elongation of the spindle as a tion applied to the biceps tendon at the elbow.
whole. But, second, the Ia output is also related to The blindfolded subject was asked to "track" the
the stretch placed on the equatorial region by the (experimenter-determined) passive movements
intrafusal fibers via the gamma motor neurons. of the vibrated arm with corresponding active
Therefore, the central nervous system would have movements of the other arm; thus, the subject
difficulty in interpreting changes in the Ia output had to perceive where the right arm was and
as due to (a) changes in the overall muscle length match the (consciously felt) position with move-
with a constant gamma motor neuron activity, (b) ments of the left arm. The vibration of the tendon
changes in gamma motor neuron activity with a produces a small, rapid, alternating stretch and
constant muscle length, or perhaps (c) changes release of the tendon, which affects the muscle
in both (see "Muscle Spindles and the Gamma spindle and distorts the output of the Ia afferents
Loop" on p. 164). from the spindles located in the vibrated muscle.
The system works to disentangle these factors Goodwin and colleagues (1972) found as much
by making use of the gamma motor system and as 40° misalignment of the vibrated arm with the
the intrafusal muscle fibers. If the muscle should nonvibrated arm. The interpretation was that the
contract actively (and thus shorten), the stretch vibration distorted the Ia afferent information
of the equatorial region of the spindles would coming from the same muscle, which led to a
be reduced, and the Ia afferents would decrease misperception of that limb's position and hence
their activity. Under such conditions, the spindle to improper (conscious) decisions about the
would become insensitive to any additional, per- positioning of the opposite limb. The argument,
haps unexpected, muscle stretch. However, at the then, is that this information from the Ia afferent
same time the overall muscle is activated (via the actually did reach consciousness and thus the
alpha motor neurons), the gamma motor system Ia afferent was the basis for knowing the limb's
activates the intrafusal fibers, which applies ten- position. (To control for the possibility that the
sion to the equatorial region of the spindles. This vibration merely influenced the structures in the
process has been termed alpha-gamma coactivation. joint capsule of the elbow, the authors placed
This process acts to compensate for the (active) the vibrator over the triceps tendon; now, the
change in the overall muscle, so that the spindle misalignment occurred in the opposite direction,
can register changes in muscle length that are as would be expected if the perception of the Ia
not caused by active shortening (e.g., from an afferent output from the triceps muscle were
unexpected external source). being disrupted.) Such evidence supports the
Another problem was a lack of strong evi- idea that the muscle spindle provides informa-
dence that the Ia afferent fibers actually send tion about limb position and velocity (Sittig, 1986;
156 111 Motor Control and Learning

Sittig et al., 1985a, 1985b)-quite a different view Joint Receptors

from that held earlier. There is still some question
Each of the joints of the various limbs is sur-
whether or not the spindle is sufficiently sensi-
tive to detect small positional changes, and thus rounded by a sheath called a joint capsule, which
it may be only one of a number of sources for is primarily responsible for holding the lubricating
fluid for the joint. Embedded within the joint cap-
detecting position (see Kelso & Stelmach, 1976).
sules are different kinds of receptor cells known
Golgi Tendon Organs as the joint receptors. They are located primarily on
The other receptor for muscle information is the the parts of the joint capsule that are stretched the
Golgi tendon organ; these organs are tiny recep- most when the joint is moved, which originally
tors located in the junction where the muscle led investigators to believe that these receptors
"blends into" the tendon (figure 5.13b). They were involved in the perception of joint position.
seem to be ideally located to provide information Studying the cat hindlimb, Skoglund (1956) found
about muscle tension because they lie in series individual receptors that were active at very spe-
with (i.e., between) the force-producing contrac- cific locations in the range of limb movement (e.g.,
tile elements in the muscle and the tendon that from 150° to 180° of joint angle for a particular
attaches the muscle to the bone. cell). Another receptor would fire at a different set
The Golgi tendon organ has been shown to of joint angles, and so on. Presumably, the central
produce an inhibition of the muscle in which it nervous system could "know" the joint angles by
is located, so that a stretch to the active muscle detecting which of the joint receptors were active.
would cause the same muscle to decrease its ten- These conclusions have been seriously chal-
sion somewhat. Also, the early finding that a very lenged, however (see Kelso & Stelmach, 1976, for
large stretch of the muscle (near physiological a review). A number of investigators (e.g., Bur-
limits) appeared to be required to induce activity gess & Clark, 1969) have found that only a small
in the Golgi tendon organ led to speculation that proportion of the joint receptors fire at specific
the sensor was primarily a protective device that angles; rather, most of the joint receptors tend to
would prevent the muscle from contracting so fire near the extremes of joint movement. Further,
forcefully that a tendon would rupture. other researchers have found that the nature of
However, the work of Houk and Henneman the firing pattern is dependent on whether the
(1967) and Stuart (e.g., Stuart, Mosher, Gerlack, & movement is active or passive (Boyd & Roberts,
Reinking, 1972) has provided a different picture of 1953) and on the direction of motion of the joint
the functioning of the Golgi tendon organ. First, (see Smith, 1977). The fact that the firing pattern
anatomical evidence revealed that each organ is of the joint receptors is dependent on factors
attached (in series) to only a small group of from other than the simple position of the limb has
3 to 25 muscle fibers-not to the entire muscle as dimmed enthusiasm for the hypothesis that the
had been suspected. Thus, the various receptors joint receptors are the means by which the system
were sensing forces produced in different parts senses joint position.
of the muscle. Moreover, only a few (up to 15)
different motor units5 were represented in the Cutaneous Receptors
muscle fibers attached to a single tendon organ, Other receptors related to movement perception
so that the tendon organ now appeared to be in are located in various places in the skin. Although
a very good position to sense the tensions pro- such receptors can signal many separate states
duced in a limited number of individual motor of the body, such as pain, pressure, heat, cold, or
units-not from the whole muscle. This work has chemical stimuli, the important ones for move-
also shown, contrary to earlier beliefs, that the ment control are those receptors that signal infor-
tendon organs can respond to forces of less than mation about touch and, to some extent, deep
0.1 g (Houk & Henneman, 1967). Such evidence pressure. In addition, pain sensations certainly
suggests that the Golgi tendon organs are very constitute important information for specific
sensitive detectors for active tension in localized kinds of movement behaviors.
portions of a muscle, in addition to having the Different kinds of cutaneous receptors exist:
well-known protective function noted earlier (see some close to the surface and others much deeper,
Jami, 1992, for further discussion). some in glabrous (hairless) skin and others
 Sensory Contributions to Motor Control m 157

particular to hairy skin. One of these, called the spinal cord at each segment. Each segment serves
Pacinian corpuscle, is located deep in the skin and a particular region of the body.
is stimulated by deep deformation such as would
be produced by a blow or heavy pressure. Other Ensemble Characteristics
kinds of receptors in the skin include the Meissner Proprioception enables us to tell with remarkable
corpuscles, Merkel's discs, Ruffini's corpuscles, accuracy where our limbs are and how they are
and "free nerve endings." The last provide espe- acting, but how do the various receptors men-
cially strong signals when hairs on the body are tioned in the previous sections contribute to our
deformed by light touch, as they are located close motor control capabilities? An important concept
to the hair follicles. Near the surface of glabrous is that any one of the receptors in isolation from
skin (such as that of the lips and the palms of the the others is generally ineffective in signaling
hands) is a particularly strong concentration of information about the movements of the body.
Meissner corpuscles and Merkel' s discs. The fin- This is so because the various receptors are often
gertips have one of the highest concentrations of sensitive to a variety of aspects of body motion
cutaneous receptors on the body, which provide at the same time. For example, the Golgi tendon
information about the surfaces of objects through organs probably cannot signal information about
touch (Martin & Jessell, 1991). movement, because they cannot differentiate
between the forces produced in a static contraction
Input to the Central Nervous
and the same forces produced when the limb is
System moving. The spindle is sensitive to muscle length,
Input from the various receptors comes together but it is also sensitive to the rate of change in
in the periphery into spinal nerves, collections of length (velocity) and to the activity in the intra-
individual neurons (both sensory, or afferent, and fusal fibers that are active during active contrac-
motor, or efferent) that carry information toward tions; so the spindle confounds information about
and away from the spinal cord. These nerves position of the limb and the level of contraction
branch into two roots near the cord, called the of the muscles (force). And the joint receptors are
dorsal roots (posterior, or back) and ventral roots sensitive to joint position, but their output can be
(anterior, or front), where they contact the spinal affected by the tensions applied and by the direc-
cord separately. At this point, there is almost tion of movement, or by whether the movement
complete division of the neurons into afferent (or is active or passive (Paillard & Bruchon, 1968).
sensory) neurons that enter via the dorsal roots As a solution to this problem, many have sug-
and efferent (or motor) neurons that leave the gested that the central nervous system combines
cord via the ventral roots. Once inside the cord, and integrates information in some way to resolve
the afferent neurons can either synapse (connect) the kind of ambiguity in the signals produced by
with other neurons whose cell bodies are in the any one of these receptors (e.g., Wetzel & Stuart,
central gray matter, or travel to higher or lower 1976). Producing an ensemble (meaning a group-
levels in the cord or to the brain in one of the ing together) of information by combining the
many tracts that form the white matter adjacent various separate sources could enable the genera-
to the gray matter. tion of less ambiguous information about move-
The major pathways for transmitting signals ment (Gandevia & Burke, 1992). How the central
from the periphery to the brain are the spinal nervous system does this, and which sources of
tracts, located alongside the vertebrae that make information are most strongly represented in
up the spinal column. There are 8 cervical, 12 tho- which situations, are important questions for
racic, 5 lumbar, and 5 sacral vertebrae, defining the future. It is easy to imagine how all or most
a number of segments of the spinal cord. Except of these sensory inputs would be involved in a
for the input from the structures in the head and skill such as that shown in figure 5.14. We tum
neck (for our purposes here, mainly from the eyes, now to the ways in which these various sources
ears, and vestibular apparatus, entering through of sensory information contribute to movement
one or more of the 12 cranial nerves), the input to control (see the various chapters in the volume
the central nervous system is through the dorsal by Kandell, Schwartz, & Jessell, 2000, for a good
roots, which collect and guide the input to the tutorial review).
158 m; Motor Control and Learning

discussed in the previous two chapters. This

idea is illustrated in figure 5.15. An original com-
mand for action, such as an external stimulus
or an internal self-generated "go" signal, starts
the action by progressing through the stimulus-
identification, response-selection, and response-
prograrnrning stages, eventually leading to evoca-
tion of the movement commands to the muscles.
This portion of the closed-loop model is similar
to that involved in an open-loop system.
The difference becomes apparent, however,
when one considers the actions just subsequent to
this first aspect of the movement. First, a reference
of correctness is generated that will serve as the
standard against which the feedback from the
performance is judged. This reference of correct-
ness is also termed feedforward information (see
"When Can We Tickle Ourselves?" on p. 175). We
can think of this reference as a representation of
the feedback qualities associated with moving
as planned or as intended; it is analogous to the
value to which you set your thermostat in your
home heating system or to the compass heading
for the automatic pilot of a ship. The reference of
correctness represents the state of the feedback
associated with the correct movements of the
limbs during the intended action; it specifies
the sensory qualities of a goal. According to the
model presented in figure 5.15, muscle contrac-
tions cause the muscles, limbs, and body to
move, producing changes in the environment
FIGURE 5.14 An inverted iron cross move on the still rings-
as a result. Each of these actions generates
where optical flow and other sources of feedback are almost cer- information. The contracting muscles and the
tainly critical. movement of the body produce sensations from
the various receptor systems described earlier.
Then, via the reference of correctness, the system
Proprioception can compare the feedback it receives with the
and Motor Control feedback it expects to receive. If feedback from
the two sources is the same, the implication is
that the movement is correct and that no adjust-
The closed-loop ideas presented earlier regard-
ments are necessary. But if a difference exists
ing human performance have been considered in
between the feedback received and the refer-
various ways (see Krakauer & Ghez, 2000), but
ence, then an error is signaled and a correction is
one of the more common is to think of the closed-
required.
loop system as a system that contains the stages
of information processing and conscious decision
Influence of Movement Duration
making discussed in chapter 3. (Of course, it is
clear-to be discussed later in this chapter-that Closed-loop models such as the one in figure
there are other embedded closed-loop systems 5.15 are conceptualized in essentially two ways,
that do not involve consciousness.) It is useful depending on the nature of the motor skill. For
to consider the executive level of this system as rapid movements, feedback provides a basis for
consisting of the information-processing stages knowing (i.e., consciously) whether a movement
 Sensory Contributions to Motor Control Iii 159

lnput-------.l
Feedback error

Executive

Error

Feedback (Muscle sensations)

Feedback (Movement sensations)

Feedback (Environmental sensations)

FIGURE 5.15 The expanded closed-loop model for movement control.

produced was correct or not. For example, after Control of Long-Duration Movements
a golf ball is hit, the sensations from the swing The closed-loop model presented in figure 5.15
are available in memory for a brief period of time has been very useful for describing certain
and can be analyzed. A professional golfer prob- kinds of movements. The model seems to have
ably can tell a great deal about the direction and the most relevance for tasks that require a great
distance of the golf shot just from the feel and deal of time, because the processes involved
sound of it. A second way in which the closed- in the analysis of the error information are
loop ideas in figure 5.15 are used concerns the relatively slow. Also, the model best relates to
control of ongoing movements. These kinds of movements in which something is regulated at
models have obvious relevance to continuous some constant value, such as keeping the car at
skills, such as steering a car down a highway. a particular speed by monitoring the speedom-
Think of the reference as evaluating the set of eter or keeping an airplane on the proper glide
sensations associated with moving at a particu- path when guiding it onto the runway. These are
lar speed or with maintaining a certain distance called tracking movements (chapter 2), and they
behind another car. If one of these goals is not constitute an important class of motor behav-
met, the feedback received and the reference do ior. Tracking tasks have received considerable
not match, and an error is fed back to the execu- study, much of this work having been directed
tive level to compute a correction. Thus, these to problems in vehicle control, gunnery, and
closed-loop models view the control of a car on the like. An excellent review of the research on
a highway as a series of corrections that keep the tracking was provided by Poulton (1974); it has
vehicle safely on the road. been updated by Hammerton (1989), Wickens
160 Motor Control and Learning

and Hollands (2000), and Jagacinski and Flach tion of the limb at each point in space and at each
(2003). time in the movement is represented by a refer-
There are many different mathematical and ence of correctness, and the system can be made
physical models of tracking behavior; the major to track this set of positions across time to produce
differences relate to how the system uses feedback an action with a particular form. But the system
information and how the system initiates a cor- must process information very rapidly, even for
rection when errors are detected (see Jagacinski the simplest of movements. All these references of
& Flach, 2003, for an excellent review). The most correctness must be stored somewhere, which cre-
important generalization from this research ates difficulties when we realize that each point
is that if the models are used in computer or will be different if the movement begins from a
mechanical simulations of the human (in which the slightly different place or if it is to take a slightly
device is controlled in ways analogous to those different pathway through space.
in figure 5.15), these nonliving devices seem to Engineers have generally found these methods
"come alive" to behave in ways nearly indistin- very inefficient for machine (robot) control; such
guishable from their human counterparts. For findings have led many motor behavior research-
example, when we perform a laboratory tracking ers (see Greene, 1972; Kelso, 1995; Turvey, 1977)
task, approximately 200 ms elapses between the away from these kinds of control processes to
appearance of an error and the initiation of a cor- explain human skills. But there is still the possi-
rection back toward the center of the track. Such bility that the system might operate in this way at
lags and the character of the correction can be certain times or for certain skills that demand very
modeled very well, and the statistical agreement high precision (e.g., threading a needle or slicing a
between the actual and simulated movements is loaf of bread). Also, such a mechanism might serve
quite good for this kind of task. This evidence as the basis for recognizing errors at various places
does not prove that humans actually track this in the movement as it is carried out, without actu-
way, but the agreement between theoretical pre- ally being the basis for controlling it. After a tennis
dictions and data is very strong, and alternative stroke the performer could say that the elbow was
theories cannot boast of similar success. bent too much on the backswing, and thus have the
basis for making a correction in the movement on
Changing the Reference of Correctness the next attempt. Finally, there is ample evidence
We can extend this idea somewhat (as Adams, that the system makes use of reflexive mechanisms
1971, 1976a, 1977, has done) to account for how (i.e., without using the information-processing
the individual makes a limb movement such stages), mechanical mechanisms (Bizzi, Accornero,
as would be involved in sawing a board or in Chapple, & Hogan, 1982), or both to hold itself on
reaching for a mug of beer. Here, the reference of the proper track; these possibilities are discussed
correctness is not a single state as in the earlier later in the chapter.
examples, but rather a set of states that changes A compromise view is that only certain posi-
at each moment in time. Because the reference is tions in the movement are represented by refer-
constantly changing, it can be matched against ences of correctness. One view is that feedback
the feedback from the moving limb, which is from the movement when it is at its end point is
also changing as the movement progresses, so checked against a reference of correctness; then
that errors in the movement's trajectory can be corrections are initiated to move the limb to the
detected and corrected. This kind of mechanism proper position if it is not already there. These
is the basis for Adams' (1971) theory of learning, views of motor control hold that the limb is more
according to which the subject learns a set of refer- or less "thrown" in the direction of the end point
ences of correctness that the closed-loop system by some kind of open-loop control, and that the
is to "track" during the movement. We will have limb then "homes in on" the target by closed-
more to say about Adams' ideas later in the book loop control (a view reminiscent of Woodworth,
when we discuss learning theory (chapter 13). 1899; see "R.S. Woodworth on Manual Aiming"
But these kinds of models have serious limi- on p. 143). Here, the actual trajectory of the limb
tations. Engineers can design robots and other is determined by how the limb is "thrown," in
machines to behave in this way, using what they combination with mechanical factors such as
call point-to-point computation methods. The posi- gravity, friction, and muscle forces. In this view,
 Sensory Contributions to Motor Control Ill 161

the trajectory is not determined by point-to-point produced by conscious information-processing

computation as would be explained by a purely mechanisms, without considering the idea that
closed-loop system. the central nervous system contains closed-loop
mechanisms that do not require any attention.
Control of Rapid Movements
Many examples are possible, such as the control of
One of the most important points to have emerged body temperature and the regulation of breathing
from the evidence presented in chapters 3 and 4 during sleep. In this section we discuss evidence
was that the information-processing mechanisms, that these nonconscious mechanisms are involved
which lie at the very heart of the closed-loop in the control of voluntary movements as well.
system in figure 5.15, require a great deal of time
in order for stimuli to be processed to yield a Latencies of Corrections
response. So far we have assumed that each error An experiment by Dewhurst (1967) is representa-
signal the system receives must be processed in tive of a number of studies on this problem. The
these stages and that the response (a correction) subject was asked to hold the elbow at a right
can follow only after all the stages of processing angle to support a light weight attached to the
have been completed. Thus, a correction is seen in hand. The subject could monitor the performance
the same way as any other response to a stimulus. through vision of a display that signaled elbow
It requires a great deal of time and attention. angle. The experimenter recorded the position of
But there is a problem. In the closed-loop the arm together with the rectified electromyo-
models such as that shown in figure 5.15, rapid graphical (EMG) activity in the biceps muscle as
actions do not provide sufficient time for the the subject performed. Unexpectedly the weight
system to (a) generate an error, (b) detect the error, attached to the hand was then increased, and nat-
(c) determine the correction, (d) initiate the correc- urally the hand began to move downward. After
tion, and (e) correct the movement before a rapid a brief period, the subject increased the EMG
movement is completed. The left jab of the great activity to the biceps muscle, which increased
boxer, Muhammad Ali, is a good example. The its force output and brought the limb back to
movement itself was about 40 ms; yet, according the right-angle position. Given that the lowered
to our discussion earlier in this chapter, visually arm represents an error relative to the goal (of
detecting an aiming error and correcting it during keeping the arm at a right angle), how much
the same movement should require about 100 time will elapse before a correction is made, as
to 150 ms-the time necessary to complete the would be seen by an increase in the elbow flexors'
activities of the stages of information processing. EMG? Note that if the subject must process the
The movement is finished before the correction visual or kinesthetic feedback (or both) from the
can begin. For this reason, the closed-loop models arm through the information-processing stages,
of movement behavior do not seem well suited there should be no change in the biceps EMG for
to explaining rapid movements. approximately 150 to 200 ms.
Chapter 6 will raise this and other limitations Figure 5.16 shows the essential results. The
to the closed-loop models again. For now, suffice weight was added at the point in time indicated
it to say that the closed-loop mechanisms involv- by the arrow, and the limb began to move down-
ing the stages of processing appear to have a ward immediately. The records show a small
very difficult time explaining rapid movements. burst of EMG about 35 to 40 ms after the weight
Because these models have much credibility with was added and a larger irregular burst beginning
respect to very slow movement and posture and about 50 ms afterward. Just after this second burst
have little with respect to rapid movement, it is of EMG, the limb began to move back to the target
possible that there are essentially two fundamen- position. This change in EMG represents a clear
tally different kinds of movements: fast and slow. correction for the added weight, yet this correc-
We return to this distinction in chapter 6. tion was initiated far more quickly than can be
explained by a closed-loop process that requires
Reflexive Closed-Loop Control information processing as shown in figure 5.15.
In considering closed-loop control of movement, Rather, the correction is thought to be due to the
we dealt only with the kind of closed-loop model operation of reflexes in lower, probably spinal,
in which the determination of the correction was levels in the central nervous system.
162 Motor Control and Learning

Up

c:
~
'iii
0
Q.

E
~

Down

-50 0 50 100 150 200 250 300

t Time (ms)

FIGURE 5.16 Movement and rectified electromyogram record showing the latencies of two reflex-based corrections.
Reprinted, by permission, from D.J. Dewhurst, 1967, "Neuromuscular control system,'' IEEE Transactions on Biomedical Engineering 14: 170, © 1967 IEEE.

Consciousness and Reflexive handle. When the handle pushed against the
Corrections subject, the correct response was to "ease up,"
Another aspect of reflexive corrections for errors, so that the pressure between the handle and the
aside from their apparent rapidity, is that they subject's hand was held constant. In the second
might not require attention as other corrections condition, the subject was to compensate for the
seem to. Evidence for this notion was provided changing pressures exerted by the handle so that
in a study by Henry (1953), in which subjects had a constant position of the handle was maintained,
to regulate the force they applied to a handle. but with the exertion of constantly changing
The basic arrangement is shown in figure 5.17. pressure. A third condition was used to assess
The standing subject (blindfolded in the experi- the conscious perception of change; the subject
ment) is pushing against a handle attached to a attempted to hold the arm immobile, reporting
mechanical device that could alter the position of through a left-finger movement when a change in
the handle continuously. The arrangement was the pressure exerted by the apparatus was sensed.
such that if the subject was not pressing against The pressure changes were different for different
the handle, the handle would move forward and segments in the testing period, and Henry could
backward unpredictably. But a spring had been obtain an estimate of the amount of change that
placed between the machine and the handle, so was required for conscious perception of change.
that by modulating the force produced at each Henry (1953) found that the "threshold" force
moment, the subject could maintain the handle needed for an appropriate adjustment depended
in a constant position. strongly on what the subject was asked to control.
Henry used three conditions. In one condition, When the subject was asked to report a conscious
the subject's task was to keep the pressure against change, a force of 0.559 dynes was required for
the handle fixed by varying the position of the detection. But when the subject was asked to hold
 Sensory Contributions to Motor Control 163

Spring
1-----~f\f\NVV\NV ~-_,,

To recorder

FIGURE 5.17 Apparatus and general arrangement in Henry's experiment.

Reprinted with permission from Research Quarterly for Exercise and Sport, vol. 24, pg. 177. Copyright© 1953 by the American Alliance for Health, Physical Education,
Recreation and Dance, 1990 Association Drive, Reston, VA 20191.

the position constant, a force change of only 0.296 Experiments like the ones just described show
dynes was successfully detected and responded two important things about movement control.
to; apparently, in this constant-position condition, First, studies like Dewhurst' s (1967; see also Houk
subjects were responding to forces too small to & Rymer, 1981; see Lee, Murphy, & Tatton, 1983,
be detected consciously. Even more striking was for a review) demonstrate that the corrections for
the finding that position changes associated with suddenly presented changes in position can be
a force change of only 0.029 dynes produced suc- initiated with correction latencies of 30 to 80 ms-
cessful modifications in movement control. Thus, far more rapidly than would be expected based
the motor system, in holding a constant position on conscious-processing estimates. This kind of
against the apparatus, could respond to a change result suggests that the information-processing
that was considerably less than the change neces- stages, at least as shown in figure 5.15, are not
sary for conscious awareness (i.e., 0.029 vs. 0.559 involved in these actions. Second, the data from
dynes). In the constant-pressure condition-and Henry (1953) and others show that subjects can
particularly in the constant-position condition- make adjustments for changes in position-and
the motor system was responding to stimuli perhaps for changes in tension-so small that the
that were too small to be detected consciously. subject cannot even perceive them consciously.
As compared to the change required to adjust These data also indicate that the stages of infor-
to a change in position, the change had to be 19 mation processing are not involved, because at
times larger for conscious detection. These small least some of these stages are thought to entail
adjustments were apparently made without the conscious processing and attention in the sense
subject's awareness. discussed in chapter 4. Both of these lines of
164 ill Motor Control and Learning

evidence suggest that these kinds of corrections Cortcospinal (and other)

descending pathways
are produced via reflexive mechanisms that do
not concern the stages of processing. 11
_I I
I I
I I
Muscle Spindles and the Gamma Loop : I
The mechanisms responsible for the effects just
described probably involve the muscle spindle,
the small, cigar-shaped structure located between Spinal
cord
and in parallel to the main fibers of the skeletal
muscles (review figure 5.13a). The (simplified) la afferent
neurological connections of the spindle to the
spinal cord are illustrated in figure 5.18. Recall
from figure 5.13a and associated text that the
--Alpha
muscle spindles are specialized to detect changes motor neuron
in muscle length. The tension on the equatorial
(sensory) region, and thus the output of the Ia
afferent neurons, is adjusted via the gamma motor
system (and intrafusal muscle fibers) for contrac-
tions of the main muscle. Thus, at essentially any
point in the action's trajectory, the muscle spindles Extensor
are set to respond to a muscle length that is dif- muscle
ferent from that anticipated (or intended). Such
changes in the muscle length could be caused by
unexpected loads applied to (or removed from)
the limb's movement, or by the inherent vari-
ability of the main muscle's contractions (see the
discussion of force variability in chapter 7), or
both. Thus, it appears that the muscle spindles and FIGURE 5.18 Simplified diagram of the alpha and
gamma motor system connections.
the gamma system provide a means for correcting
Reprinted from K. Pearson and J. Gordon, 2000, Spinal reflexes. In Principles of
small changes in the external loads-and that such neural science, 4th ed., edited by E.R. Kandel, J.H. Schwartz, and T.M. Jessell
changes are fast, are not mediated by the stages of (New York: McGraw-Hill), 722. With permission o!The McGraw-Hill Company.

processing in figure 5.15), and are automatic in the

sense that they do not require attention.
The information from the Ia afferent is sent "hammer" just below the kneecap with the knee
to several places: From figure 5.18, we see that flexed. The resulting "knee-jerk reflex" is caused
the Ia afferent signals synapse with the alpha by the rapid stretch of the quadriceps muscle,
motor neurons in the same muscle to generate which stretches the spindles in the muscle. This
greater (or lesser) muscle force to compensate for increases the firing of the Ia afferent transmitted to
small changes in load. These signals are also sent the cord, increasing the alpha motor neuron firing
upward to various sensory regions in the brain rate, which causes the muscle to contract. This is
that could mediate more elaborate and complex an autogenetic reflex, because it causes an action in
corrections (e.g., the cortex, cerebellum, or both; the same muscle that was stimulated. The loop
Ghez & Thach, 2000). These compensations are time, or the time from the initial stretch until the
effective mainly for small changes in loads; we extrafusal fibers are increased in their activation,
all know that if we attempt to lift an "empty" is about 30 to 40 ms in humans. Because this 30 to
box that is actually full of bricks, we stop lifting 40 ms value corresponds with the latency for the
almost immediately and attempt the action again first burst of EMG shown in the Dewhurst (1967)
with a completely new pattern of action. experiment (see figure 5.16), this monosynaptic
This process is the basis of what has been called mechanism is probably responsible for this first
the monosynaptic stretch reflex. You experience an compensation for the added weight. This informa-
example of this reflex when a neurologist lightly tion may explain how Dewhurst' s (1967) subjects
strikes your patellar tendon with a small rubber responded so quickly to the added weight. It also
 Sensory Contributions to Motor Control Iii 165

helps to explain why Henry's (1953, figure 5.17) perturbation occurred at any location or time
subjects were able to make corrections for position in the thumb's cycle. Similar findings have
that they could not even perceive; some of the been shown in the breathing cycle by Sears and
corrections for changes in position occur in the Newsom-Davis (1968); when the resistance to
spinal cord, without the conscious involvement airflow was suddenly changed at various places
of the information-processing stages. in the cycle, the EMG in the intercostal muscles
Notice also that the reflex activity (figure 5.18) (which control the rib cage volume) increased
is more complicated than simply activating the in activity with a latency of 30 ms, regardless of
one muscle that was stretched. When the neu- when in the cycle the resistance to airflow was
rologist's hammer strikes the tendon, that muscle changed. Nashner and Woollacott (1979) have
is stretched, leading to increased firing in the Ia presented similar findings for the situation in
afferent fibers from its spindles. These Ia afferents which the ankle joint is unexpectedly perturbed
transmit their signal into the dorsal roots, where during normal walking. One interpretation of
the Ia synapses with several other neurons. First, these findings is that there is a reference of cor-
it synapses with the alpha motor neurons that rectness that "moves with" the overall move-
activate that muscle via the monosynaptic stretch ment so that, at any time in the movement, the
reflex. Second, the Ia synapses with other alpha limb's desired or planned location can be speci-
motor neurons (not shown) for muscles acting fied (Schmidt, 1976a). Thus, it appears that the
as synergists for this action (muscles oriented to muscle spindles-because they work in concert
aid in the action). Third, the Ia synapses with Ia with the alpha and gamma systems as described
inhibitory interneurons (not shown), which in earlier-are able to provide information about
tum synapses with the alpha motor neuron for the whether the limb is in a position different from
antagonist muscle (here, the triceps) and inhibits that specified, even during the execution of an
its action. This reflex-based inhibition of the action, and can induce rapid, automatic, reflex-
muscle acting in opposition to the muscle stimu- based corrections if it is not.
lated is called reciprocal inhibition. Thus, the stretch More recent research has added considerably
from striking the tendon with a rubber hammer to our understanding of these reflexive processes
has (a) increased activation in the muscle that was in movement control. There is considerable evi-
stretched (the agonist), (b) increased activation in dence that the reflexive response is not always
synergistic muscle(s), and (c) inhibited activity in autogenetic, in that the effects of a stimulus (such
the muscle acting as an antagonist-all rapidly, as a sudden stretch) can be seen in many of the
automatically, and without consciousness. Also, participating muscles-not just in the one(s)
if this had occurred in the legs, the stretch reflex stimulated (Pearson & Gordon, 2000b ). Thus,
(say, for right-knee flexion) would have produced these reflexive activities can be viewed as rapid,
increased activities in the left-knee extensors, and, automatic, involuntary, and coordinated responses
via reciprocal inhibition, decreased activity in the to disruptive stimuli.
left-knee flexors (see Pearson & Gordon, 2000b, for We can conceptualize the function of these
more details), presumably to allow the individual reflexes as directed at ensuring that the original
to provide postural support when the right knee goal of the action is achieved in the event of a per-
was flexed reflexively. turbation to it during execution. Consider a task in
This kind of reflexive adjustment is applicable which you are to repeatedly touch the tip of your
to situations other than those in which the limbs right thumb to the tip of your right index finger.
are being held in a static position. Numerous That is, both digit tips are moving toward the goal
investigations have shown that these processes of touching each other. Now, imagine the applica-
seem to keep an ongoing movement on course. tion of a perturbing force that resists (only) the
For example, Marsden, Merton, and Morton thumb's movement somewhat. We would expect
(1972) had subjects move the last joint of the automatic, rapid changes in the thumb muscula-
thumb back and forth in time to a metronome. ture-and this is what you see. But, in addition,
At an unpredictable time, the movement of the we will observe changes in the index finger mus-
thumb was resisted. The result was an additional culature that are coordinated with the changes in
burst of EMG within 30 ms of the perturbation. the thumb musculature; this happens even though
It is impressive that this occurred when the the index finger was not perturbed at all. One way
166 Motor Control and Learning

to explain such findings is that the motor system influence of the gamma loop just described. This is
has set up the goal of touching the two digit tips; conceptualized in figure 5.19, which shows a bone
the thumb's progress toward that goal is slowed in equilibrium being supported by two opposing
by the perturbation. So, the index finger's action muscles producing force (equal and opposite joint
is also adjusted (sped up), presumably so that the torques). If the system is perturbed, causing the
goal of digit touching is achieved (see Pearson & bone to move downward, the muscle on the right
Gordon, 2000b, for a more complete discussion). side of the diagram is lengthened slightly, caus-
Control of Muscle Stiffness We have seen how ing the stretch reflex described earlier. Perhaps
the muscle spindles are involved not only in the even more important, because the contracting
maintenance of static positions (e.g., posture), muscle is a "springy" substance, more tension
but also in maintaining the trajectories of various is produced in it by purely mechanical means as
limbs involved in a goal-directed action. But the its length increases. Furthermore, this mechani-
spindle seems also to be related to the control of cal change in tension is instantaneous, just as the
muscle stiffness, which is probably very important change in tension in a metal spring would be if
in the control of posture and other movements. it were stretched. Such increases in tension have
Stiffness, one of the measures used by engineers the effect of opposing the perturbation, bringing
(and others) to describe the characteristics of the system back to the original position. Nichols
elastic materials (e.g., springs), is defined in terms and Houk (1976) have provided evidence that the
of the amount of tension increase that is required muscle spindle is responsible for the maintenance
to increase the length of the object by a certain of muscle stiffness when the muscle is stretched,
amount (one unit). Engineers define stiffness so that the muscle can continue to act as a spring
more precisely as the change in tension divided in the control of posture and similar movements
by the resulting change in length, or the slope of (see also Houk, 1979; Houk & Rymer, 1981).
the length-tension relationship. If a spring is very Long-Loop Reflexes Have another look at
stiff, a great deal of tension is needed to increase figure 5.16. In addition to the monosynaptic
its length by a given amount; for a less stiff spring, reflex activity (the EMG activity that began with
less tension is required. This is important in our a latency of about 30 ms), there is another kind of
context because the muscle seems to provide a activity responsible for the more sustained burst
compliant (springy) interface between the per- that occurred about 50 ms after the weight was
former and the environment. added, and yet more starting at about 130 ms.
When we maintain posture, the muscles sup- This second burst at about 50 to 70 ms occurs too
porting the skeleton are contracting under the rapidly to be explained by the stages of informa-

Decrease in Increase in
tension recorded tension recorded
here'\ /here

FIGURE 5.19 Muscle-spring model with gauges for measuring tension in the "tendons" as a perturbation is applied.
 Sensory Contributions to Motor Control 167

tion processing, yet it is apparently too slow to constructed in this fashion, as there are situations
be accounted for by the monosynaptic stretch in which we must resist very strongly when per-
reflex. The early burst at 30 ms was very brief and turbations occur. Other situations arise in which
did not result in much actual increase in force, a very strong resistance would mean a serious
whereas the burst at 50 to 80 ms was larger, was accident; in skiing over bumps, for example,
more sustained, and probably resulted in the force failing to "let go" would result in a very stiff leg
changes necessary to move the limb back toward musculature when a very compliant (springy and
the horizontal position. The response at 130 ms is supple) one would be more desirable.
probably the RT response, mediated by the stages
of processing. The response to muscle stretch with Triggered Reactions
a latency of 50 to 70 ms has been termed the long- So far, we have seen three distinct processes lead-
loop reflex (also called the functional stretch reflex). ing to increased EMG in response to perturbations
The long-loop reflex fits neatly into the over- such as an added weight. These are the mono-
all picture of segmental limb control. When the synaptic stretch reflex (30 to 50 ms latency), the
spindle is stretched and the Ia afferent is increased long-loop reflex (or functional stretch reflex, 50-70
in its activation, the information is fed back to the ms), and of course the voluntary RT response dis-
spinal cord where it activates the alpha motor cussed in chapter 4 that begins at about 130 ms in
neuron and the same (and functionally related) Dewhurst's (figure 5.16) study. 6 Crago, Houk, and
muscle (review figure 5.18). This signal is also Hasan (1976) argue, however, that there is yet a
sent to higher segmental levels in the cord, to the fourth kind of response that falls between the 50
brain, or both. Information from the Ia afferent to 70 ms long-loop response and the voluntary
is integrated with other information in sensory RT latency. They call these responses triggered
and motor centers in the brain that can initiate a reactions: prestructured, coordinated reactions in
more complete or more complex response to the the same or in closely related musculature that
imposed stretch. Because the information travels are "triggered" into action by various receptors.
to a higher center to be organized, and because Such reactions have latencies from 80 to perhaps
more information needs to be considered and 200 ms, and the latencies are far more variable
more synapses are involved, the reflex requires than are those of the faster reflexes. Presumably
more time-50 to 70 ms rather than 30 to 40 ms the triggered reaction is like a "very fast RT," per-
for the monosynaptic reflex. haps bypassing some of the stages of information
Something appears to be regained, however, processing because the reaction to the stretch is
with the loss in time. First, the EMG activity stereotyped, predictable, and well practiced. The
from the long-loop reflex is far stronger (produc- performer does not have to spend much time in
ing more force change) than that involved in the processes like response selection and program-
monosynaptic stretch reflex. Second, because the ming, and the reaction is just "triggered off"
reflex is organized in a higher center, it is more almost as if it were automatic (see Schmidt, 1987).
flexible than the monosynaptic reflex. For example, What evidence is there for this kind of control?
Evarts (1973) has shown that, if the subject in a Crago and colleagues (1976) have shown that
task similar to Dewhurst's (1967) is told to "resist" portions of the response to an unexpected stretch
the stretch, a burst pattern like that in figure 5.16 perturbation were faster than RT (as we have
occurs. If the subject is told to "let go," so that seen before), but also that the latencies increased
when the weight is added the subject simply lets as the number of stimulus-response alternatives
her arm be moved by it, the second burst (presum- increased from one to two; here, the perturba-
ably due to the long-loop reflex) nearly disappears tion was (a) always a flexion (i.e., a one-choice
but the first burst remains, unaffected. It appears task) or (b) either a flexion or an extension (i.e., a
that prior instructions can change the reflexive two-choice task). We usually think of processes
response to a given stimulus (the added weight) so involved in resolving choice as being "located"
that the reaction is appropriate for the particular in a response-selection stage (Hick's law, chapter
situation. The monosynaptic reflex, residing at a 3); these results suggest that, unlike the mono-
very low level in the spinal cord, can probably synaptic stretch reflexes, the responses might be
not be modulated by prior instructions (at least mediated in some way by the stages of informa-
to any great degree). It is fortunate that we are tion processing. Perhaps some of the processes
168 111 Motor Control and Learning

are bypassed, leading to latencies shorter than finger gripping force, and with a latency of only
"normal" RT latencies. Table 5.1 summarizes about 80 ms (see also Cole & Abbs, 1988). These
these several processes. More research is clearly reflexes were fast enough to prevent a noticeable
needed to clarify these issues. movement of the object, and often the subject did
not even know that a slip had occurred. Some-
Tactile Feedback times, several of these slips and catches were seen
Recently, we have seen evidence of the initiation in a single lift of the object, each with very short
of reflex-like control by the cutaneous receptors latencies. In addition to the increase in gripping
in the skin. We have long known that the cutane- forces, there was at the same time a corresponding
ous receptors can be the trigger for withdrawal decrease in the EMG in the elbow flexors, as if the
responses-as in jerking your hand from a hot system were reducing the slippage by decreasing
surface (Johansson & Westling, 1991; Johansson the hand's upward acceleration. All this resulted
& Flanagan, 2009). in a beautifully coordinated response to the
The Wineglass Effect Suppose you are washing slipping, which was evidenced in a number of
dishes and raise an object to check its cleanliness- joints-not just in the structures directly affected
for example, holding the stem of an expensive by the stimulus. The reaction was very fast and
wineglass between your. fingertips. If the glass probably nonconscious, and seemed to have the
begins to tip or slip (because of the wet surface), overall "goal" of reorganizing the system slightly
a common reaction is to increase the grip of force to complete the action successfully (i.e., lifting
with your fingers to stop it. Using a laboratory without dropping). Thus, in this example, as well
analog of this basic idea, Johansson and Westling as in others mentioned here, the stimulus does
(1984, 1988, 1990; W~stling & Johansson, 1984; not (only) directly affect the muscles that make
see also Ghez & Kr&.er, 2000, and Johansson & the response as in the example from Dewhurst
Westling, 1991) have studied the motor reactions (1967); that is, the response to the stimulation is
to stimuli indicating loss of grip. They asked sub- not (simply) autogenetic, but rather is coordinated
jects to lift small objects (having various degrees across several muscles and joints.
of surface friction) between the thumb and index Object Manipulation In addition, it appears
finger and hoist them with an elbow flexion action. that cutaneous receptors can provide sensitive
The stimuli indicating that the object is slipping are feedback about the fingers' contact with objects
a set of tiny vibrations in the skin of the fingers, being manipulated. One example involves
which we detect through the cutaneous receptors. anesthetizing (via needle injections) the tips of
After the onset of a slip (which can be measured by the thumb, index, and middle finger. Before the
vibration sensors in apparatus), subjects showed injection, the subject could remove a wooden
an increase in the EMG in muscles responsible for match from a matchbox and light it by strik-
 Sensory Contributions to Motor Control 169

ing it against the side of the matchbox-a very

common, unremarkable action requiring about
8 s. However, after the anesthetic was injected,
which eliminated cutaneous feedback from the
fingertips temporarily, this task became very
difficult; the subject dropped matches several
times and produced very clumsy and "labored"
actions, eventually succeeding in lighting the
match in about 29 s. In addition, it appeared that
the subject made extensive use of visual feedback
after the injections, but this apparently did not
make up for the loss of tactile sensation, perhaps
because the visual feedback was far slower than
the reflex-like feedback from the cutaneous recep-
tors (Johansson & Flanagan, 2009). 7
Other Reflexes
Many reflexes can be elicited in humans, and
some scientists have suggested that these reflexes
play important roles in movement control. For
example, Fukuda (1961) collected photographs of
athletes, dancers, and other performers in specific
situations, such as the baseball player illustrated
in figure 5.20 who is jumping and stretching his
gloved hand high in the air to catch a ball. In this
example the positioning of the outstretched hand
and extended leg on the one side of the body,
together with the flexed arm and leg on the other
side of the body, closely resembles the tonic neck
reflex, which is often seen in infants. When the
head is turned to the left, the left arm becomes
extended and the right arm curls up alongside
the neck. Fukuda's observations, illustrated in
examples from baseball among others, perhaps
FIGURE 5.20 A baseball player reaching for a ball,
suggest that built-in reflex patterns might repre- showing a pattern resembling the tonic neck reflex.
sent basic components of various skilled actions Reproduced with permission from T Fukuda et al., Acta Oto-Laryngo/ogica, 1961;
in adult motor behavior. In support of Fukuda's 161: 1261-1263. ©1961 lnforma Healthcare. "Studies on human dynamic
postures from the viewpoint of postural reflexes."
argument, if subjects turned their head toward the
hand that was active in a force production task,
the force was increased as compared to when dence is lacking to link these reflexes as observed
the head was turned away (Hellebrandt, Houtz, in isolation and in childhood directly with the
Partridge, & Walters, 1956; Shea, Guadagnoli, & movement patterns seen in adults. A review of the
Dean, 1995). The interpretation was that turning possible role of reflexes in movement is provided
the head supposedly activates the tonic neck reflex by Easton (1972, 1978).
that provides a reflex-based facilitation of the
alpha motor neurons on both sides of the body. Speech Perturbations
However, in this baseball example (figure 5.20), Another example of triggered reactions was pro-
it is possible that the player is in this position vided by Abbs and Gracco (1983; Abbs, Gracco, &
simply because he is looking at and reaching for Cole, 1984) and by Kelso, Tuller, Vatikiotis-Bate-
the ball, not because the reflexes are producing son, and Fowler (1984). In Abbs' work, the sub-
the classic tonic neck pattern. Fukuda presents jects were asked to utter nonsense syllables such
a suggestive set of findings, but the crucial evi- as I afa/ and I aba/. Try I aba/, for example; the
170 111 Motor Control and Learning

lips must be open to make the initial vowel sound, ensuring that the original action is carried out
and then they must come together briefly to make faithfully even in the face of small perturbations.
the stop consonant /b I. In their experiments, This idea can be summarized from the flow-
Abbs and colleagues (1984) would occasionally chart in figure 5.21. Here, we attempt to show
perturb the lower jaw (during the utterance of . that many simultaneous feedback loops appear to
I aba/) with a small downward force pulse that be involved in one way or another in movement
prevented the lower lip from rising to make control. One of these-the outermost loop, shown
contact with the upper lip in its normal position, as a heavy line in figure 5.21-involves the stages
and the EMG from the musculature in the upper of processing, reaction time (M3 response), and
lip and its position were measured. When the attention as we argued in chapter 3. However, in
perturbation was applied, EMG increased in the the present chapter we have provided evidence
upper lip musculature, which moved the upper for various reflex-like processes that appear not
lip downward, all with a latency of about 25 to to involve the stages of processing. The inner-
70 ms. Notice that, as in the wineglass example, most loop is the M 1 response, probably the most
this is not an autogenetic response, because the rapid of all, traversing from an involved muscle
response to the stimulation does not occur (only) to the spinal cord and back to that muscle. Feed-
in the stimulated muscle and seems to be orga- back from cutaneous receptors in the fingertips
nized with the "purpose" of completing the action is involved in nonconscious control, and this
of making the proper sound (which required lip feedback can also be conceptualized as operating
closure for success). Furthermore, the reaction mainly nonconsciously, as we have discussed
seems to be dependent on practice, as it is difficult (Johansson & Flanagan, 2009). The M 2 response
to argue that making the particular sound I aba/ is also travels to the spinal cord, but extends as well
genetically determined; this feature of susceptibil- to somewhat higher levels than the M1 response.
ity to practice might be yet another distinguishing Triggered reactions arising from receptors in vari-
characteristic of triggered reactions. However, this ous muscles or from the environment can also be
effect is analogous to the so-called reflex-reversal thought of as providing input to some level of the
phenomenon in the study of locomotion; in loco- spinal cord. Finally, ambient vision can influence
motion, it is difficult to argue that learning plays (at least) balance nonconsciously, certainly not
a large role. In this situation, the same stimulus involving the stages of processing. 8
elicits one reflex when the leg is in the swing phase,
and a different reflex when the leg is in the stance
phase. See chapter 6 for a fuller discussion. Feedforward Influences
The Role of Reflexes in Skilled Actions on Motor Control
It is interesting to think about the possible role
of the various reflexes in the control of move- In this section we consider evidence that the
ment, and the evidence for such contributions is motor system operates with a feedforward control
mounting steadily. At this point, there is consider- mode. This term-coined to contrast with the
able evidence for the involvement of reflex-like concept of feedback-is usually defined as the
processes in voluntary motor behavior. We have sending of a signal that (a) readies the system
seen stretch reflexes, reflex reversals, cutaneous for the upcoming motor command or (b) read-
feedback effects, triggered reactions, and ambient ies the system for the receipt of some particular
vision effects. All of these have the property of kind of feedback information (or does both). Such
providing stereotyped (if sometimes somewhat processes appear to occur frequently, and we will
complex) responses to stimuli, and all with consider a few examples. Notice that the reference
latencies far shorter than our usual estimates of mechanisms illustrated in figures 5.15 and 5.21
reaction time. Thus, it is difficult to argue that the are examples of a feedforward process.
stages of processing (identification, selection, pro-
gramming; see chapter 3) have any role in these
Saccadic Eye Movements
reactions. This evidence, taken together, suggests Numerous situations exist in which the idea of
that these reactions are automatic (nonconscious) feedforward control appears to be involved in the
responses, which seem to have the property of production and evaluation of human behavior.
 Sensory Contributions to Motor Control 171

Error
M3

Desired state

Triggered reactions

Actual state
M2
Cutaneous c
0
feedback ·c;;
·;;;
M1 cQ)
i5
E
<(

Proprioceptive feedback

Exteroceptive feedback
Focal vision

FIGURE 5.21 Expanded conceptual model of closed-loop processes.

One of the earliest notions of feedforward control and others (e.g., Gallistel, 1980) was that the
concerned the mechanisms of visual perception visual perceptual system was informed about the
after the eye made a saccade-a very rapid, jerky upcoming movement of the eye ahead of time, so
movement of the eyes from one position to a new that the pattern of changed visual input could be
position. Of course, the light patterns falling on evaluated properly. This advance (feedforward)
the retina are different before, during, and after the information was termed corollary discharge, or
saccade. But how does the person know whether efference copy, by these authors (see also Evarts,
the eye moved in a stable world or whether the 1973, and Kelso, 1982, for reviews).
world moved, with the eye remaining stationary? The idea is that a "copy" of the motor (effer-
The pattern of stimulation (optical flow pattern) on ent) command to the eye muscles is also sent to
the retina could be exactly the same in both cases. some other location in the brain, where it is used
The answer suggested by von Holst (1954; to evaluate the incoming visual feedback signals
von Holst & Mittelstaedt, 1950), Sperry (1950), and to correct for" the fact that the image on
/1
172 Motor Control and Learning

the retina is about to move. Thus, the individual Preparatory Postural Reactions
perceives the environment as stable because she
"knows" that the eye has moved via a saccade. Consider a situation in which a standing subject
How such a system works is the subject of much awaits a signal to raise the arm (with a weight
debate, but many scientists have argued that in the hand) quickly from a relaxed position at
some such mechanism must exist in order for the the side to a position in front of the body, as if to
individual to interpret incoming visual signals point straight ahead. According to earlier discus-
correctly. sions about such actions, the commands for the
shoulder muscles are generated after an RT of
Efference Copy in Limb Control about 150 to 200 ms or so following an external
"go" signal. But if the subject is in a balanced
The efference copy mechanism appears to have stance, a sudden movement of the (loaded) arm
a parallel in the control or evaluation of limb forward and upward will cause a shift in the per-
movements. First, as pointed out by Evarts son's center of gravity, and he will lose balance
(1973), there is neurological evidence that infor-
unless some compensation is provided prior to
mation destined for the muscles is also sent to
or along with the movement itself. When is such
places in the brain that are primarily sensory in
compensation produced-before, during, or after
nature. Perhaps the purpose of such activities is
the action?
to "tell" the sensory system what was ordered
Belen'kii, Gurfinkel, and Pal'tsev (1967; see
by the motor system and to ready it for receipt
also Cardo & Nashner, 1982) recorded the EMG
of the feedback. Thus, the idea of efference copy
activity from the support muscles of the legs as
is much like the establishment of the reference
well as the prime-moving shoulder muscles in
of correctness against which the feedback sig-
nals will be compared. One component of this this rapid arm-raise task (performed as a reaction
feedforward must simply be the knowledge to a stimulus). After the stimulus came on, the
that the person moved voluntarily (as opposed first signs of EMG activity occurred in the large
to passively), so that the person can distinguish muscles in the back of the leg (biceps femoris) on
feedback from movement as due to active motion the opposite side of the body from the intended
versus passive motion. action-and these changes occurred about 60 ms
Related to this example is the well-known before any EMG was seen in the shoulder muscles.
heightened kinesthetic sensitivity when the sub- The actions of the EMGs in the legs could not
ject is moving actively versus passively (Brodie have been caused by an imbalance resulting
& Ross, 1985). Do this experiment. Take a few from the movement of the arm, because these
different grades of sandpaper, and rank them changes occurred before the first EMG changes
in terms of roughness by rubbing them with in the shoulder did and even longer before any
your index finger (eyes closed). First, move movement in the shoulder occurred. It is pos-
your finger actively over the surfaces. Then sible to consider these changes as an example of
have someone hold your finger and move it over feedforward control, in which the motor system
the surface in the same way, but without your sends commands to the spinal levels associated
active muscular involvement. You will likely with the leg musculature prior to the arm action;
find that your perception of roughness is much the purpose is to "prepare" the legs so that the
impaired when you are moved passively. Why? body does not lose balance when the arm finally
One answer is that when the motor system sends moves, or to ready the motor unit pools in the
the commands to move actively, it also sends spinal cord for the upcoming signal to contract
an efference copy of the commands to sensory (called spinal tuning), or both.
areas in the brain to enable proper evaluation Another, related view is that the changes in
of the feedback. But when the finger is moved the patterns of EMGs in the legs prior to action
passively, no motor commands are issued to the could be considered a part of the coordination of
muscles; hence there is no efference copy, and the the entire action, beginning with change in the
"same" feedback signals from the finger are not legs. W. Lee (1980), using this same task, found
perceived as accurately (for further evidence and that the temporal aspects of the EMGs are quite
discussion, see Lederman & Klatzky, 1997, 2009). closely linked, with the various muscles acting in
the same order and with a nearly constant pat-
 Sensory Contributions to Motor Control 1111 173

tern of action for various trials (see chapter 8 for reference indicated that there was going to be an
more evidence related to the processes involved error, a correction could be initiated and a new
in coordination). This supports the idea that the movement command could be sent. This has
feedforward information is an integral part of the advantage of initiating the correction much
the overall control system (discussed later in this more quickly than would be the case if feedback
chapter), or part of the motor program for action from the (errant) movement was evaluated; and
as we discuss in the next chapter. the correction could even be given before the
errant movement was initiated, or at least before
Detection and Correction of Errors the movement could "do much damage" in the
What role does feedforward information play environment.
in the detection and correction of errors? It is Various experiments appear to provide evi-
thought that this feedforward information is, first dence for such processes. For example, Angel and
of all, information that a movement was ordered. Higgins (1969) used a step tracking task in which
Also, this feedforward information presumably the target would move suddenly in discrete
contains the expected sensory consequences of steps to the left or right; the subject's task was
the movement just ordered. As an example, if to follow the target movement with appropriate
you ask a friend to call your home phone while limb movements. When the subject is highly moti-
you look on, then in some way you "know" that vated to minimize RT, she will move in the wrong
the phone at home is ringing even though you direction occasionally, reverse the move, and then
cannot hear it directly. Knowing that your friend move rapidly to the correct target. Figure 5.22 is a
has placed the call is almost as reliable as hearing diagram of a typical corrected trial, showing the
the telephone ring. In this way, feedforward infor- incorrect initial movement (beginning at point B)
mation can be thought of as a variant of feedback. and the subsequent correction (beginning at point
If feedforward (or efference copy) could, in C). Interestingly, the correction times, measured
this general way, be evaluated as a kind of sen- as the interval from the beginning of the incorrect
sory information, then the idea is that it can be movement until the beginning of the correction
evaluated against a reference of correctness just (from B to C in the figure), were as short as 90
as feedback can be evaluated. If the subsequent ms. The subjects could not have been process-
analysis of the efference copy in relation to this ing movement feedback from proprioception or

FIGURE 5.22 Position (top) and velocity (center) traces showing an error with "rapid" correction toward the target.
Reprinted, by permission, from R.W. Angell and J.R. Higgins, 1969, "Correction of false moves in pursuit tracking;' Journal of Experimental Psychology 82: 186. Copyright© 1969
by the American Psychological Association.
174 Motor Control and Learning

vision in the stages of processing, because the Forward Models of Motor Control
correction for the false move was made more
rapidly than could be accounted for by the usual Consider the simple example of being tickled.
feedback-processing mechanisms. Perhaps the Most of us have body parts that are responsive to
subjects were using a central feedback loop based touch and that result in mild to extreme sensations
on efference copy as a basis for initiating these when brushed lightly. With only a few exceptions
corrections. (see "When Can We Tickle Ourselves?"), those
We discussed these so-called "rapid error cor- responses are completely avoided when the
rections" earlier (in chapter 3). But, the hypothesis same tickling movements are produced by our
of eff~rence copy-based "rapid" corrections just own actions. Why? A class of theories of motor
mentioned can also be explained in other ways. control known as forward models provides a fairly
For example, when the subject moves, he feeds simple explanation of why we are unable to tickle
forward the expected sensory consequences of the ourselves (for overviews see Desmurget & Graf-
display. That is, if the subject guesses "left," he will ton, 2000; Shadmehr & Krakauer, 2008; Wolpert,
feed forward the expected sensory (visual) con- Ghahramani, & Flanagan, 2001; Wolpert, Miall, &
sequences of the left signal coming on, and then Kawato, 1998). The basic idea is that when a limb
trigger off the (incorrect) action with a very fast movement is initiated, a "model" of the predicted
RT (152 ms). However, if the proper response is consequences of the action is also generated. This
"right" on that trial, then what the subject expects is_ c~lled a "forward" model because it is a pre-
to se~ (the left signal being on) is not matched by diction of the movement's sensory consequences
the signal that actually appeared (the right signal). before the actual sensory consequences have been
This mismatch is the "stimulus" indicating that generated. If the actual sensory consequences
the response will surely be an error, and corrective match the expected consequences (as they would
if we try to tickle ourselves), this evokes no tick-
actions to move left are initiated then, using the
ling response. But, when someone else tickles us,
stages of processing in the "usual" way. If so, the
latency of the onset of the correction is not really the actual sensory consequences do not match
the expected consequences, because the expected
very "rapid." In the Schmidt-Gordon data (see
consequences of the (other person's movement)
figure 3.17 on p. 84), this correction had a latency
do not exist for comparison, and the result is the
of 276 ms (from signal onset to reversal in direc-
tickling sensation.
tion, or from A to C in figure 5.22). Thus, in this
In other kinds of actions, once the movement
interpretation, these "rapid" corrections are not
is begun, the predicted and actual consequences
rapid at all, and we need not invoke efference copy
are compared and corrections for discrepancies
explanations to explain them.
between them can be initiated with minimal time
The notion of efference copy also has the dif-
delays. Thus, forward models of motor control
ficulty of explaining how efferent (or motor)
avoid many of the processing delays that are
~nformat~on can be compared against sensory
implicit in traditional feedback-based theories of
information (see Schmidt & Gordon, 1977). The
control. And, unlike feedforward-based theories
problem is that the "language" of the motor
forward models have the advantage of usin~
information is in terms of which muscles to con-
actual feedback in error correction processes to
tract, when, in what order and timing, and with
avoid the problem of motor and sensory coding
what force (discussed further in chapter 6). The
states as mentioned in the previous section.
sensory information arising from the action is in
the form of limb movement, joint positions, and
muscle forGes. So, how can the system compare Summary
the information about motor commands with the
information about movement feedback if these Closed-loop systems involve the comparison of
signals are in different codes? For several authors feedback against a reference of correctness, the
this problem requires that the feedforward determination of an error, and a subsequent cor-
~nformation be in the same code as the sensory rection. The receptors for the feedback supplied
information, which is the basis for the reference to closed-loop systems are the eyes, ears, and
of correctness seen in figure 5.15 (see Schmidt, vestibular apparatus, as well as the Golgi tendon
1975b) and a class of models discussed next.
 Sensory Contributions to Motor Control 175

Most of us have certain areas of our bodies that are particularly swscEiotible
of these areas results in a mixture of feelings that often includes
Perhaps somewhat surprisingly, the act of tickling has become a study of interest
scientists. The reason is simple: Although we respondto tickling by someone else,
when we attempt to tickle ourselves (Claxton, 1975; Weiskrantz, Elliott, &Darlington,
this is done under special circumstances.Why is this so?
Some researchers suggestthat the inability to tickle ourselves is relater.I to nn::idi•r::tii,,rfFprooes~;es.
involved in motor control. According to Blakemore, Wolpert, and
ing a movement that would normally stimulate a ticklish part
the sensory consequences of the action. This expectation has the
that would normally arise if the area is stimulated by the subject, but cannot
that arise when the subject is stimulated by another agent. Support for th.is comes
functional magnetic resonance imaging (fMRI); brain activity during tickles shows IA<::~:... nti'11~t1nr1
under self-generated than externally generated tickles (Blakemore et al., 2000).
In another study, however, Blakemore, Frith, and Wolpert ( 1999) employed a robotic device that
could be used to generate a tickling sensation externally or could be •. manipulated bythe J11~ividual.
As expected, higher ratings of "tickliness" were found on external- than onself~generatedUckleJrials,
even when the machine provided the tickles. However, in some trials the robot did not executeJhe
movement until a brief period of time (100, 200or 300 ms) had elaps~d·after thesubjecthadcornc
manded the tickle. The 100 ms lag resulted in a greaterfeelingof ticklinesscompared toimrn~dlately
experienced selHickles. Moreover, lags of 200 and 300 ms produced as much sensationoftickliness
as externally produced tickles. One explanation forthis finding was that the predictive inforrnationfed
forward in expectation of certain sensory consequences had a rather short "shelflife," essentially as
short-term {visual) sensory store (STSS) does, as discussed in chapter 3: Perhapsthi'$ rapid decay
in memory quality is characteristic of feedforward information generally. Other explanations are
equally plausible, however, and further research is warranted to sort out this interesting phenprnenon.

organs, the muscle spindles, the joint receptors, different kinds of movements exist: slow and
and touch receptors in various places in the skin. fast. However, strong evidence exists for closed-
All these sources provide input to the central loop reflexive control in limb movements. Most
nervous system, and then the information is of this work suggests involvement of the muscle
presumably combined for the purpose of analysis spindle and the gamma loop, but other receptors
of movement. are involved as well. Such reflexive corrections
Vision provides the richest source of informa- can be classified as (a) the monosynaptic stretch
tion for closed-loop control. Vision can be used reflex (latency= 30-50 ms), (b) the long-loop or
in a variety of ways, providing information about transcortical (or functional) stretch reflex (latency
errors in movement as well as predictive infor- = 50-80 ms), (c) the triggered reaction (latency=
mation so that potential errors can be anticipated 80-120 ms), and (d) RT (latency= 120-180 ms or
and avoided. Closed-loop control models seem to longer).
have their greatest strength in explaining move- Feedforward control models involve the
ments that are slow in time or that have very delivery of information to some other part of
high accuracy requirements (or both). Tracking the system to "prepare it" for incoming sensory
tasks are most obviously related to closed-loop information or for an upcoming motor command
processes. These closed-loop models have dif- (as in spinal tuning). Thus, feedforward informa-
ficulty explaining the kinds of corrections seen tion serves an important role in error detection
in very rapid movements, however, and this fact and correction, often occurring in anticipation
leads to the suggestion that two fundamentally of the error.
176 Motor Control and Learning

3 Performance in the various vision conditions may

Student Assignments have reached a ceiling level. Thus, direct comparison to
performance in the control condition with full room lighting
1. Prepare to answer the following questions is problematic because of the richer sources of contextual
during class discussion: cues provided, which appear to be particularly important
for reducing spatial errors (Montagne & Laurent, 1994).
a. Describe the closed-loop operation of
Whiting and colleagues (1970) also noted that subjects used
any human-made device of daily living a different catching strategy in the full room lighting condi-
(e.g., a furnace). tion whereby they would move with the ball and delay their
b. Describe the anatomical parts and attempts to make the catch, thereby gaining themselves an
what each contributes to the stretch additional 100 ms or so of ball flight information.
4 The root fusal means fusiform or spindle-shaped; so
reflex.
intrafusal fibers are muscle fibers within the spindle, and
c. What are the psychophysiological cues
the extrafusal fibers are those outside the spindle-that is,
that humans use for depth perception? the fibers of the muscle in which the spindle is embedded.
2. Find a recent research article (within the The Greek letter gamma("/) refers to the spindle system
past five years) that compares the effect of (the intrafusal fibers are thus innervated by the gamma
a visual illusion on perception and motor motor neurons; the alpha motor neurons innervate the
extrafusal fibers). The term Ia refers to the fact that the
control.
sensory (afferent) fiber emerging from the spindle is a
large type I afferent; the a refers to the fact that this fiber
comes from the spindle (type lb fibers come from the
Web Resources Golgi tendon organs). Type II afferents are smaller in
diameter and conduct impulses more slowly than the
How the sensory systems function: type I afferents.
5 A motor unit is defined as an alpha motor neuron and
http://pathology.mc.duke.edu/neuropath/nawr/ all of the muscle fibers that it innervates. In humans, the
sensory.html number of fibers supplied by one alpha motor neuron
A gallery of visual illusions: might vary from a few (in muscles requiring fine con-
trol-in the hand, larynx, eyes) up to several thousand
http://dragon.uml.edu/psych/illusion.html (in muscles requiring only gross control-in the trunk).
There could be from a few to several hundred motor units
Basic information about the brain and central in any one muscle (see Leob & Ghez, 2000).
nervous system: 6 This 120 ms value in figure 5.16 is considerably

shorter than the typical RT latency discussed in chapters

http://faculty.washington.edu/chudler/introb. 3 and 4. But in figure 5.16, the latency is measured by
html the EMG change, whereas in chapters 3 and 4 the RT is
measured with respect to the movement, which usually
occurs with an additional delay of at least 50 ms.
Notes 7A video of this demonstration is available at www
.hp-research.com/ videos.shtml.
1 Anatomical and neurophysiological evidence that 8 This diagram is not meant to be correct anatomically.
supports the two visual stream proposal is reviewed by For example, visual information probably travels to high
Norman (2002) and Bear, Connors, and Paradiso (2001, levels in the central nervous system (visual cortex) and is
chapter 10). certainly not "housed" in the spinal cord, per se. Rather,
2 Some authors refer to this illusion as the Ebbinghaus this diagram should be taken as a set of functional path-
illusion while others refer to it as the Titchener illusion ways-some of which involve the stages of processing
(see also footnote 1 in Fischer, 2001). and some of which do not.
 CHAPTER 6
cENTRAIY\\;~ i~;;i ·; ;;:
"'' «]'' {' ''-<', :~> :~,~~:1~,~::::;

CONTRIBU·~l0N;S~J.~t12 ·
TO MoToR cONiil~·

he focus in the last chapter was primarily ferent. But the only way modifications in timing
on the role of sensory mechanisms. Motor can occur is for the programmer to structure them
control was considered as a closed-loop system, into the program in advance.
dependent on feedback either (a) for online cor-
rections mediated by the stages of processing, or
(b) for corrections or compensations in the action
Open-Loop Processes
that are reflex based and not mediated by the
stages of processing. In contrast to this closed- A diagram of a typical open-loop system is shown
loop viewpoint is an open-loop system, in which in figure 6.1. The executive and effector mecha-
the instructions for movement are structured in nisms can be thought of in the same way as for the
advance and are executed without regard to the closed-loop system in figure 5.1 on page 136, but
effects they may have on the environment. That the feedback loop and the reference of correctness
is, the behavior of the open-loop system is not are missing (the feedback pathway is "cut" or
sensitive to feedback.
A good example of an open-loop system is
the traffic signal at a major intersection. The pat- Input
tern of red and green lights is controlled from
a program that handles this sequence without
regard to moment-to-moment variations in traffic
patterns. Because there is no feedback from the
traffic conditions back to the executive, there can
be no immediate modification in the pattern if an
accident occurs or if traffic is particularly heavy.
However, even though the program for the traffic
lights is inflexible, we should not get the idea that
it must be simple. The program can be structured Output
so that the north-south street has a 20% longer
green light duration than the east-west street
during rush hours, with this relation being altered FIGURE 6.1 Elements of the typical open-loop control
in midday when the traffic pattern is often dif- system.

177
178 Motor Control and Learning

"open," as when a switch is open-hence the label appear in the proper order in skills, as the chain
"open loop"). The executive is "programmed" to ensured that the second contraction did not occur
send certain instructions at particular times to the before the first one. Also, James thought that this
effector, and the effector carries them out with- mechanism could account for the timing among
out the possibility of modification if something the various contractions so important for skilled
goes wrong. In this chapter, we emphasize the actions; such timing (or more precisely, relative
open-loop processes and central representations timing) would be determined by the temporal
involved in motor control, for which sensory delays in the various sensory processes and could
influences play a much less dominant role. be relatively consistent from response to response
to produce stereotyped actions. Although such a
Response-Chaining Hypothesis model seems appropriate for serial tasks (starting
One of the earliest explanations of movement the car, buttoning a shirt), there is no conceptual
control was the response-chaining hypothesis reason why the model could not explain dis-
(sometimes called the reflex-chaining hypothesis) crete actions, such as speech and throwing, by
proposed by the 19th-century psychologist Wil- assuming that the responses triggered are the
liam James (1890). The basic idea is illustrated contractions of individual motor units. Viewed
in figure 6.2. James assumed that a movement in this way, the response chain shown in figure
began with a muscular contraction caused by an 6.2, which consists of four units of behavior,
external or internal signal. This first contraction might last only 100 ms, or it might continue for
generated sensory information (which he termed a few seconds.
response-produced feedback), which we discussed in James (1890) recognized that when skilled
detail in chapter 5. James regarded this feedback movements were produced, they did not seem
as stimulus information (just like that from any to require much consciousness for their control.
other stimulus such as light or sound), which Under the response-chaining hypothesis, move-
served as the trigger for the next contraction in the ments could be viewed as requiring attention
"chain." The second contraction then produced only for the initiation of the first action, with the
its own response-produced feedback, which trig- remainder of the actions being run off "auto-
gered the third contraction, and so on until all the matically." Also, inJames's view of learning motor
contractions in the sequence were completed. The skills, the acquisition of the associations between
feedback could come from various sources (e.g., a given feedback event and the next action is the
muscle spindles, joint receptors, or even vision fundamental basis for improvement in skill.
or audition), and it could trigger responses in the The response-chaining hypothesis is really a
same or in different limbs. With such a mecha- variant of an open-loop control mode, in spite
nism, James hypothesized how certain actions of the presence of feedback. Remember that in a

FIGURE 6.2 The response-chaining hypothesis. (The response-produced feedback from earlier portions of the action
serves as a trigger for later portions.)
 Central Contributions to Motor Control 179

closed-loop system (chapter 5), the executive level unilaterally or bilaterally, thereby eliminating
is acting on the error that is produced, and such the feedback from one or both sides of the body.
errors are computed as the difference between the These procedures have been performed on mice,
actual state and the desired state defined by the cats, monkeys, and numerous other species in
reference of correctness. In the response-chaining order to study the movement control that occurs
hypothesis, though, there is no reference of cor- in the deafferented state.
rectness against which feedback is compared- While normally these procedures are limited
the feedback simply serves as the trigger for the to animal experiments, a number of examples of
next act in the sequence. It is open loop because deafferented humans have also been reported.
the original stimulus sets the chain in motion, Lashley (1917), in a study we will describe in
and the following events are determined by the detail later, assessed a patient with a gunshot
learned associations between feedback and the wound to the lower spine. The lesion had the
next act in the sequence. Also, open-loop move- same effects as surgical deafferentation, and it
ments cannot be modified if something goes left the motor innervation of the subject intact.
wrong or if the environment changes, as is the Also, patients with complete or near-complete
case for a closed-loop model. loss of sensory information due to degenerated
One way to test the response-chaining hypoth- afferent pathways (sensory neuropathy), but with
esis is to examine the role of sensory information intact motor systems, have made significant
in the production of movement. Of course, if the contributions to this area (e.g., Bonnet, Carello,
sensory information is eliminated (or delayed or & Turvey, 2009; Gordon, Ghilardi, & Ghez, 1995;
degraded in quality), then the result should be a Rothwell et al., 1982; Sanes, 1990; Teasdale et
loss of skill, or even paralysis, because the trigger al., 1993). Finally, Kelso, Holt, and Flatt (1980)
mechanisms have been disrupted. In the next sec- have studied arthritic patients who have had the
tions, we review some evidence about feedback joints of the fingers replaced with artificial ones.
degradation in relation to this hypothesis. This operation removes the joint and the joint
capsule in which the joint receptors are located.
Deafferentation Thus, while this is not really a deafferentation
procedure in the strictest sense, it does provide
There are both temporary methods for interrupt- a situation in which feedback from the moving
ing the flow of sensory information into the spinal limb is disrupted.
cord (e.g., the blood pressure cuff technique, the
injection of local anesthetics) and permanent Deafferentation Studies
procedures, called deafferentation. Nearly all One of the earliest investigations using surgical
of the afferent input to the spinal cord enters deafferentation was conducted by Sherrington
through the dorsal roots on the posterior side of (1906). He severed the dorsal roots in a monkey
the cord. In an operation called a dorsal rhizotomy, so that only the sensations from a single forelimb
the back of the animal is entered surgically, and were lost; the remainder of the body had normal
the muscles are carefully moved to expose the sensory feedback. A major finding was that after
dorsal roots. Then, at the particular spinal level recovery from surgery, the monkey never used
of interest, the dorsal roots are cut, essentially the limb, keeping it tucked against the chest and
preventing any sensory information at that spinal using the other limbs to eat and ambulate. For
level from reaching the cord in the future. Note decades, this finding was regarded as support
that this procedure leaves intact the efferent (or for the response-chaining hypothesis, because
motor) part of the system, which exits the cord eliminating the feedback seemed to eliminate
via the ventral roots, and allows the muscles to be movement altogether, as it should if the hypoth-
activated as before the operation; but, of course, esis is correct.
it eliminates the feedback from the muscles. This But Sherrington's conclusions were later chal-
procedure can be done at a single spinal level or at lenged by a number of separate lines of evidence.
multiple levels-with additional deafferentation On the one hand, considerable research was
at each successive level progressively eliminat- completed on the control of locomotion in lower
ing more and more of the animal's sensations organisms such as fish, snakes, frogs, insects, and
from the periphery. The operation can be done birds (for reviews of this early work, see Grillner,
180 Motor Control and Learning

1975; Pearson, 1976). Some of this research to move the limbs nearly normally-activities
involved severing the afferent (sensory) path- such as climbing, swinging, eating, and grooming
ways for various segments of the animal's system. were different only in minor ways from those of
The conclusions generally were that movements the normal animals. The deafferented monkeys
are not seriously disrupted. For example, Wilson did, however, show some deficiencies in very
(1961) deafferented locusts, stimulating the insect fine manipulations, such as those that would be
electrically with a pulse near the head region, required to pick up a small piece of food. Perhaps
and wing movement patterns resembling flying this is related to the role of the cutaneous receptors
resulted. The patterns were decreased in ampli- in these movements (see chapter 5; also, Frank,
tude and frequency as compared to normal flight Williams, & Hayes, 1977). The conclusion to be
patterns, but clear rhythmic activity nevertheless drawn from these studies is that feedback from
continued. the moving limb is not essential for movement to
Why were the locust's movements so well occur, but that it undoubtedly aids the accuracy
accomplished, when Sherrington's monkey did of movement in most situations. 1 However, these
not move the deafferented limb at all? Could findings do seem to contradict the expectations of
it be that monkeys are fundamentally different the reflex-chaining hypothesis, which claims the
from the lower species in terms of their motor necessary involvement of feedback in the normal
systems? This is probably not the answer, as conduct of movement.
studies subsequent to Sherrington' s on humans
Deafferentation in Humans
and monkeys have tended to show that some
movements are not strongly interrupted by deaf- Provins (1958) studied the role of joint receptors
ferentation. For example, Lashley (1917), in his from the fingers by injecting anesthetic directly
study of the patient in whom a gunshot wound into the joint capsule. Although the subjects could
to the spine had rendered the legs deafferented, not feel the movements of their fingers, there was
asked the patient to perform various positioning nevertheless a strong capability to move; but the
movements without vision. While sitting on the accuracy suffered somewhat compared to that in
edge of an examination table, the patient was the condition without the anesthetic. Very similar
asked to extend the knee to 45°, and the error in findings were obtained by Kelso (1977; Kelso et
producing the movement was compared to that al., 1980) in studies involving the joint afferents
of a "normal" control subject. Lashley found that from the hand. When the feedback from the joint
the deafferented subject and the normal subject afferents was blocked either by Laszlo's (1967)
could do the positioning task about equally well. cuff technique (see below, and Kelso, 1977) or in
How can the apparently contradictory find- patients who had artificial finger joints (Kelso et
ings of Sherrington and Lashley be reconciled? al., 1980), little or no loss in movement position-
One possibility is that the deafferented monkey ing accuracy occurred without vision. Of course,
chose not to use the affected limb, which is quite these studies involved normal afferent feedback
different from saying that the monkey could not from the muscle spindles located in the finger
use it. You know how it feels when you sleep on muscles in the forearm, and it could be that
your arm the "wrong way," or when your jaw is this was the source of feedback that allowed for
anesthetized after a trip to the dentist; the sensa- accurate control. All these studies imply that the
tion is strange and unpleasant, and we might joint afferents are not essential for movement, as
prefer not use these effectors in these situations is often believed (e.g., Adams, 1977).
unless it was important to do so. Using an anesthetic block of the gamma
loop and other pathways from the right arm in
Later Deafferentation Studies humans, J. Smith (1969; Smith, Roberts, & Atkins,
In a series of studies, Taub and his colleagues (see 1972) found that dart throwing and grip strength
Taub & Berman, 1968, or Taub, 1976, for reviews) tasks were only minimally disrupted by this kind
and Bizzi (e.g., Polit & Bizzi, 1978) used surgi- of deafferentation. Although some impairments
cal deafferentation affecting various portions in performance occurred, the most important
of monkeys' bodies. After Taub and Berman's point is that the movement could be produced
monkeys had both forelimbs deafferented and even though feedback was not available from
had recovered from the operation, they were able the moving limb.
 Central Contributions to Motor Control 181

Many studies have been done using the so- Studies involving patients with sensory
called cuff technique popularized by Laszlo neuropathy have yielded similar conclusions,
(1967). Here, a blood pressure cuff is inflated although these individuals often have severe
around the upper arm, blocking blood flow to difficulties in performing many activities of
the lower arm; after about 20 min, the feedback daily living. For example, the patient studied by
sensations from the lower arm and hand are Rothwell and colleagues (1982) reported that he
temporarily lost; but for a few minutes, the motor had trouble feeding and dressing himself and
capabilities remain largely intact. In a series of could not hold a pen in his hand. Yet figure 6.3a
studies using tasks such as rapid finger tapping, illustrates that he could control his hands to touch
handwriting, aiming, and limb positioning, the thumb to each fingertip sequentially without
impairments were found in performance under visual feedback, at least for a short period of time.
the cuff conditions. However, the movements, The photos shown in figure 6.3b reveal that this
although impaired, still could be produced, con- activity could not be maintained for very long,
trary to the expectations from a response-chaining however; performance had deteriorated consid-
hypothesis (see also Chambers & Schumsky, 1978; erably after 30 s, when these photos were taken.
Kelso, Stelmach, & Wannamaker, 1976; Laszlo & Another example involves a woman known in
Bairstow, 1979). the literature as GL, who is completely without

FIGURE 6.3 (a) A patient with sensory neuropathy could touch his thumb and fingers without vision quite well early in the
sequence, but (b) performance was disrupted markedly after 30 s.
Reprinted, by permission, from J.C. Rothwell et al., 1982, "Manual motor performance in a deafferented man;• Brain 105: 523.
182 Motor Control and Learning

information about touch, tendon vibration, and for study in the various paradigms, although
proprioception but has normal visual feedback none seems to operate independently during
and efferent pathways. GL is able to carry out most everyday activities. It is important for stu-
her activities of daily living quite well using dents to understand these mechanisms, but the
constantvisualfeedback (Bard et al., 1999; Blouin, real problem in understanding motor behavior is
Gauthier, Vercher, & Cole, 1996; Hepp-Reymond, to appreciate how these processes work together
Chakarov, Schulte-Monting, Huethe, & Kristeva, toward smooth, elegant, and energy-efficient
2009). From the perspective of the response- performance (see also Cruse, Dean, Heuer, &
chaining hypothesis, sensory neuropathy patients Schmidt, 1990).
demonstrate remarkable capabilities to perform
skilled actions (Sanes, Mauritz, Dalakas, & Evarts, Central Pattern Generators
1985).
A number of extensive reviews summarize the
Implications for the Response- evidence about the control of locomotion and
gait in a variety of animals (Dietz & Duysens,
Chaining Hypothesis 2000; Grillner, 2007; Grillner & Wallen, 1985, 2002;
Even though work with various kinds of deaf- Goulding, 2009; Marder, Bucher, Schulz, & Taylor,
ferentation has shown that feedback from the 2005; Pearson & Gordon, 2000a; Zehr & Stein,
responding limbs is not necessary in order for 1999) and humans (Zehr, 2005). One important
actions to occur, the evidence was sometimes topic in this literature concerns so-called spinal
taken incorrectly to mean that feedback in general preparations in cats and other animals. In this
is never used in movement control. The deaf- procedure, the spinal cord is cut at a level below
ferented animals were not completely normal in the brain so that the higher (supraspinal) centers
their movement, especially when the fine control cannot influence lower ones, and often the cord
of finger action was required. We saw (in chapter is deafferented below the level of the cut as well
5) the effects of having the fingertips anesthetized (see Pearson & Gordon, 2000b). If the prepared
on performance of simple tasks such as striking a cord is then stimulated only briefly below the cut,
match. Also, it is possible, as Adams (1971, 1976b) the cord displays a definite periodicity in terms of
has said, that other kinds of feedback (e.g., vision) the activity in the efferent fibers emerging from
could be substituted for the lost sensations in the its ventral side. Thus, the spinal cord is seemingly
deafferented animal. And finally, there are many capable of producing a rhythm that can be pres-
cases in which feedback is almost certainly used ent even without input from the brain or higher
in movement, such as those that we discussed in centers, and without feedback from the limbs.
chapter 5. On strict experimental grounds, the With reference to gait patterns, a popular view,
evidence does not really say that the response- termed the half-center model (Brown, 1911), sug-
chaining hypothesis is incorrect. But the fact gests that interneurons (wholly within the spinal
that movements can occur in the absence of any cord) alternately stimulate the flexor and exten-
movement-produced feedback at all strongly sor motor neurons in a pattern more or less like
indicates that the response-chaining hypothesis is the one that would be displayed in locomotion.
not a very complete account of movement control. Apparently, the spinal cord has complex neural
Other mechanisms of motor control have to be circuitry that is capable of producing these oscilla-
used to explain the available evidence. tions. These circuits have come to be called central
pattern generators.
A schematic diagram of how such a simple
Central Control Mechanisms spinal generator might be structured is shown in
figure 6.4. Many alternatives exist, and the central
We need motor control mechanisms that explain pattern generator illustrated in figure 6.4 is only
how movements can occur in the absence of sen- a simple illustration. In this structure there could
sory feedback in order to deal with some of the be a neural network in the cord made up of four
evidence just described. In the next sections, the interneurons (the cord undoubtedly uses many
role of sensory processes is more or less reduced more). Input from some higher center (a chemical
in prominence. Keep in mind, however, that vari- or electrical signal for the spinal animal) initiates
ous mechanisms of motor control can be isolated a cyclical pattern of excitatory motor neural activ-
 Central Contributions to Motor Control 183

FIGURE 6.4 A simple possibility for the connections of interneurons forming a central pattern generator.

ity. Neuron 1 is activated, which activates neuron at the level of the cut, the animal on a moving
2, and so on, until neuron 4 activates neuron 1 treadmill began to produce stepping movements
again to start the cycle all over again. This con- that resembled normal locomotion in cats. This
tinuous cycling process would go on indefinitely stepping continued for some time after the stimu-
or until some other process turned it off. Further, lus was discontinued. As the treadmill sped up,
if neuron 1 synapses with another neuron that the cat walked faster, even trotting or galloping.
drives the flexor muscles, and neuron 3 synapses It appears that some spinal generator(s) for walk-
with another that drives the extensors, every time ing must be activated by a higher source and that,
around the cycle the flexors and extensors will be once initiated, the pattern of flexion and extension
activated. This basic concept of simple oscillating continues without further involvement from the
circuits helps to illustrate how a neural network supraspinal centers. Because the mesencephalic
could be expected to produce rhythmic patterns animal cannot sense the activity occurring in its
of activity such as gait or breathing in animals. limbs, such stepping activity must be indepen-
Showing that the spinal cord has some slow dent of the animal's feedback of the activity.
rhythmic capability is interesting, but to what As it turns out, a stimulus from the higher
extent is this activity involved in gait control? A center in the midbrain is not the only way to
very important surgical preparation in cats has initiate the spinal generators for stepping. Using
allowed considerable insight into this process. the same apparatus, Shik and Orlovskii (1976)
This preparation is called the mesencephalic studied the cat's behavior when the treadmill was
(midbrain) preparation (Shik, Orlovskii, & Sev- turned on. At first, the legs would trail off behind
erin, 1968). The cat receives a cut of the spinal the animal; then suddenly the animal would ini-
cord in the midbrain, which totally severs the tiate stepping, with the total pattern of activity
lower levels of the cord from the higher, supra- generated as a unit. As the treadmill increased
spinal centers. The cerebellum, the small structure in speed, the animal would walk faster, with
behind the midbrain, is left intact, connected to only minor differences in the pattern of activity
the spinal cord side of the cut. In this state, the cat from that observed in the normal cat, plus some
is (presumably) unable to sense any stimulation unsteadiness. As the treadmill increased in speed
from the body (because the afferent pathways to further, the cat would suddenly break into a trot
the cortex are severed) and is unable to perform pattern. Occasionally the cat could be made to
voluntary movements of the legs. Shik and col- gallop. (Remember, there is no control from the
leagues used a mechanism that supported the cat higher centers and no stimulus from higher levels
above a treadmill, as shown in figure 6.5. in the cord to turn on the spinal generators.) These
A number of important observations have results indicate that the afferent input from the
come from this preparation. First, when stimu- feet and legs, which are at first dragged by the
lated with a brief electrical current or a chemical treadmill, is sufficient to initiate stepping. Once
184 Motor Control and Learning

FIGURE 6.5 Mesencephalic (midbrain) cat supported on a treadmill as used in the study of spinal mechanisms in gait.
Reprinted from Biofizika, Vol. 11, M.L. Shik, G.N. Orlovskii, and F.V. Severin, "Locomotion of the mesencephalic cat elicited by stimulation of the pyramids;' Copyright 1966, with
permission from Elsevier Science.

the pattern generators are activated, the speed of output; this is shown by the two-way arrows from
the oscillation appears to be controlled by the rate the various muscles to the spinal generator. And,
at which the treadmill moves the cat's legs and finally, a number of spinal generators are thought
feet. When the limbs are being moved so rapidly to exist, perhaps one for each of the four limbs in
by the treadmill that a walk pattern is no longer the stepping cycle of the cat, so that the operation
effective in keeping up, the afferent information of the separate oscillators must be coordinated
presumably triggers a new pattern-the trot. An (coupled) by interneurons. Thus, in the diagram
analogous set of findings has been produced by a connection is shown from another oscillator to
Smith and her colleagues (Carter & Smith, 1986; indicate this kind of control.
Smith, 1978; Smith et al., 1986). A number of important concepts emerge from
As a result of this evidence, as well as the this work on pattern generators. First, for the
evidence reviewed by Grillner (1975), several control of gait and other stereotyped actions in a
models of gait have emerged. The general fea- variety of species (e.g., swimming in lamprey; tail
tures of these models are shown in figure 6.6.
The box in the center of the diagram represents a
Higher
central pattern generator, and it can be activated centers
or deactivated by supraspinal centers. In some
cases, this higher-level input appears to be but
a single pulse that will turn on the generator,
with no further higher-level activity necessary in
order for the oscillator to continue to operate. In Extensors
other cases, a continuous input (not necessarily a
rhythmic one) appears to be necessary, with the
action in the generator continuing only as long
as the input continues.
Sensory
The activity in the generator can also be turned input
on by sensory input. While the generator is oper-
ating, the activities in the flexor and extensor FIGURE 6.6 Spinal generator model for gait, showing
muscles are coordinated, and feedback from the input to oscillators from sensory sources, from higher cen-
responding limbs also can serve to modify the ters, from other oscillators, and from moving limbs.
 Central Contributions to Motor Control Ii!! 185

flipping or escape reactions in lobsters; groom- reflexes is reciprocal inhibition, whereby the flexors
ing in mice), strong evidence exists that these of a joint tend to be automatically inhibited when
patterns are controlled by "prewired" pattern the extensors are activated. With the crossed-exten-
generators that can handle most of the details of sor reflex, the extensors of one knee are activated
the actions. They can be turned on by a variety when the flexors of the opposite knee are called
of sources of stimulation, and they can continue into action. When we step on a tack, the flexors in
until they "run down" or are stopped by some the affected leg take the weight off the tack, while
other source of input. While the basic pattern is the extensors in the opposite leg help to prevent
quite stereotyped, in "higher" animals (such as falling. Another example is the tonic neck reflex, in
cats), extensive modification of the basic pattern which turning the head to the right causes facilita-
is possible, either from higher centers to make tion in the arm flexors on the left and in the arm
the whole pattern more rapid or more forceful, extensors on the right (see chapter 5).
or from lower feedback sources (e.g., from the That these reflexes exist is not in question. They
leg or foot) that serve to alter the particular pat- are especially easy to identify in infants, and
tern of force applied to conform to variations in they have been used in the diagnosis of various
terrain. And, finally, these pattern generators neurological disorders. But to say that gait and
do not require the conscious awareness of the other movement behaviors consist of the vari-
animal in order to operate. Once initiated, they ous reflexes implies a model in which the motor
can apparently continue without involvement of system is always reacting to peripheral stimula-
the higher centers. However, in the operation of tion. A more reasonable viewpoint, based on the
these generators during running, for example, evidence on central pattern generators, is that
attention seems to be required, perhaps to evalu- reflexes ensure that the pattern of activity speci-
ate the upcoming terrain or to keep the oscillator fied by the central pattern generator is carried out
running (e.g., Duysens & Van de Crommert, 1998; effectively in the face of unexpected changes in
Van de Crommert, Mulder, & Duysens, 1998). the environment. The muscle spindle and gamma
systems seem to fill this role. One function of the
Reflex Involvement in Locomotion gamma system is to maintain muscle stiffness
Throughout the history of motor skills research, (i.e., its mechanical, spring-like properties) in the
considerable debate has centered on the role of face of various unexpected changes in muscle
reflexes, and our discussion, about genetically length because of changes in joint position(s)
defined activities such as locomotion would be (Nichols & Houk, 1976). If an animal steps on
incomplete unless we considered reflex involve- a patch of ground that is higher than expected,
ment (see also chapter 5). The following sections the "springy" muscles allow extensors to yield
present some of these diverse viewpoints. without collapsing, maintaining a smooth gait
without a fall. The view is that reflexes are pre-
Maintaining the Original Pattern pared to operate if the system is perturbed but
The focus in the next few sections concerns the do not have a particularly important role under
ways in which feedback activity of various kinds normal circumstances (Grillner, 1975).
cooperates with the central programs for action.
We are only now beginning to understand the Reflex-Reversal Phenomena
important principles of these sensory-motor An important variation of the concept of trig-
interactions. gered reactions (discussed in chapter 5) is the
An older notion of control in locomotion was concept of reflex reversals described in relation to
that the patterns of limb action consisted of fun- locomotion by Forssberg, Grillner, and Rossignol
damental reflex activities (e.g., Easton, 1972). This (1975). In the locomoting cat, when a light touch
was somewhat different from the reflex-chaining or a weak electrical shock is applied to the top of
hypothesis in that the component reflexes were the foot during the flexion portion of the swing
thought to be the identifiable, genetically defined phase of the gait cycle (i.e., the time at which the
patterns that we see so often in infants, whereas animal is lifting the foot in preparation for the
the reflex-chaining hypothesis involves any swing forward), an abrupt increase in the flexion
chained activity-even those that are learned. response occurs (with an extension response in
A good example of these genetically defined the opposite foot), as if the cat were trying to lift
186 111 Motor Control and Learning

its foot to avoid an obstacle (such as a rock) that stance phase), then the reflex that would serve
would cause it to trip. (This crossed-extensor pat- to lift the leg in response to a tap is inhibited by
tern is not voluntary, as it can be shown to exist the central generator. However, this pathway is
in the mesencephalic cats described earlier; thus, activated when the flexors are activated (in the
the response is spinal in origin.) However, when flexion phase). In this way, the pattern generators
the same stimulus is applied to the foot during involve the already structured reflex pathways
the phase of the gait cycle in which the foot is so that they contribute maximally to the ani-
on the ground (the stance phase), essentially no mal's overall movement goals (see also Hasan
reaction, or perhaps a slight extra extension, takes & Enoka, 1985).
place in the stimulated foot-a response opposite The role of reflexes in human posture and
that shown in the swing phase of the step cycle. locomotion is also starting to emerge. Zehr and
Because the same stimulus causes two different Stein's (1999) excellent review on the topic identi-
patterns of action depending on the phase of the fied a number of situations in which the role of
stepping cycle, this effect has been termed the reflexes changes either as a function of the task
reflex-reversal phenomenon. being performed or due to a specific phase of the
Frequently (but see some exceptions in chapter activity within the current task. For example, pain
5), a reflex is thought of as a stereotyped response reflex responses are diminished during stand-
caused by a particular stimulus (e.g., the blink ing as postural load is increased (a task-related
of an eye when a puff of air is presented). Yet modulation of the cutaneous reflex), whereas the
the evidence just cited in locomotion indicates functional importance of cutaneous and muscle
that the response to the stimulus depends on the reflexes changes between the parts of the step
location of the limb in the stepping cycle and is cycles within the task of locomotion. Figure 6.7
not simple and stereotyped. Thus, a simple view illustrates these changes within locomotion as
of reflex control cannot explain these effects. summarized in Zehr and Stein's (1999) review.
This kind of evidence has been explained (e.g., Cutaneous reflexes involved in corrections for
Grillner, 1975) based on the assumption that the a stumble and foot-placing reactions dominate
spinal generators for locomotion, in addition to the swing and swing-to-stance phases of the step
providing efferent commands to the relevant cycle. Muscle reflexes involved in weight sup-
musculature, also provide feedforward signals to port, stability, timing, and unloading dominaje
other locations in the cord that serve to modify the stance and stance-to-swing phases (see Zehr
the actions of various reflexes. The sense of this & Stein, 1999, for details).
control is that if the pathways to the extensors Other studies of the role of reflexes in humans
of the right leg are being activated (during the have been conducted in research on the motor

FIGURE 6.7 The phase-dependent role of reflexes in locomotion. Cutaneous reflexes dominate the swing and swing-to-
stance phases, and muscle reflexes dominate in the stance and stance-to-swing phases.
Reprinted from Progress in Neurobiology, Vol. 58, E.P. Zehr and R.B. Stein, 'What functions do reflexes serve during human locomotion?" 185-205, Copyright 1999, with
permission from Elsevier.
 Central Contributions to Motor Control mm 187

control of speech (Abbs & Gracco, 1983; Abbs, The "Smart" Spinal Cord
Gracco, & Cole, 1984; Kelso, Tuller, Vatikiotis-
Bateson, & Fowler, 1984). These are also dis- Early in the thinking about motor control, the
cussed in chapter 5. Subjects in the study by spinal cord tended to be viewed as a "freeway"
Kelso and colleagues said simple sentences, that simply carried impulses back and forth from
such as "It's a /baez/ again" or "It's a /baeb/ the brain to the peripheral receptors and muscles.
again," and received an unexpected downward Gradually, as many spinal activities were iso-
perturbation to the jaw during the target syl- lated and studied (e.g., Sherrington, 1906), the
lable /baez I or /baeb I. When /baez I was to be spinal cord came to be regarded as considerably
spoken, a perturbation to the jaw evoked a rapid more complex-and has even been considered
compensation in the tongue muscles but not in as an organ. The evidence that the spinal cord
the lip, whereas the same perturbation during contains central pattern generators for gait and
/baeb I produced the reverse pattern-a com- other movements continues to point toward the
pensation in the lip but not the tongue; this com- cord as a complex organ where much of motor
pensation occurred on the very first trial. As in control is structured. Further, evidence suggests
the work on locomotion, the same stimulus (here that the spinal cord is responsible for considerable
the jaw perturbation) produced rapid responses integration of sensory and motor information, as
(about 30 ms latency) that were very different shown by the following example.
depending on the action being produced (i.e., the Figure 6.8 shows a frog making a wiping
specific word). But, unlike the changes observed response to a noxious stimulus placed on the
in the locomotion studies, these speech modifica- "elbow." Fukson, Berkinblit, and Feldman (1980;
tions were probably learned along with speech, Berkinblit, Feldman, & Fukson, 1986), like others
as it is difficult to imagine how this capability before them, showed that the frog is capable of
could have been a part of an inherited pattern performing these hindlimb responses when spi-
as was the case for locomotion. These rapid sen- nalized (i.e., with a transection that separates the
sory-motor processes seem to serve the general cortex from the intact spinal cord). The response
purpose of ensuring that the overall goal of the always begins with a movement of the hindlimb
particular movement being attempted at the time toe to the region of the shoulder area, followed by
is maintained, with the modifications necessar- a rapid wiping action that is aimed at the elbow.
ily being different for different movement goals It is interesting that the animal can use sensory
(Schmidt, 1987). information from one part of the body (the elbow)

FIGURE 6.8 In the spinal frog, the hindlimb response to wipe an acid stimulus from the "elbow" is aimed to various elbow
positions without the involvement of voluntary control from the cortex.
Reprinted with permission from 0.1. Fukson, M.B. Berkinblit, A.G. and Feldman, 1980, ''The spinal frog takes into account the scheme of its body during the wiping reflex;'
Science 209: 1261. Reprinted with permission of AAAS.
188 Ill Motor Control and Learning

to trigger an action pattern in some other part (the tured in the same way? Are there programs in
hindlimb ), even when spinalized. What is of more the spinal cord that can handle the production
interest, however, is that the animal produces dif- of a football pass if they are activated by a pulse
ferent wiping movements depending on the loca- from the midbrain, or do programs that are not
tion of the elbow at which the response is aimed. genetically defined have some different origin?
That is, the central pattern generator for this These questions are difficult to answer, as almost
response appears to modify its action depending no research with animals has used tasks or skills
on the sensory information from the forelimb that we could consider learned, or that are not
indicating the position of the stimulus-the cord genetically defined. One hypothesis is that the
"knows" where the limbs are. Remember, the frog control of learned and genetically defined actions
had no cortical involvement in this response, and is fundamentally the same, but no good evidence
thus no awareness of the limbs' actions, so this is available on this question. Instead, we tum next
integration of sensory information was done at to evidence about the role of central control mecha-
very low levels, perhaps completely within the nisms-motor programs, specifically-that has been
spinal cord. Such observations indicate that the generated in behavioral studies with humans.
spinal cord is a very smart" organ indeed.
/1

Another example, from the work of Smith and

colleagues (1986) on spinalized cats, is particu- Central Control
larly impressive. When a piece of tape was placed of Rapid Movements
on the walking cat's hind paw, a stereotyped
paw-shake program was initiated in which the cat Consider a very rapid limb movement in which
lifted the foot and shook it rapidly and violently the pattern of action is initiated and completed
for about 10 to 13 cycles, apparently for the pur- in less than 150 ms. There are many examples of
pose of shaking the tape loose. Of course, because
movements like this, such as a key press in typing
of the spinal section, this stereotyped program (<100 ms) and, in sport, the bat swing in baseball
must have been initiated through peripheral
(~ 140 ms) and boxer Muhammad Ali's left jab (40
stimulation from the foot and controlled by the ms). Because these discrete tasks are so highly
spinal cord. But even more remarkably, the spinal represented in our everyday activities, they have
cat could walk on the treadmill (another program) been studied in laboratory settings in an attempt
and shake the paw at the same time, triggering
to understand how they are controlled.
the paw shake when the limb was in the swing
A laboratory study might involve beginning
phase of locomotion and turning it off when the
with the elbow in one position and then rapidly
foot was on the ground in support. Somehow the extending it so that the hand comes to rest at or
spinal cord, without the help of a higher center,
near a target 30 cm away. Although at first this
"knew" how to coordinate these two actions
movement appears to be very simple, the kinds
simultaneously to achieve the double goals of
of neurological activities associated with it are
removing the tape and walking without falling.
elegant and complex. Although many examples
could be presented, one in a study by Wadman,
Human Skills Denier van der Gon, Geuze, and Mol (1979) makes
The evidence and ideas presented that support the point particularly well. Figure 6.9 shows a
the concept of a central pattern generator were pattern of electromyographic (EMG) activity in
produced in animals. Suggestions that central this elbow extension task; this kind of pattern is
pattern generators also operate in humans have common in many investigations using fast move-
sparked considerable debate, and to date the evi- ments such as this. The EMGs from the triceps
dence is not entirely clear on the issue (MacKay- (the agonist) and the biceps (the antagonist) are
Lyons, 2002). We might expect that central pattern shown. The "raw" EMGs from the muscles have
generators likely would operate in humans to been rectified, meaning that the negative swings
some extent (see Zehr, 2005, for more discussion). of the EMG signals have been changed to positive
The movements studied in the cat mentioned values, and these positive values are plotted as a
previously are probably genetically defined and function of time in the action. The occurrence of
"prewired." To what extent are motor programs peaks in the rectified EMGs represents periods
for movements like throwing a football struc- of heightened activity in the muscle in question.
 Central Contributions to Motor Control 1111 189

- - Normal

0.0 0.1 0.2 0.3

Time (s)

FIGURE 6.9 Agonist (triceps) and antagonist (biceps) electromyographic activity in a rapid elbow extension action.
Reprinted from W.J. Wadman, 1979, "Control of last goal-directed arm movements," Journal of Human Movement Studies 5: 10. By permission of W.J. Wadman.

Also, the record for the biceps (the antagonist) has activity? This question becomes much more
been turned "upside down" so it can more easily complex when applied to skills for which many
be compared to the triceps record. muscles tum on and off at particular times, such
The distinctive triple-burst EMG pattern is as in pole-vaulting and swinging a sledgeham-
prominent in figure 6.9, first identified in the 1920s mer. The closed-loop account of movement
by the German physiologist Wachholder (see behavior involving the stages of information
Stemad, 2001; Stemad & Coreas, 2001). A burst of processing could account for these features of the
the agonist (triceps) muscle occurs first, and then movement by indicating that the system monitors
the agonist is turned off and the antagonist (the the position of the limb (perhaps by sensing the
biceps here) is turned on, presumably to bring the joint angle) and waits until the limb is at some
limb to a stop. Then, near the end of the action, particular position before turning off the triceps
the antagonist muscle is turned off and the agonist and turning on the biceps. That is, the system
comes on again, probably to cause the limb to be could use the feedback from the responding limb
"clamped" at the target, dampening the oscilla- to trigger the end of activity in one muscle and
tions that could be produced. The exact nature of initiate activity in the other. 2 However, a number
this pattern of EMG activation depends on various of fundamental difficulties arise with this idea
task-related parameters (Stemad & Coreas, 2001). about the control of actions, as discussed next.
A question of interest for these kinds of actions
is how the motor system controls the timing of The Degrees of Freedom Problem
these events-how does it "know" when to turn One difficulty for the closed-loop model, and for
off the triceps activity and tum on the biceps any other model holding that the contractions
190 i!! Motor Control and Learning

of the various muscles are handled by direct if the executive does not have the responsibility
commands from higher centers, was raised by for controlling the degrees of freedom, how are
the Russian physiologist Nikolai Bernstein (see the many degrees of freedom controlled, and
"Nikolai Bernstein" on p. 8). Bernstein's idea was how can the elegant, skilled organization of the
that if the information-processing system were muscles and joints be achieved?
involved in the production of all the decisions This general question has been of fundamen-
about each of the muscles participating in a motor tal concern since scientists began to think about
act, it would be difficult to imagine how this movement skills. One solution has been to pos-
would explain all of the mental work involved tulate a structure, subordinate to the executive,
in producing even a simple act like that shown which can account for the particular organization
in figure 6.9. The fundamental concern is that the required among the separate degrees of freedom.
system has too many independent states that must These theoretical structures, called motor pro-
be controlled at the same time. These independent grams by many (e.g., Brooks, 1979; 1986; Henry &
states are called degrees offreedom. Rogers, 1960; Keele, 1968, 1986; Keele, Cohen, &
As an example, consider a simple arm move- Ivry, 1990; Lashley, 1917; Schmidt, 1975b, 1988),
ment. Each joint is capable of moving indepen- have the capability to influence the activity of the
dently, having at least one degree of freedom many independent degrees of freedom so that
that must be controlled in an action. And some they act as a single unit. If this temporary orga-
joints have two degrees of freedom, such as the nization can be imposed on the system, then the
shoulder, which can (a) allow the hand to move problem for the executive will have been simpli-
in a half sphere with the elbow "locked" and (b) fied so that only a single degree of freedom will
rotate the shaft of the arm, independently. The need to be controlled. In other words, the execu-
degrees of freedom problem is compounded fur- tive level is thought to control the selection of a
ther because each joint has a number of muscles motor program, ready it for action, and initiate it
acting on it, and each of these muscles is made at the proper time. Once under way, the program
up of hundreds of motor units that also must be controls the activity of the individual degrees of
controlled. All of the independently moving parts freedom involved in the movement. In this sense,
would lead to an impossible situation for the the executive is freed from the task of solving the
central nervous system if it had to control these degrees of freedom problem.
degrees of freedom individually by conscious This kind of solution to the degrees of freedom
decisions (see also Greene, 1972; Whiting, 1984). problem sounds very simple-in fact, too simple.
In searching for answers about how we con- Explaining how all the degrees of freedom are
trol movement, a good question is "How can the controlled by inventing some theoretical structure
very many degrees of freedom of the body be (here, the motor program) does not really answer
regulated in the course of activity by a minimally the question of control at all. The question of
intelligent executive intervening minimally?" exactly how the various degrees of freedom are
(Kugler, Kelso, & Turvey, 1980, p. 4). The idea is coordinated still must be addressed. The debate
that an executive level should not be thought of
about the nature of this reduction in controllable
as having much responsibility for the control of
degrees of freedom-in terms of what variables
the many degrees of freedom; there are simply too
are controlled, when, and how-represents a
many degrees of freedom for even an intelligent
truly fundamental problem for researchers in
executive to control and yet still have capability
motor control. We will consider some of these
left over to do anything else.
questions later in this chapter and again in chap-
On other grounds, for many years scientists
ters 7 and 8.
have thought that actions are controlled, not the
individual degrees of freedom such as muscles
Agonist-Antagonist Patterning
or motor units. When we perform an action, we
seem to be aware of the goal of the movement, Another argument against the idea that some
its global action pattern, and its effect on the executive level directly terminates the triceps
environment; but we are hardly aware at all of burst of EMG and initiates the biceps burst (figure
the particular muscles used and are never aware 6.9) relates to the time available for generating
of the particular motor units involved. Therefore, these events. The information-processing stages
 Central Contributions to Motor Control li!i 191

require considerable time in order for an indi- impresses us is essentially the reverse of this:
vidual to produce a response to an environmental The antagonist (here, biceps) EMG is turned on
stimulus, such as the offset of the triceps EMG even though no movement of the limb occurred.
and the onset of the biceps EMG. Although this The fascinating question is thus, What did turn
model might be acceptable for very slow move- on the EMG in the antagonist? Feedback (e.g.,
ments, if the action is very fast (as in the example vision) from the movement could not have
in figure 6.9), the movement will be over before been processed by the stages of processing; the
one of these stimulus-response processes can be stages are too slow and, in any case, there was
completed. Something else must be involved in no movement. Reflex activities associated with
the activation of the various muscle groups. joint movement could not do this, as there was
This argument, however, does not rule out the no movement and hence no usable output from
possibility that the patterning has been influenced the various proprioceptors. The dynamics of the
by reflexive adjustments, handled at lower levels moving limb cannot be responsible for activating
in the motor system. As discussed in chapter 5, the antagonist, as the limb did not move and, in
such processes are far faster than those involved addition, had its dynamics severely disrupted.
in the stages of processing, and they might be We are left with one theory that can handle these
rapid enough to account for the control entailed findings: The motor program for action was
in these very fast actions. A number of models prepared in advance to govern the timing in the
based on lower-level reflex control have been action, and this program continued to "run" even
proposed (e.g., Adamovich & Feldman, 1984; though the limb was blocked. That is, the answer
Feldman, 1986), and considerable debate has to the question "What turned the antagonist on
centered on this question. Some available data, at the proper time (even though the limb was
however, seem to suggest that these models are blocked and prevented from moving)?" is The
not correct either. motor program did.
Wadman and colleagues' (1979) experiment Using considerably slower movements, Angel
had a very important, additional condition that (1977) found that the antagonist EMG disap-
we have not yet described. When the subject peared when the movement was blocked; this
had become accustomed to the apparatus and suggests that sufficient time was involved for
movement task (a rapid, 150 ms, elbow exten- feedback to higher levels, or reflex activities, or
sion to a target), on particular trials the lever was both to have an effect on the action. This also
unexpectedly locked mechanically at the starting provides evidence for the long-held view that
position so that no movement of it could occur. "fast" and "slow" movements are controlled
Figure 6.9 shows the EMG patterns available from in fundamentally differently ways. Simply, fast
these "blocked" trials, superimposed on the pat- movements seem to be controlled open loop,
terns from the "normal" trials. For approximately whereas slow movements appear not to be. We
110 ms after the initial agonist EMG burst, the two will return to these ideas several more times later.
EMG patterns were nearly identical, with only
minor modifications in the patterning afterward. Startled Actions
The most important feature of figure 6.9 is that the
onset of the antagonist (biceps) EMG occurred at Another line of evidence, using a very different
the expected time on these "blocked" trials-that experimental paradigm, supports the conclusions
is, it occurred at approximately the same time in we have just drawn from the study by Wadman
the blocked trials as it did in the "normal" trials. and colleagues (1979). Our usual reaction to a
Magill, Schmidt, Young, and Shapiro (1987) and loud, unexpected sound signal is a startle response,
Shapiro and Walter (1982) obtained similar results which is characterized by a sudden, fast, auto-
for fast movements. matic, generalized, stereotypical reaction that
Several have argued that unexpectedly block- includes a rapid eye blink and flexion of the
ing the limb from moving would result in reflex- neck muscles; this response is clearly one of the
like responses from various proprioceptors, and reflex-like actions that result from certain stimuli.
hence the EMG patterning of the blocked action Valls-Sole, Rothwell, Goulart, Cossu, and Mufi.oz
cannot tell us about the originally planned pat- (1999) asked subjects to perform a large number
tern of action. We agree, to a point. But, what of trials involving a movement task similar to
192 Motor Control and Learning

the task used by Wadman and colleagues (1979).

On a few of these trials, the visual stimulus was 120
accompanied by a very loud (130 dB) auditory
signal. As expected, the loud stimulus produced 110
U)
the typical startle reflex responses. Unexpectedly, g
however, the startle response was also accompa- I- 100
cc
nied by the action that the subjects were prepared ...
0
to make-all of the features of the triple-burst 0 90
E
(I)
pattern seen in "normal" trials, including magni- ....
a.
tude and timing of the EMG patterns, were also 80
observed in these "startle" trials. However, the
response was initiated about 100 ms earlier than in 70
the "normal" trials! The premotor reaction times
(RTs) (see chapter 2), which had averaged 171 ms 83 93 103 113 123
Stimulus intensity (dB)
on the normal trials, were reduced to 77 ms on the
startle trials. And, even though the subject also
FIGURE 6.10 Increases in stimulus intensity produce
produced the typical startle response (eye blink
faster premotor reaction times (RTs) in the absence of a
and neck reflexes), all of the movement events, startle response. In the presence of a startle response
including magnitude and timing of the EMG pat- indicator (sternocleidomastoid muscle electromyographic
terns, were identical to those in the normal trials. activity), premotor RTs were reduced by 20 to 30 ms and
immune to stimulus intensity effects.
One interpretation of these findings is that
With kind permission from Springer Science+Business Media: Experimental Brain
the prepared movement was still released "vol- Research, "Startle produces early response latencies that are distinct from
untarily," just much earlier because auditory stimulus intensity effects," Vol. 176, 2007, pgs. 199-205, A.N. Carlsen, C.J. Dakin,
R. Chua, and J.M. Franks, figure 1.
stimuli result in faster RTs than do visual stimuli
or because the more intense stimuli produce
faster RTs than less intense stimuli (Woodworth two pieces of evidence argue against such an
& Schlosberg, 1954). Carlsen, Dakin, Chua, and interpretation. First, the premotor RTs (77 ms) and
Franks (2007) provided evidence against such RTs (104 ms) were much shorter than would be
an interpretation, however, when they separated expected for a voluntary response. And, second,
startle trials that produced an accompanying in the postexperiment questioning, two-thirds of
EMG activation in the neck (sternocleidomastoid the subjects claimed that "something other than
muscle) from those trials that did not (presum- their own will was making them move" (Valls-
ably, the former trials were ones in which an Sole et al., 1999, p. 935), even though none of the
actual startle response occurred). Figure 6.10 subjects was aware of the speed-up.
illustrates their main results. Even though louder Valls-Sole and colleagues (1999) argued that
auditory stimuli produced faster premotor RTs the motor program is stored in subcortical
than softer stimuli (by about 20 ms), the trials that structures of the brain (see Valls-Sole, Kumru, &
produced a neck response (indicating a startle) Kofler, 2008, for a review of the evidence). The
resulted in premotor RTs that were 20 to 30 ms presentation of the startle stimulus results in an
shorter than in trials without the startle response involuntary "release" of the motor program, in
indication. Also, responses in these "true" startle advance of the voluntary initiation that would
trials were independent of the loudness of the have occurred in response to the regular (visual)
stimulus-all stimuli that produced a true startle stimulus. Valls-Sole and colleagues' (1999)
response had premotor RTs of about 80 ms. This procedures and findings have been replicated
suggests that a startle had a role in these findings. using similar tasks (e.g., Carlsen, Chua, Inglis,
Another possible explanation for the results of Sanderson, & Franks, 2004; Oude Nijhuis et al.,
Valls-Sole and colleagues is that, since the visual 2007; Siegmund, Inglis, & Sanderson, 2001) and
and auditory stimuli were presented together, also using saccadic eye movement responses
there was an intersensory facilitation effect, in (Castellote, Kumru, Queralt, & Valls-Sole, 2007).
which the two stimuli somehow "combined" to These results provide another convincing line
produce a facilitation (Gielen, Schmidt, & van of support for the idea that fast movements are
den Heuvel 1983; Schmidt et al., 1984). However, organized in advance via motor programs.
 Central Contributions to Motor Control 193

Inhibiting Actions that sometimes continues in its fine detail even

if a button is missing. Again, feedback from the
In the next sections we examine some of the evi- fingers is probably not critically involved in the
dence about the performer's capabilities to inhibit production of this sequence of movements. But,
a movement, even before it has been initiated. this seems to contradict the finding that striking
This information is closely related to the findings a match was seriously impaired by anesthetic
presented in chapter 4 (e.g., psychological refrac- applied to the cutaneous receptors of the finger-
toriness), and the two kinds of findings seem to tips. More work is needed in this area to resolve
blend well to provide a picture of what happens these uncertainties.
during the course of a rapid action. The British psychologist Bartlett expressed
Anecdotal Evidence similar views about the lack of role for feedback
in certain ball sports. Bartlett (1958) suggested
Evidence from personal experience is quite dif-
that the "launching" of an action signals a point
ficult to interpret, for the strong possibility exists
(or region) of no return-beyond which attempts
that what we think we do is not what actually
to modify the action are largely unsuccessful. In
occurs, leading to a false picture of movement
the next sections, we describe some experimental
control processes. Even so, some of our common
approaches to examine these issues.
observations guide us to experiments in which
the ideas can be studied more carefully. Inhibiting Rapid Discrete Actions
Long after his work with the wounded patient, Henry and Harrison (1961) presented one of the
Lashley (1951) provided an example of a skilled first experimental analyses of these questions.
pianist playing a piano with a broken key that They asked subjects to begin with a finger on a
could not be depressed. As the pianist played a key located at their hip and, at a "go" signal, to
string of notes, the attempts to press the broken move the arm forward and upward to trip a string
key did not interrupt the sequence of actions at located in front of their right shoulder. Subjects
all. In fact, only after the entire sequence was were instructed to do this as quickly as possible.
completed did the individual notice and remark The simple RT in these control trials was 214 ms
that the key was broken. This suggests that the on the average, and the movement time (MT)
actions do not appear to be structured with feed- was slightly shorter, at 199 ms. On some trials,
back to "verify" that a certain finger movement a second, "stop" signal would occur, indicating
has been made before the next one is commanded. that the subject should avoid tripping the string
The feedback appeared to be only minimally or at least try to begin to slow the limb as quickly
involved in the production of the movement as possible. The "stop" signal could come on at
sequence (see also Gordon & Soechting, 1995). one of four times: 110, 190, 270, and 350 ms after
Another example is buttoning a shirt, an activity the "go" signal. Figure 6.11 shows the timing of

FIGURE 6.11 A time line showing the critical events in the Henry-Harrison experiment. ("Stop" signals were presented at
various times after an initial "go" signal.)
Reproduced and adapted from Tables 1 and 2 with permission of author and publisher from: Henry, F.M. and Harrison, J.S. "Refractoriness of a fast movement;' Perceptual and
Motor Skills, 1961, 13, 351·354. ©Southern Universities Press 1961.
194 Motor Control and Learning

the essential details of the experiment, indicating this occurred even when the "stop" signal was
where the "stop" signals could come on relative provided on the last letter of the previous word.
to the observed RT and MT durations. Given that "the" is the most frequently used word
The primary result of interest to Henry and in the English language (Kucera & Francis, 1967),
Harrison was the time taken to begin to decelerate typists have probably developed a very highly
the limb after the "stop" signal was presented. overlearned motor program for typing it-one
They found that only when the "stop" signal whose entire execution is difficult to inhibit after
was given at the very earliest of the presentation the internal "go" signal has been issued. At what
times (the 110 ms location) was there a tendency point in the RT to a signal are we committed to
for the subjects to even start to slow the move- action? In other words, when does the point of no
ment before it had been completed. But the more return occur? An experiment by Slater-Hammel
interesting feature of these data is the subject's (1960) helps to answer this question; as well, it
response in the 190 ms condition. Notice here supports some of the other points just made about
that the "stop" signal came on 24 ms before the movement programming.
movement even started, and yet the movement
Inhibiting Anticipatory Actions
was carried out without interruption. That is, a
signal presented before the movement was not Slater-Hammel (1960) asked subjects to watch a
effective in modifying that particular movement, sweep timer that made one revolution per second
even when the movement lasted for 199 ms (see and to respond by lifting a finger from a key to
also Gao & Zelaznik, 1991). stop the clock at the moment when the timer
If the information-processing stages are too reached "8" on the second revolution (i.e., 1.8
slow to be involved in the details of a particular revolutions after the clock hand started). The sub-
action, and segmental reflexive control is not ject could not, of course, wait until the clock hand
involved either, then (as we have argued earlier had actually arrived at "8" before planning and
in this chapter) these movements must be prepro- initiating the movement, because the finger lift
grammed, structured in advance, and run off as a would be far too late. So the subject's task on these
unit without much modification from events in trials was to anticipate the movement of the clock
the environment. An additional interpretation is hand, together with the lags in the information-
that, once some internal "go" signal is issued, the processing and neuromuscular systems, so that
action occurs and cannot be stopped, as in pulling the finger was lifted at precisely the correct time.
the trigger on a gun. The initiation of the internal Slater-Hammel added an interesting condition,
"go" signal represents the point of no return. however. Occasionally, unpredictably from the
When the (external) "stop" signal was presented viewpoint of the subject, the experimenter would
104 ms before the overall response was to begin, stop the clock hand before it reached "8." The
the response began anyway, and an additional 110 subject was instructed to not lift the finger from
ms or so was required to even begin to modify it. the key if this happened-that is, to do "nothing."
Logan has performed many of the recent Slater-Hammel set up the experimental condi-
experimental and theoretical analyses using this tions such that the primary task (stopping at the
"stop signal" paradigm (reviewed in Boucher, "8") was required on about 75% of the trials. On
Palmeri, Logan, & Schall, 2007; Logan, 1994; the remaining trials, the clock hand could stop
Verbruggen & Logan, 2008). In one of these stud- randomly at some point prior to the target point
ies, skilled typists were asked to type phrases as at "8." The critical events of the experiment are
rapidly as possible, but to stop their typing upon illustrated in figure 6.12.
presentation of a "stop" tone (Logan, 1982). In The measure of most interest to Slater-Hammel
general, the typists produced one to two addi- was the probability of the subject's inhibiting the
tional letters after the "stop" tone occurred (about finger lift successfully as a function of the time
200 ms of MT), regardless of the specific position before "8" when the clock hand was stopped by
within the word at which the tone was presented. the experimenter. Figure 6.13 presents the prob-
However, there were several notable exceptions; ability of successful inhibition as a function of the
one occurred with the word "the." Logan (1982) time before "8" when the clock hand stopped. If
found that typists would almost always produce the clock hand stopped 250 ms before "8," the
the entire "the," and the space after it, and that subject would probably have had no trouble
 Central Contributions to Motor Control Iii! 195

1.0

0.9

~ 0.8
:c 0.7
:c
·=
0
0.6

~ 0.5
:c
1l 0.4
2
CL 0.3

0.2

0.1

Time before response (ms)

FIGURE 6.13 The probability of successfully inhibiting

an anticipated finger lift as a function of the interval before
the critical event.
FIGURE 6.12 The task used by Slater-Hammel (1960). Reprinted with permission from Research Quarterly tor Exercise and Sport, vol. 31,
The sweep hand of the chronoscope normally made a full pg. 226. Copyright 1960 by the American Alliance for Health, Physical Education,
revolution in 1 s (1,000 ms). The subject's primary task was Recreation and Dance, 1900 Association Drive, Reston, VA 20191.
to stop the sweep hand of the chronoscope on its second
revolution precisely on the 800 ms mark by lifting a finger
off a response key. If, however, the sweep hand stopped
prior to the 800 ms point, then the subject was to try to responded to the primary task with no error (i.e.,
inhibit the finger lift. assuming a CE= 0), we must therefore add (as
Slater-Hammel [1960] did) the 26mstothe 140ms
to obtain 166 ms as the estimate of the time before
inhibiting the movement, as the stages of process-
ing would seem to have ample time to inhibit it. "8" that the internal "go" signal must have been
Conversely, if the clock hand stopped only 50 ms produced. Thus, these data provide one estimate
before the clock reached "8," the subject should of a point of no return at 166 ms before the action.
never have been able to inhibit the movement. A number of important interpretations may be
That is essentially what Slater-Hammel found. derived from this thinking. First, the finding that
But notice that as the time before "8" decreased the subject could not inhibit a movement once it
from about 170 ms, the probability of successfully was internally planned and initiated supports the
inhibiting the movement decreased sharply; the observations made about the Henry and Harrison
probability was about 0.5 when the interval before (1961) study and the typing of the word "the"
"8" was 140 ms. Another way to state this finding shown by Logan (1982). Apparently, once the
is to say that if the clock hand stopped 140 ms subject has committed to action, the movement
before "8," the subject could successfully inhibit occurs even when some signal is presented in the
the response about half the time. In our interpre- environment shortly afterward indicating that the
tation, this result says that, at 140 ms before the action should not be performed. As mentioned in
finger lift (on average), the subject irrevocably chapter 4, inhibiting preplanned actions requires
triggered the action. time and attention, an idea supported by these
However, this conclusion needs to be adjusted data.
slightly. Although this median (50%) point Students do this experiment in the laboratory
occurred at 140 ms before "8," Slater-Hammel's sections of our courses; they see that the clock
subjects responded a little late on average (con- hand has stopped, but their hand then responds
stant error= +26 ms, responding at the 826 ms anyway. The subjective feeling is that they do not
point on the clock, rather than at the "8," or 800 have control over the hand, where "they" refers
ms mark). Because our first estimate of the point to "their consciousness." The point of no return
of no return (140 ms) presumed that subjects appears to represent the point at which conscious
196 Motor Control and Learning

control of the hand has been transferred, with Programming Rapid Movements
the hand now subservient to the motor program in Advance
instead (Osman, Kornblum, & Meyer, 1990; see
also De Jong, Coles, Logan, & Gratton, 1990). In chapter 3, we introduced the idea that in a
These ideas of conscious control, and related con- response-programming stage the person selects,
cepts such as "intention" or "will," which were organizes, and initiates a program of action
the source of philosophical debate for centuries, that will produce a series of muscular activities
have resurfaced in recent years (e.g., Haggard, resulting in an action. According to this model,
2008; Libet, 1985) thanks to modern techniques the program must be structured completely (or
used to record what the subject is doing in vari- almost completely) in advance, before the move-
ous "stop signal" tasks. Among such measures ment can be initiated, and very little modification
are the EMG activity of the effectors (McGarry will occur in the movement for the next few hun-
& Franks, 1997), including activities in the motor dred milliseconds or so. We saw evidence for this
neuron pools in the cord (H-reflexes; McGarry, also in situations in which the limb was blocked
Inglis, & Franks, 2000), and various measures of from moving, with the pattern of EMGs being
neurophysiological activity (Band & van Boxtel, unaffected (as compared to unblocked moves)
1999). Sometimes actions are either c0mpletely for the next 100 ms or so (Wadman et al., 1979).
inhibited or not (i.e., they are performed as We also saw evidence that a startling stimulus
originally intended). In other instances, however, can "release" a complete, preplanned action far
researchers have identified responses that have earlier than would be the case if it was released
been interrupted, for example by means of a "normally" via the stages of processing (Valls-
partial reduction of motor neuron pool activation Sole et al., 1999). Another line of support for this
(e.g., McGarry & Franks, 1997). Findings such hypothesis is the evidence that certain variables,
as these have resulted in a controversy regard- related to the "complexity" of the movement to
ing the location of the point of no return-some be made (i.e., the number of limbs or movement
arguing that the location is at a high level of the segments, the movement's duration, or both),
central nervous system such as the cortex (Band & tend to affect the time between stimulus and the
van Boxtel, 1999; van Boxtel & Band, 2000), with beginning of the movement (i.e., the RT; Henry
others suggesting that it is at a lower level in the & Rogers, 1960). More complex, and longer-dura-
central nervous system (McGarry & Franks, 2000). tion (but still rapid) movements produce longer
Regardless of the exact location of the point RTs (Christina, 1992; Klapp, 1996). It has not been
of no return, a theory of how the process works possible to explain how these effects could occur
applies equally well. The theory, developed by except by the hypothesis that the movement is
Logan (1994; Band, van der Molen, & Logan, programmed in advance, with these variables
2003; Boucher et al., 2007; Logan & Cowan, 1984), affecting the duration of the stage necessary
regards the inhibition of action as a kind of horse for completing this preprogramming (Henry &
race. One horse, the "start horse," represents a Rogers, 1960; Schmidt, 1972b; Schmidt & Russell,
command that is sent to the motor neuron pool 1972). Many of these ideas are far from new; the
to execute the preprogrammed response. The original notion dates back to thinking by James
other horse, the "stop horse" represents a com- (1890) and Lashley (1917), and more recently to
mand sent to halt the execution. The horse that Henry and Rogers (1960), Keele (1968), Schmidt
reaches the "finish line" first (here, the lower (1976a), and Brooks (1979). Early thinking on how
centers involved with movement production) motor programs might be structured is credited
"wins" -thereby determining whether or not to Karl Lashley (see "K.S. Lashley on 'Motor Pro-
the movement is executed or inhibited. In this grams'" on p. 198; see also Rosenbaum, Cohen,
way, it is possible to model situations in which Jax, Weiss, & van der Wel, 2007).
total and partial inhibition and execution occur As a result of this thinking, at least two levels
(Boucher et al., 2007; McGarry, Chua, & Franks, can be distinguished in the motor system: (a)
2003; McGarry & Franks, 1997). Extensions of this an executive level (including the information-
research have been used to explain more complex processing stages) for selecting, organizing,
actions, such as "checking" a swing in baseball and preparing and initiating a complex pattern
batting (see "The Checked Swing in Baseball"). of muscular activities and (b) an effector level
Batting in baseball is a good example of a case in which inhibitory orc1ce:sse!S
In baseball it is not uncommon to observe "checked swings"-the
some point after the pitcher has released the ball, the batter decides
swing instead. If the "inhibit swing" decision occurs soon enough, the bat can
crosses the plate (after this point, if the ball is not contacted, the
swing and is called a strike). But how is this inhibition process carriedout? ·
Recall that Slater-Hammel measured only two kinds of behavior.-....:release
baseball batting, these two behaviors would be analogous to (a) the full
swing, or fully inhibited swing. However, studies of baseball batting have rA\rA~l!=!rlt\i~c'nnth1A'r•hir'1es
of swing behavior, called (c) partial response and (d) interrupted outcome
& Franks, 2003). The partial response occurs when the batter initiates bat ..,,,,.,,..,,""'..,"'+ *•"''"' '".w
contact point with the ball (above the plate) but with less velocity than. in the
lighter dashed line in figure 6.14). With the interrupted outcome, the swing is initiated
velocity as the full swing but undergoes a sudden alteration in kinema.tics prior to
(see the darker dashed line in figure 6.14).
Using a batting simulation task in which batters
were instructed to swing only at pitches that were
predicted to arrive in the strike zone, Gray (2002,
g-
2009a) found that "stop signals" were executed based -m 1
a.
on pitch predictions and three distinct types of ball 0
information. Gray modeled pitch predictions (pitch 'C
speed and location) using variables that baseball ,g
players frequently rely upon to make predictions about ~ -1
an upcoming pitch (e.g., the ball-strike count; loca-
tion and speed of recent pitches). The earliest ball
information used to determine if a pitch would arrive
out of the strike zone was the launch angle of the
ball as it left the pitcher's hand; this was used most
frequently for pitches that resulted in a fully inhibited
swing. The discrepancy between the predicted and
actual speed of the pitch as it approached the strike
zone was used in the majority of cases that resulted FIGURE 6.14 Sample oat displacernent,-timeprofitfu~torfc>µf:·. /
types of baseball batting swings: (a) full, (b) )ntercFup~d. (c) .' ·• ..··
in a partial response. And the discrepancy between
partial, and (d) fully inhibit@d. The;irrowrepres@f\tp.llJepOip:t < ·•
the actual and predicted horizontal height (above when the ball crossed theJ>late. FqP =!r?QlPf rnaj~L . . .
the plate) of the pitch was the information picked up Adapted, by permission, from R. Gray, 2009, ')\ model °'frrt~l~r iJlhit)ltl~~fu~~ c(/mptex<<' ..
skill~ baseball batting;• Journal of Eixperimf)ntal Psych<ilbl7k::il'ppfi~8j$::9'\~1C(5, ..
latest in the ball flight trajectory, resulting in the most
Copyright© 2009 by American Psycholo~i.calAss!Jci~ti9!1·, ••. · · ·
frequent occurrence of interrupted outcomes. 3
Of most importance for the batter, however, were . .
the results of these attempts to inhibit the swing-95% of the partial responses resulted in asud-
cessfully checked swing (i.e., stopped before the bat crossed the plate), wh.ereas only 41% the of
interrupted outcome swings were checked successfully. Therefore, different ball flight char;:lCtefistips
(speed and location) were most frequently associated with specific inhibitory processes (partial/
responses and interrupted outcomes, respectively), which dramatically affected the proba.billtypf
swinging at a pitch that was out of the strike zone. These analyses providean understal"l9ir:Jg,of ..
how inhibition processes are regulated in everyday, complex activities based. on the iriterlicti~hJ>l .·
visual information and prior expectations.

00 197
198 !Iii Motor Control and Learning

as.~~9~r~i~1vas<l~H:~\n6~~~i1'~!~0~,ffu~'peffotmanC:e ~'~ef:¥ .....· .,~ven1~n~s.

.
.. .Wdic~testheir independence ofp~rtentcOf1tto~.:·¥1gZsn(lPPin~· r1oveme~ts .of .~fle hand can be
a1sQ ·
.regul(lted accvratel~ in extent. yet the entire movement, from .initiation toQ~ll)pfetion, requires
less. than the reaction time for tactile o.r kin.esthetic stimulation of the arm, which. is about one-
eighth of a second, even when no disC:rlmination is involved. Such tacts force the conclusion
that an· effector mechanism can be preset or primed to discharge at a given intensity or for a
given duration, in independence of any sensory controls" (pp. 122-123).
He also said, "I believe that there exist in the nervous organization, elaborate systems of
interrelated neurons capable of imposing certain types of integration upon a large number of
widely spaced effector elements .... These systems are in constant action. They form a sort of
substratum upon which other activity is built" (pp. 127-128).

(motor programs and the muscular system) for Prior to the Movement
actually controlling or producing the patterns as One of the major roles of sensory information is
they unfold. We can further distinguish these two almost certainly to provide information about
levels by examining two distinct types of errors the initial state of the motor system prior to the
that can occur in performance, which we describe action. Consider this simple example: You must
later in the chapter. That said, however, we have know whether you are standing with your left or
already presented evidence that such a view can right foot forward in order to initiate a walking
explain only a limited set of movement situa- pattern (Keele, 1973). The spinal frog (figure 6.8)
tions, as many examples can be cited in which requires sensory information from the forelimb
feedback processes seem to interact with open- in order to direct the hindlimb to the elbow
loop processes in the production of movement. A
during the wiping response. Such information is
more complete approach to motor programming
presumably provided by afferent feedback from
would be to ask how the sensory processes operate
the various proprioceptors, and it would seem to
together with (i.e., cooperate with) these open-
be critical for the selection and adjustment of the
loop processes to produce skilled actions. We
proper action. We argued in chapters 2 and 3 that
tum to some of these ideas in the next sections.
these processes are very important for open types
Sensory Information and Motor of skills, for which the nature of the environment
is unpredictable or constantly changing.
Programs Polit and Bizzi (1979), using deafferented
The next sections deal with various functions of monkeys, showed that when the initial position
feedback in movement control. We can conceptu- of the shoulder changed prior to the elbow action,
alize these functions as acting before a movement, a systematic error in pointing to the target posi-
during a movement, and after a movement. tion occurred. This is understandable from figure
 Central Contributions to Motor Control 199

6.15, because changing the shoulder angle as the motor apparatus. In figure 6.16 are two dia-
shown necessarily affects the elbow angle (from 01 grams of a hypothetical rapid movement. In both
to 02) required for pointing at a target in a given cases, the movement involves flexion of the elbow
position in space. If the monkey programmed a distance of 45°, beginning with the arm straight.
a given elbow angle, then the equilibrium-point In figure 6.16a, the upper arm is positioned 45°
mechanism (chapter 7) would achieve that angle, below the horizontal, so that a flexion of the elbow
and the arm would not be pointing to the proper will result in the forearm's being horizontal at the
target. These monkeys did not learn to point end. In figure 6.16b, the upper arm is 45° above
to the target, even after considerable practice. horizontal, so that the forearm will be vertical
By contrast, normal, intact monkeys learned in at the end. The same command signal delivered
a few trials to compensate for the shifts in the to the biceps muscle group will not "work" in
shoulder position. The interpretation is that the both situations, for two reasons. First, a force is
intact animals had feedback from the shoulder required to hold the forearm against gravity at the
joint and could adjust the angle at the elbow to target position in the first situation, but not in the
compensate for the change in the shoulder angle. second. Second, more force is required to move
Thus, these data suggest that feedback about the the forearm against gravity in the first example
initial positions of the joints is required when relative to the second. A logical conclusion from
pointing to a position in space if the environment this simple example is that the motor system
is not perfectly predictable. must "know" the position the shoulder is in prior
Another role of afferent information involves to the action so that the command to the elbow
what a number of authors have called functional flexors can produce the required 45° movement.
tuning (Fitch, Tuller, & Turvey, 1982; Turvey, 1977). How this happens is not entirely clear, but that it
Recall that the spinal apparatus and resulting happens seems nearly obvious.
limb force output could be affected by change in Consider another complicating factor facing
the head position, much as would be expected on the motor system in producing a movement.
the basis of the idea that the tonic neck reflex was Figure 6.17 is a schematic diagram of the muscle
involved in the action (Hellebrandt, Houtz, Par- attachments involved in a simple movement. This
tridge, & Walters, 1956). In this example, afferent time, imagine that the movement is an extension
information from the neck influences the spinal movement in which the elbow is to be moved
mechanisms prior to action, thereby facilitating or through 45°. Notice thatthe triceps muscle, which
inhibiting them. But a more compelling reason for is the primary elbow extensor, is attached to the
assuming that premovement tuning must occur humerus in two places (internal and external
relates to some simple facts about the nature of heads) and to the scapula of the shoulder area (the

Normal position Elbow forward

t
FIGURE 6.15 In pointing to a target, the equilibrium point of the elbow is dependent on the angle at the shoulder.
JOURNAL OF NEUROPHYSIOLOGY by M.H. Polit and E. Bizzi. Copyright 1979 by AMERICAN PHYSIOLOGICAL SOCIETY. Reproduced with permission of AMERICAN
PHYSIOLOGICAL SOCIETY in the format Textbook via Copyright Clearance Center.
200 Ill Motor Control and Learning

FIGURE 6.16 Two 45° elbow flexion movements that appear to require different commands for the action and different
forces at their end points because of the effects of gravity.

long head). Thus, the triceps muscle performs is involved and produce some compensatory
two actions when it contracts: It extends the stabilization. The amount of stabilization will be
elbow and it tends to extend the shoulder joint, dependent on the shoulder angle because of the
pulling the humerus back. Therefore, when the length-tension relation (chapter 7).
triceps is contracting to produce the 45° move- The picture that emerges from these obser-
ment, one of the muscles that flexes the shoulder vations is that a "simple" 45° movement of the
must contract so that the shoulder joint is stabi- elbow joint is not really that simple at all, at
lized and only the elbow moves. Thus, during least in terms of the motor system. In addition,
this simple extension movement, the motor other complicated aspects of the muscle need
system must "know" that a two-jointed muscle to be considered by the motor system, such as

Elbow extension

Top view

Shoulder
) horizontal
extension

FIGURE 6.17 Complexity in a 45° elbow extension movement caused by the fact that the triceps muscle both extends the
elbow and horizontally extends the shoulder.
 Central Contributions to Motor Control 111 201

the nonlinear relationship between the muscle perhaps to reduce the errors made on the present
force and limb velocity, together with aspects trial. As such, this information has a considerable
of the contraction process that make the motor relevance to the acquisition of skills, as discussed
system very difficult to predict and control (Par- in the final part of this book dealing with motor
tridge, 1979, 1983). Weallknowthatournervous learning (chapters 12 and 13 in particular).
system controls our limbs beautifully in these
simple" situations. How it does so is exciting to
11
Types of Motor Program Errors
ponder.
If, as we suspect, the motor system can be con-
One role that feedback seems to have during
ceptualized as consisting of two major parts (an
movement production is a monitoring function,
executive level and an effector level), we can ask
whereby the feedback from the movement is
about the origins of errors in each-including
taken in and processed but not necessarily used
their causes, their detection, and their correction.
in the control of the action unless something goes
We conceptualize the motor system as being
wrong. It is probable that a long string of actions
capable of making essentially two distinct types
dealing with finger movements in piano play-
of errors. Each of these errors involves feedback
ing is programmed and carried out open loop.
in distinctly different ways for their detection and
Feedback fro)TI the fingers is returned to the cen-
tral nervous system for analysis, as if the central correction. We discuss these two kinds of errors
nervous system were checking" for errors. If no
/1
in the next sections.
errors appear, then the feedback is ignored. But if When a person makes a rapid movement,
the feedback indicates that an error has occurred, there are really two goals (Schmidt, 1976a). First,
attention can be directed to that feedback source, there is an environmentally defined goal, such as
and an appropriate correction may be initiated. changing gears in a standard transmission car
Reflexive corrections may also be generated, as or doing a somersault from a diving board. A
discussed in chapter 5. second goal (or subgoal) can be defined in terms
A second way to view feedback is that it may of the muscular activities required to produce the
be intricately involved in the physical control of desired outcomes in the environment-that is,
the limb. We mentioned a number of examples to produce the first goal. For example, a person
of this in the preceding chapter. The possibility must contract the muscles in the arm and torso
exists that a constantly changing reference of in one of a limited number of ways in order to
correctness is specified by the gamma motor change gears smoothly, and only certain patterns
neurons to the muscle spindles and that their of muscular activity will result in a somersault.
actions result in a continuous set of corrections Essentially, how to generate such a subgoal is the
to keep the movement on the proper course. The problem facing the performer.
feedback could be involved in the determination We can view the subgoal as a pattern of action
of the end location of a movement if the reference that is structured in both space and time. Such a
of correctness were set for this position. And in pattern of action will determine where a particu-
repetitive movements, the feedback from early lar part of the body will be at a particular time
segments of the sequence can provide adjust- after the movement starts. If this spatial-temporal
ments for the later segments. pattern in the muscles and limbs (the subgoal)
is produced accurately, then the environmental
Following the Movement goal will be achieved. This process can go astray
Extensive feedback is also delivered to the central in essentially two ways.
nervous system after a movement. Such informa-
tion can be evaluated, presumably, by the stages Errors in Program Selection
of information processing in order to determine Given the assumptions about the spatial-tem-
the nature (or "quality") of the movement just poral goal, the first kind of error that the person
made. Information about whether or not the might produce can be defined as an error in
movement achieved the environmental goal and program selection. This kind of failure to achieve
about its smoothness, its level of force or effort, or the environmental goal results from the per-
its form or style is derived from sensory feedback. former's choice of action. This error can happen
A major role for such information is in the adjust- in a number of ways (Reason & Mycielska, 1982).
ment of the movement on the subsequent trial, First, the person can produce the wrong pattern
202 Motor Control and Learning

of action: for example, heading versus kicking a unexpectedly, slowing the action and making it
ball in soccer, moving right when a left move is miss its mark. Many other examples can be used
appropriate, or moving when it might be impor- of course, including one that occurs occasionally
tant to stand still. The outcomes are the result in driving (see "Pedal Misapplication Errors").
of apparently intentional decisions about what These influences do not make the originally
to do and, in our terms, represent inappropriate intended movement pattern incorrect, as some
choices of a program of action. Such errors can compensation that will achieve the originally
arise from a variety of sources, such as a faulty planned spatial-temporal goal will still result in
perception of the environment (e.g., the speed the achievement of the environmental goal. Thus,
of an oncoming car by a pedestrian crossing the the correction for an error in movement execution
street) or an error in the choice of a goal (e.g., to may not require a new motor program, since the
pass to teammate A when a pass to teammate B original pattern of action defined by the "old"
would have been better). program will be correct if the motor system can
How does a person make a correction for an compensate for the unexpected environmental
error in program selection? According to the influences. This implies that because the system
evidence presented earlier, the person must issue does not have to select a new motor program,
a new motor program, as the "old" one will not the correction for an error in execution does not
achieve the goal in the environment. Hence, the require all the stages of information processing
information-processing stages must be initiated and will be far more rapid than correcting for an
again; a new program must be selected in the error in program selection.
response-selection stage, and it must be pro- What is the evidence for this kind of correc-
grammed in the response-programming stage. tion, and is the correction for error in execution
Because all of these stages are relatively slow, the fundamentally different than that for an error in
result is that a new pattern of action in a rapid selection? Consider the example from Dewhurst
motor skill usually cannot be selected before the (1967; figure 5.16) presented in the preceding
movement has been completed and the error has chapter; recall that the subject was instructed to
occurred. If the movement has a somewhat longer hold a weight with the elbow at a right angle.
MT, however, then a correction for an error in When weight was suddenly added, a correction
selection is possible during the action. followed in the biceps EMG within about 30 ms,
and a more sustained correction followed in about
Errors in Program Execution 50 to 80 ms. The corrections were far faster than
An error in program execution is fundamentally can be explained by the production of a new
different from an error in program selection program of action; hence it seems reasonable to
(Schmidt, 1976a). An error in execution can occur believe that the original program of action was
if the person produces a program of action appro- modified in some way. More importantly, the
priate for the environment but some unexpected person did not have to select a new program of
event occurs that disrupts the movement. This can action to compensate for the added weight, as the
happen, for example, if the contractions specified original spatial-temporal goal was still appropri-
by the motor program are not quite achieved by ate; the goal before the weight was added was
the muscles, perhaps because of inconsistencies "Hold the elbow at right angles," and the goal
in the spinal cord where it is determined which of afterward was still the same; the subject seemed
the many motor units are to be activated. The bas- only to require additional muscular tension in
ketball player attempting a free throw shot selects order to comply with the "old" goal. Thus, it
the proper action (making no selection error), appears that the corrections served the purpose
but random noise in the central nervous system of maintaining the original pattern of action and
often causes the shot to miss its target because did not result in the generation of a new one.
of an error in execution. Sometimes the mass or Consequently, the corrections had a far shorter
friction of the objects in the environment is not latency than would be expected if it had been
estimated quite accurately before the action, such necessary to produce a new pattern (see chapter
as when you pick up a nearly empty milk carton 5 for other examples).
that you thought was nearly full; the initial part What about the role of consciousness in cor-
of the action is much too rapid. Sometimes the rections for errors in selection? As you will recall
resistance offered by the environment increases from chapter 5, Henry (1953; figure 5.17) asked
 Central Contributions to Motor Control 203

Pedal Misapplication Errors

The concept of two types of errors (selection and execution) has a wealth of applications to real-
world situations, one of which is automobile accidents. Consider the following situation.
A driver enters his car (with automatic transmission) and prepares to drive forward. He reports
that he started the car, placed his right foot on the brake, and shifted the transmission to Drive. The
car roared away under what seemed to be full throttle, and the brakes were completely ineffective
in stopping the car ("The brake pedal went to the floor") until the car struck a wall after about 8 s of
travel. The driver claims that the car malfunctioned, but a subsequent inspection of the vehicle (by
experts, car makers, three federal government agencies, or independent research firms) reveals
nothing wrong with the car. Incidents essentially like this have occurred more than 10,000 times
since the mid-1980s and are now referred to as unintended acceleration cases, involving pedal
misapplication errors. What can be going on here?
A leading theory to explain these events is that the driver in some way placed his foot on the
accelerator pedal rather than the brake (see Reinhart, 1994; Schmidt, 1989b, 1993). But why would a
driver with, in some cases, 50 years of driving experience choose the wrong pedal after never having
done so before? The important word in this question is "choose." It is certainly difficult to imagine
that the driver made an error in the choice of the accelerator pedal rather than the brake-what we
call here a selection error-and, in fact, the driver emphatically claims not to have done so.
Rather, imagine that the driver intends to press the brake as usual, and the executive level pro-
grams and initiates an action to do so. The executive passes control to the program, and the program
controls the muscles and limbs, while the executive shifts attention to vision or other sources of
information relevant for driving. However, (a) because the driver is slightly misaligned after having just
entered the car or (b) because of variability in the neuromuscular processes (see chapter 7, impulse
variability)-or because of both-the foot on this occasion turns out to be aimed at the accelerator
rather than the brake that was the intended goal, and a pedal error occurs. This represents an error
in execution, as the correct action was selected and produced but it was executed or aimed (or both)
inappropriately. Now, when the shift lever reaches Drive, the car lurches forward, and the startled
driver's response is to press harder on the "brake." But, because his foot is actually on the accelera-
tor, the car goes faster, he presses harder, it goes even faster, and so on, until the car hits the wall.
In these situations, the driver "knows" (incorrectly, however) that his foot was on the brake because
that is where the executive level ordered the foot to go. Lower-level processes have caused the error,
and the executive level has no information about them. The driver's response to the unexpected
acceleration is to make a very fast correction at the execution level (to press the "brake'' harder), a
kind of triggered reaction that actually makes the situation worse. Because of the hypervigilance
(or "panic"; see the related discussion in chapter 4) associated with the unexpected acceleration,
he does not consider that the foot is on the accelerator, hence does not make any of a number of
actions that would stop the car (e.g., switch off the ignition, shift to Neutral, etc.), which would have
required initiation of a new action (see also Schmidt, 2010).
Such views are now commonly accepted in examples of unintended acceleration (e.g., Depart-
ment of Transportation, 2000; Lococo &Tucker, 2011 ). These analyses of such automobile accidents
have involved our current understanding of the separation between errors in selection and errors in
execution, the factors that control their corrections, and the separation between levels of hierarchical
control. These concepts are critical in a number of analogous real-world situations.

subjects to try to maintain the position of a lever changes they could consciously detect. The
in response to unexpected changes in the pres- subjects were responding to changes in position
sure it exerted against the hand. He showed that that they could not perceive-that is, they were
subjects were able to compensate for changes in responding unconsciously. Also, Johansson and
position that were some 20 times smaller than Westling (1984; Westling & Johansson, 1984)
204 Motor Control and Learning

showed that if subjects began to lose grip on an in chapter 4 and that this process will interfere
object held between the fingers, compensations greatly with certain other (cognitive) processes
(in the fingers and in the elbow) could be made attempted at the same time; hence only one such
in approximately 30 ms, apparently without correction can be done at a time. Correcting for an
conscious awareness. In these examples, the error in execution, on the other hand, is automatic
person did not have to select a new program of with respect to cognitive information-processing
action when the stimulus occurred, because the activities, and many such corrections could pre-
original pattern of action was still acceptable. sumably be done at the same time in parallel and
Force changes within the context of this pattern without interference-some in the arms, some
of behavior were required, however, to maintain in the legs, and so on. Hick's law clearly applies
the movement's overall goals. These were accom- when one is correcting for errors in selection
plished very quickly and without awareness, (chapter 3), with the latency increasing as the
and can be thought of as corrections for errors number of possible corrections increases. For
in execution. errors in execution, on the other hand, the number
of possible errors is probably not a factor, and so
Program Selection Errors Versus
Hick's law would not be expected to apply. 4 All
Execution Errors these differences, taken together, clearly argue
Table 6.1 summarizes some of the fundamental that corrections of motor program errors are of
features of errors in program selection and pro- at least two fundamental types.
gram execution, listed so that their differences
can be seen more easily. These differences are Triggered Reactions
important, because without testable distinctions The classification scheme in table 6.1 is now rather
between these error categories it would make little old (Schmidt, 1976a, 1983), and newer research
sense to consider them as separate classes (see also suggests it may be somewhat too simple to
classifications of reflex responses in chapter 5 for account for all the evidence. One good example
a similar analysis, especially table 5.1). involves triggered reactions, as discussed in
From table 6.1, we can see that the latencies of chapter 5. We saw that triggered reactions were
the two kinds of corrections are quite different; faster than RT, did not seem to require conscious
the selection errors require 120 to 200 ms or more processing, and did not seem to involve the selec-
in order for a correction to begin, and the execu- tion of a new movement program-all of which
tion errors can be corrected far more quickly, in would at first glance seem to place them into the
30 to 50 ms. Also, a new spatial-temporal goal is category of corrections for errors in execution.
needed to correct an error in selection, whereas But the notion of errors in execution implies that
the original pattern of action can continue while the correction serves to bring the limbs back to
an error in execution is being produced. We know the original spatial-temporal goal, or trajectory
that selecting and initiating a new motor program originally selected, after a perturbation (Cooke,
(needed to correct an error in selection) requires 1980). Yet the evidence on triggered reactions
attention and consciousness in the sense defined shows that the response to various perturba-

Latency of correction? 120-200 ms 30-50 ms

Old spatial-temporal goal OK? No Yes
New program selected? Yes Noa

Attention required? Yes No

More than one at a time? No Yes
Hick's law apply? Yes No?
'Provided that the deviation from the spatial-temporal goals is not very large.
Adapted from Schmidt 1983, 1987.
 Central Contributions to Motor Control 111 205

tions is a new pattern of action, with a trajectory trajectory quickly while seeming to maintain
fundamentally different from the one that was the overall movement goal in the environment.
occurring before the stimulus. When a given Feedback from the touch receptors in the skin can
perturbation was applied, we saw altered lip also modify the ways in which the gamma loop
and tongue trajectories in human speech (Abbs functions in movement control (Merton, 1972).
et al., 1984; Kelso et al., 1984), new hindlimb tra- This large body of evidence suggests a cen-
jectories in cat locomotion (Forssberg et al., 1975), trally organized structure that is capable of han-
and different, coordinated patterns of elbow and dling most of the details of the action but is also
finger movements in lifting tasks (Johansson & very sensitive to movement-produced sensory
Westling, 1984; Westling & Johansson, 1984), all information from a variety of sources. One way
with very short latencies. to view this blending of open- and closed-loop
It is tempting to suggest that triggered reac- functioning is to consider a hierarchical control, in
tions ensure that the original goal of the action is which a higher-order, open-loop control structure
achieved-not necessarily that the original trajec- has "under" it a set of closed-loop processes that
tory of the limbs is achieved. This suggestion ensure the movement's intended goal in the face
implies that a particular trajectory of the limbs may of various perturbations. If a signal appears in
not be as important as achieving the environmen- the environment indicating that the higher-order
tal goal. Of course, there are many ways in which program is no longer relevant, the highest levels
the motor system can achieve a particular envi- in the system (the stages of processing) become
ronmental goal; and when perturbed, the system involved in stopping it, or initiating a different
seems to shift from one of these alternatives to program, or perhaps both. But if smaller pertur-
another, with a very short latency. This capability bations occur that do not involve an alteration
to solve a problem with several action patterns is in the fundamental movement goal, these can
often ter:tned motor equivalence (see figure 6.18, p. be handled by lower levels in the hierarchy, pre-
209). This combination of features seems to suggest sumably while the original higher-level program
that triggered reactions fall somewhere between continues to operate. This is a classic example of
correction for errors in execution and correction one form of hybrid control in which a closed-loop
for errors in selection, sharing features of both cat- system is embedded within an open-loop system.
egories or perhaps even forming a third category. These thoughts lead to a modified, less
It may be that when the perturbation is small, a restricted definition of a motor program, one
correction for an error in execution can occur to that is in keeping with the literature on feedback
bring the limbs back on the target trajectory; if process but yet retains the essential feature of the
the perturbation is somewhat larger, a triggered open-loop concept: The motor program is an abstract
reaction results in another trajectory, but without representation ofaction that, when activated, produces
the need for reprogramming the movement using movement without regard to sensory information
the stages of information processing; and if the indicating errors in selection. Once the program
perturbation is even larger, a correction for an error has been initiated, the pattern of action is car-
in selection is generated, which of course involves ried out even if the environmental information
the stages of processing. indicates that an error in selection has been made.
Yet during the program's execution, countless
Hierarchical Levels of Control corrections for minor errors in execution can be
From the previous sections it is clear that there produced that serve to ensure that the original
is considerable evidence for a central open-loop goal is carried out faithfully. Grillner (1975) has
mechanism, structured before the movement is said essentially the same thing with respect to
initiated, that serves to organize and control limb the control of gait:
movements in coordinated actions. Yet substan- Perhaps it is useful to regard the relevant reflexes as
tial evidence also suggests that feedback from prepared [italics added] to operate but without any
!he responding limbs can, through a variety of effect as long as the movement proceeds according
mechanisms, modify the movement in various to the set central program. At the same instant when
ways. Some of the lower-level reflex activities the locomotor movements are disturbed (small
serve to keep the movement on track," and
/1
hole, a slippery surface, etc.) the reflexes come into
triggered reactions and reflex reversals alter the operation to compensate. (p. 297)
206 ~ Motor Control and Learning

This idea is similar in many ways to the actions are defined by the initial conditions of the
concept of a coordinative structure discussed by limbs, the posture of the performer, the relations
Greene (1972), Fitch and colleagues (1982), Turvey with respect to gravity, and a host of other factors
(1977), and Berkinblit and Feldman (1988). In of which the highest level of the system is not
both the motor program and coordinative struc- aware. These lower functions then determine the
ture concepts, the many degrees of freedom in ultimate movement that will result, on the basis
the musculature are reduced by a structure that of these lower-level interactions with feedback,
organizes the limbs to act as a single unit. Also, tuning, and other factors. In short, the "ballpark
both notions involve the tuning of spinal centers, response" becomes increasingly more specified
corrections for errors in execution, and freedom of at each lower level in the motor system.
the executive level from the details of what occurs
at lower levels in the motor system. Problems With the Motor Program
Similar ideas along these lines were expressed Notion
many years ago by Bernstein (see "Nikolai Ber-
nstein" on p. 8). In one of his works, Bernstein The advantage of the motor program notion as
(1996) described in detail a four-level, hierarchical a theory of movement control is that it provides
system that was responsible for providing the order to a large number of separate findings, such
capability for dexterity without overwhelming as the inability to use certain kinds of feedback
the role of consciousness. Level A in Bernstein's and the kinds of corrections that can and cannot
scheme refers to neck and trunk tone, responsible be made. But the ideas about programs that we
for the automated control of posture. Level B (the have stated so far have other logical drawbacks
level of muscular-articular links) refers to the that must be considered. The next section deals
level at which coordinative structures, or syner- with two of the most important: the storage prob-
gies, operate (Turvey & Carello, 1996). The higher lem and the novelty problem.
levels-C (space) and D (actions)-are devoted The Storage Problem
to the problems of perception and intention,
respectively (Turvey & Carello, 1996). Given that an animal can produce a motor pro-
Greene's (1972) point of view was similar to gram "on command" and initiate it, how many
Bernstein's and emphasized the hierarchical such programs must the organism have at its
nature of motor control. Greene suggested that disposal in order to move as it does? Recall that
at the highest levels of the system, the global a motor program is thought to result in com-
aspects of the movement are represented in the mands to muscles that define a particular pat-
form of a goal (e.g., shoot a basket). The control of tern of action. In this view, if the pattern is to be
movement is passed down through progressively changed (e.g., from an overhand to a sidearm
lower levels until all the particular decisions throwing pattern), then a totally new program
about which motor units to fire are defined at the must be produced. Imagine all the ways to pro-
muscle level. The higher levels in the system do duce a throwing action, each of which must have
not have any direct control over muscle contrac- a separate program.
tions; they have control only over adjacent levels MacNeilage (1970) identified this problem in
of control that eventually result in those contrac- the context of speech production. According to
tions. This idea is related to the motor program programming theories of speech, each sound that
view, stated earlier here, in which only two levels a human can produce (called a phoneme) is gov-
exist-an executive and a program or effector. erned by a separate program; in order to speak,
Greene's view suggests that there are more than we string together these separate programs in a
these two levels. way that follows the "rules" of intelligible speech.
Along these lines, the highest level specifies This solution seemed to be a good one- since
what Greene called a "ballpark" movement, there are only about 44 sounds in English, we
which would result in any of a number of move- should require only 44 programs. The difficulty is
ments that are generally capable of producing that the actions of the mouth, jaw, tongue, and so
the intended goal; they are "in the ballpark," as forth for a particular sound are different depend-
Americans would say. As the system passes con- ing on the sound that precedes it, as well as the
trol to lower levels, the individual details of the sound that is to follow it. That is, to make the
 Central Contributions to Motor Control 207

sound of at, the musculature must make one of a movement like this: Beginning in a standing
two different movements depending on whether position, jump up from both feet, touching your
the word is eat" or "boat," as you can easily dis-
/1
head with the right hand and your leg with your
cover for yourself when you say these two words left hand before you land. Certainly, most of us
and note the actions of your own tongue. And, could do this on the first try. If you had never
depending on the previous sound, the program done that particular movement before and if the
for the next sound must vary for similar reasons. action required a previously learned program for
Thus, the 44 programs for tongue movement for its execution, then where did the program come
the various sounds must be multiplied by the from? It is difficult to assume that the program
number of different sounds that could precede was genetically defined (as walking might be),
and follow each of these sounds. This notion- because such an action does not seem particularly
called context-conditioned variability-led Mac- essential to our survival. And you could not have
Neilage to estimate that a very large number of learned it through practice, as this was the first
programs must be stored in memory in order for time that you produced this action. A logical
us to speak as we do. Considering all the various dilemma arises about the motor program notion
accents, inflections, and combinations, as well as when we consider novel movements.
any foreign language sounds, he estimated that The same sort of problem exists for more
about 100,000 programs would be required for common skills. If you were to study a series of
speech alone. 50 shots in tennis, examining the fine details of
It is possible that the brain can store 100,000 the feet, hands, and body, you would probably
programs for speaking, of course, as long-term find that no two movements were exactly the
memory has a very large capacity. But when we same. This is compounded by the fact that the
consider the number of ways in which we move ball never has exactly the same velocity, the same
other than for speech, and the interaction of previ- location on the court, or the same height on two
ous and subsequent states for each of these move- occasions. Therefore, it is unlikely that any two
ments, there would have to be a nearly countless tennis strokes could be exactly the same. If no
number of programs in long-term memory. How two shots are exactly the same, then the programs
(and/or where) are all of these programs stored must also be different. Thus, according to this
in the central nervous system? analysis at least, every shot is "novel" in the
This original motor program notion stated sense that it has never been produced in exactly
earlier seems unwise for several reasons. First, that way before. When you make a movement,
many mechanical or electronic control systems you do not simply repeat a movement that has
can have a serious storage problem, and it is been learned earlier.
crippling to them; examples are libraries that On the other hand, a given golf or tennis stroke
have to cope with tons of paper and computer is certainly very similar to strokes that you have
systems that have to store programs for every made in the past. For example, some people have
kind of computation. They simply run out of an atypical style of hitting a golf ball that is char-
room for storage. A second reason relates to the acteristic of them and no one else, and a popular
belief that our motor system evolved in such a touring (golf) professional's style is easily recog-
way that it was simple to operate and efficient nized (e.g., Jim Furyk or Ryan Moore). Thus, it is
in terms of storage. To store a complex program not fair to say that every golf stroke is absolutely
for every movement is not a simple and elegant new, as considerable practice and experience
way for a system to have evolved (e.g., Schmidt, have led to the production of that action, and this
1975b; Turvey, 1977). There must be a better way experience tends to make the actions similar and
to conceptualize the storage of motor programs. characteristic of that individual.
Writing more than 70 years ago, Bartlett (1932)
The Novelty Problem made the following observation about tennis
The next concern about this original motor pro- strokes: "When I make the stroke I do not, as a
gram notion is related to the storage problem, but matter of fact, produce something absolutely new,
it takes a slightly different form. The basic issue and I never repeat something old" (p. 202). His
is not how we retain learned movements, but point summarizes the issues in this section very
rather how we make new movements. Consider well. When making a stroke, you do not make a
208 Motor Control and Learning

movement that is absolutely new, because that A Computer Model

movement will depend on your past learning.
But you do not exactly repeat an old movement Perhaps the best example of a generalized pro-
either, as any particular movement will be slightly gram comes from computer science. Consider
different from all the others that you have made a program that calculates means, such as the
in the past. In this sense, the stroke is considered function" feature in Microsoft Excel. This type of
fl

novel, yet at the same time, dependent on previ- program is generalized" in such a way that it will
fl

ous experience. calculate the mean for sets of values that differ
One weakness of the earlier idea about the in terms of the number (n) of them to be consid-
motor program notion is that it does not explain ered. Note that the set of operations that is to be
how the individual can produce a novel move- done with these values is fundamentally similar
ment or how a movement such as a particular regardless of the number of scores we have; that
tennis stroke is somehow slightly different, yet is, add up all of the values and then divide by the
characteristic of all earlier ones. If our theo- number of them you have. To run the function
ries about movement programs are to have an you simply specify the number of data points
application to everyday motor behavior, then over which the mean is to be calculated-the
they must be able to explain these common number of values is one of the parameters that
phenomena. must be applied to the execution of the statistical
function. Once these are specified, the program
The Need for Revision is executed for this particular number of scores.
These two rather persistent problems-the stor- How does this kind of program solve the
age problem and the novelty problem-pose storage and novelty problems? First, the stor-
rather severe limitations for the motor program age problem is reduced because, for this class
idea as it has been stated previously. One solution of computing problem, only one program needs
has been to introduce a modification to the funda- to be stored in the system for each function (i.e.,
mental programming notion, one that retains all calculating the mean), and each program can
the attractive aspects of programming that have generate its results for a wide range of possible
been discussed so far but also provides a solu- numbers of data points. So, rather than having
tion to the two problems. This kind of thinking separate programs to calculate the mean of 2, 3, 4,
led to Schmidt's idea (1975b, 1976a) that a motor 5, ... , 1,000 numbers, we need only one program
program should be considered as generalized. that is generalized; providing it with a parameter
value (e.g., 8 or 800 data points, or some differ-
ent number) enables the generalized program to
Generalized Motor produce the result easily.
Programs With respect to the novelty problem, notice that
the program for means and standard deviations
can produce results for data that it has never been
The idea of a generalized motor program (abbre-
used for previously. One simply specifies the
viated GMP) is that a motor program for a
proper parameters, and the program is executed
particular class of actions is stored in memory
perfectly. The program can compute the mean for
and that a unique pattern of activity will result
a number of scores that it has never confronted
whenever the program is executed. In order for
previously. In this sense, the generalized program
the program to be executed, certain parameters
provides one kind of solution to the novelty
must be supplied to the program that define how
problem-it can produce a result it has never
it is to be executed for that particular instance.
produced before.
The program is considered generalized because
choosing different parameters will alter the
output, in terms of movements of the limbs,
Invariant Features
but these movements will differ in only certain A motor program is thought to be responsible for
ways. But, before we describe how such a system the production of a pattern of action, expressed in
might operate, it will be helpful to consider an both space and time. When we examine patterns
example of a generalized program for a different of action carefully, we see that various aspects
application. of them are easy to change while other aspects
 Central Contributions to Motor Control ijj 209

remain almost completely fixed from movement These handwriting samples are different in
to movement. It is not always obvious which obvious ways-they are of different sizes and
aspects of the movement are fixed and which are show a different amount of shakiness" in some 11

easily changed; but examining the movement in cases. The MT taken to complete each word was
certain ways, or with certain theoretical biases, probably not the same either. But in all samples
can reveal these features (Schmidt, 1985). for each individual there are many remarkable
A classic example of ways in which movements similarities. A certain style" is seen in all of them,
11

demonstrate both fixed and modifiable features such as the little curl at the start of the m for the
is one of our most common movement patterns, subject illustrated in figure 6.18a, and the way
handwriting. This demonstration was presented the downstroke of the q is made by the subject
many years ago (apparently independently) by in figure 6.18b. Some aspects of these written
Lashley (1942; see Bruce, 1994) and Bernstein words appear to be invariant, even when the
(1947; reproduced in Keele, Cohen, & Ivry, 1990 effector used or the size or speed of the writing is
[their figure 3.5]), and more recently by Merton changed. In this figure it appears that the spatial
(1972) and Raibert (1977). All these demonstra- pattern was invariant-the shapes of the letters.
tions suggest basically the same thing. Figure Lashley noted:
6.18 is a reproduction of the handwriting samples
published by Lashley (1942). Two right-handed, In spite of the clumsiness, the general features of
blindfolded subjects wrote the words "motor the writing, individual differences in the forming
of letters and the like, are characteristically main-
equivalence" 5 normally (with the right hand),
tained. The mechanics of writing is a sequence of
with the nondominant (left) hand, and with either movements in relation to bodily position, not a set
hand attempting to produce a mirror image of pattern of special groups of muscles. (1942, p. 317)
the words (these have been reversed in the figure
to appear as normal). The subject represented in You can see a similar phenomenon if you do
figure 6.18a even wrote the words with the pencil this yourself. First, write your signature in your
held by the teeth. "normal" way, as if signing a document or a

Right hand

Left hand

Right mirror
(reverseq)
...__ "\
\fW'\4Tl.
~ L<l.Au J
- \1 ---v
.......
~
t . .. . _ +- . .
<JllO-<.O""(...
9 ~-' 1 ,... ,,.. /)..,, __,....
""1J V\.VUA-C..n./ ~

Left mirror
(reversed)
_ _\ I\.-'\ .
~"' ~~~
/1 " " . J\.~. . ~

Teeth

a b

FIGURE 6.18 Examples from two subjects writing the worc:!s "motor equivalence'' with different effectors.
From K.S. Lashley, 1942, The problem of cerebral organization in vision. In Biological symposia. Vol. VII. Visual mechanisms, edited by J. Cattell (Lancaster, PA: Jaques Cattell
Press), 301-322.
210 Motor Control and Learning

check. Second, write your signature roughly 10 musculature. These pulses produce patterns of
times larger on a blackboard with chalk. Just by contractions in the muscles that can be seen in
looking at the two signatures, you should be able EMG records or in records of force produced. The
to see that they are "the same," signed by the amount of force produced is related in a complex
same person. (If you want to be more elegant, way to the amount of neurological activity, and
take a digital photo of both signatures and enlarge the duration of the force and its temporal onset
or reduce one of them so that they are the same are determined by the duration of the neurologi-
size, and then overlay them.) Now, note that cal activity and the time of its occurrence. The
you usually write the small signature with your major role of the motor program is to "tell" the
fingers, with the "heel" of your hand fixed to the muscles when to turn on, how much force to use,
writing surface; naturally, with your hand fixed, and when to turn off. Thus, the motor program
your elbow and shoulder are fixed also. But, in ultimately controls force and time.
the blackboard signature, your fingers are mainly Impulses
fixed, and the writing is done with the elbow and
shoulder. The "same" signature was written by The combination of (or the product of) force
completely different joints and muscles. and time constitutes what is called an impulse
Although the interpretation of these kinds of in physics. A common physical principle is that
demonstrations has been called into question the amount of movement produced in an object
(Latash, 1993, 1999), in-depth analyses support (such as a limb) is determined by the force(s)
the conclusion that something in the performer's acting on it and the duration over which the force
memory is common to all these handwritten acts; the impulse is the product of force and time.
samples (Wright, 1990). Some abstract structure Therefore, the impulse-timing view assumes
has expressed itself, regardless of the variations that the GMP controls movement by controlling
impulses-bursts of force spread out over time
in handwriting speed or size or in the particular
to the appropriate muscles.
limb or muscles used to write the words. Schmidt
In figure 6.19 are three hypothetical, idealized
(1976a) theorized that those features that are
records of the forces produced by a muscle over
invariant, and that in some ways are fundamen-
the time that this muscle is acting on the limb.
tal to these written words, are structured in the
At each moment of the contraction, the muscle
GMP; those aspects of the movement that are
is producing a different force against the bone;
relatively superficial (speed, size, the limb used)
the resulting curve in figure 6.19 is called the
are thought to be parameters of the program.
force-time curve-a record of the force produced
Remember the computer analogy: The ways in
over time. The impulse is the shaded area under
which the means are calculated is invariant and
the force-time curve. From mathematics, this area
fundamental to the program-the number of
is frequently called the integral, or the integral of
scores to be averaged is not fundamental, and
force over time.
this constitutes a parameter of the program. These
In the figure, notice that the area under the
examples with handwriting seem to be showing
force-time curve for impulse a can be reduced in
something similar and perhaps invariant.
half by changing the amplitude of the force for
If these observations are correct, how can the
a given amount of time (as in impulse b), or by
structure of the motor program be conceptual- changing the duration of the impulse for a given
ized so that the invariant features of handwriting
amplitude (as in impulse c), or both. From phys-
are held constant across a wide variety of other ics, the velocity of the object (or limb, beginning
changes? In the next section, we consider one at rest) after the impulse has ended its action will
possibility that has abundant evidence to support be directly proportional to the size of the impulse.
it-the impulse-timing view. Thus, impulses band c in figure 6.19 would theo-
retically produce the same velocity at the end of
Impulse-Timing View their respective actions (because their areas are
One straightforward viewpoint about the struc- equal). And the velocity of the limb with impulse
ture of GMPs is the impulse-timing view. The fun- a would be twice as large as for the other two,
damental idea is that the motor program provides because its area is twice as large. According to
pulses of motor neuron activity to the relevant this view, the motor program controls a feature of
 Central Contributions to Motor Control im 211

2 Invariant Features
and the Impulse-Timing View
Given a model of impulses patterned in time to
produce a movement, what features of the action
must remain invariant? What aspects of these
impulses are the same from one handwriting
a
sample to another, and which of them can vary
while maintaining a given pattern of activity? The
evidence favors two, and perhaps three, features
that remain invariant.
Order of Events One aspect of the patterns
shown in figure 6.18 that seems not to vary is
the sequence or order of events (Lashley, 1951). In
each sample, some event occurred before some
other event in making a letter or word, and this
order was fixed for all of the samples. We assume
2 that the order of muscular contractions for this
sequence of events in a given program is fixed. A
basic assumption of the impulse-timing view of
motor programming is that within the structure
of a given program is an invariant order of the
various elements. Notice that this is not the same
as saying that the order of muscles contracting is
c 50 100 fixed in the program. The muscles that produced
Time (ms) the writing with the teeth are obviously different
from those that produced the writing with the
FIGURE 6.19 Hypothetical impulses seen as the area hand, and yet the sequence and the pattern were
under force-time curves. (Impulses b and c have half the same. Clearly, the motor program does not
the size that impulse a does, but impulse b is achieved contain the order of muscles; rather, it seems to
by halving the force with time constant, and impulse c is
achieved by halving the time with force constant.)
order the events (here, upstrokes, downstrokes,
etc.) that occur during the movement.
Relative Timing A second aspect of the pro-
muscular contraction that is known to be a direct gram that is thought to be invariant is the tempo-
cause of movement-impulses. ral structure of the contractions, usually termed
If it is correct that the motor program deter- relative timing. The temporal structure of a series
mines impulses, then it is reasonable to assume of events (in this case, a series of actions) can be
that the motor program is capable of produc- represented in a number of ways, but one of the
ing a group of impulses, each one in a different most common is to evaluate the structure in terms
muscle group and each one at a different time, of what is called relative time. In figure 6.20 are
resulting in a pattern of activity that produces a hypothetical examples of records taken from two
complex, skilled movement. Remember, produc- similar actions. This particular record contains
ing impulses in muscles is really nothing more EMGs, but the record could have been defined
than defining the time of onset and offset (i.e., in terms of movements of the limbs, the forces
the durations) of the relevant contractions, as produced, or other characteristics that capture
well as their forces. Once these are defined, the in some way the nature of the movement pat-
movement is defined, given a fixed environment tern produced. The hypothetical muscles whose
of course. Even so, defining these impulse sizes EMGs are shown were chosen because they act
and durations should not be seen as simple, at different times in the movement sequence. The
because many factors must be considered by the sequence begins with a strong burst of EMG from
central nervous system, as discussed earlier (see muscle 1; then muscle 1 appears to be turned off
figures 6.16 and 6.17). and muscles 2 and 3 are activated, with muscle
212 Iii! Motor Control and Learning

FIGURE 6.20 Hypothetical electromyogram (EMG) records from two similar movements differing only in movement time
(MT). (Phasing, or relative timing, is defined by the ratios of the EMG durations among various muscles, e.g., b/c, c/a, and so
on.)

2 ceasing its activity before muscle 3 does. How and one-half times as long as interval c, making
can this temporal pattern of events in these three their ratio 1.5:1, or 1.5. Similar ratios can be com-
participating muscles be described? puted for any two intervals in the sequence.
One method is to measure the durations of the Another common ratio is that of an element
various elements within the sequence. Shown in in the sequence relative to the overall length of
the figure are two similar movements, but one the sequence. For example, in the movement
of them (movement 2) has a longer MT than the 1 sequence the ratio of interval d to the overall
other. If these two records are evaluated with length of the sequence (interval a) appears to be
respect to the durations of the relevant contrac- about 0.60; thus, muscle 3 is contracting for about
tions (EMGs), then interval a can be defined as 60% of the entire movement. This is the usual use
the duration of the contraction of the muscles of the term "relative time": the duration of one
in the entire action, interval b is the duration of event divided by the duration of the entire action.
contraction of muscle 1, interval c is the duration The fundamental idea of these ratios is this:
of contraction of muscle 2, and interval d is the The temporal structure is measured by (or char-
duration of contraction of muscle 3. One way to acterized by) the values of these ratios. If all the
evaluate the temporal structure of these events ratios for analogous pairs of events are the same
is to produce ratios of these various times. The in two separate movements, then we say that
sequence for movement 1 has a ratio of interval the temporal structures are the same. Thus, any
c to interval d of 1:2, or 0.50. That is, interval dis two movements with the same order of contrac-
twice as long as interval c. Also, interval bis one tions (perhaps that shown in figure 6.20) and the
 Central Contributions to Motor Control 111 213

same ratios of muscle action to total MT (e.g., before a guitar riff, and so on. Next is the relative
0.45, 0.30, and 0.60 for muscles 1, 2, and 3) have timing. Think of relative timing as the rhythm,
the same temporal structure. Further, these two so that the time between any two events on the
movements are assumed to be produced by the record divided by the total record time is a con-
same GMP, for which relative timing is invariant stant. For phonograph records, the ratio between
(or fixed). the durations of any two events is always fixed.
Movements 1 and 2 in figure 6.20 have this Also, the relative force is fixed. For example, the
characteristic. The proportion of total MT for first drumbeat may be twice as loud as the second
each muscle is the same in the two movements, one. What is embedded on the record is a code
even though the amount of time that each muscle that is translated into sound when the record is
is contracting is different for the two movements. played on an audio system. It may be helpful to
Movements 1 and 2 are thought to be governed conceptualize GMPs as records, because in many
by the same motor program because their relative ways they behave in the same general way, and
timing is the same. According to this idea, if two these similarities allow us to visualize the GMP
movements have different relative timings, then idea more vividly. 6
they are governed by different GMPs. But we know that the record can be played in
Relative Force A third important feature of various ways to produce different sounds. It can
GMPs is relative force, which simply means that be played rapidly or slowly, loudly or softly, with
the amounts of force produced by any two mus- the sound equalizer adjusted in various ways,
cles remain in constant proportion from move- and so on. Yet a given song can still be recognized
ment to movement. If in movement 1, muscle 1 because the pattern of the sounds produced is
produced 2 kg of peak force and muscle 2 pro- invariant, even though some of the superficial
duced 4 kg, the ratio of these two forces would be features of the pattern may have varied. The
1:2, or 0.50. In another movement using the same actual muscles that produce the action (here,
program, these proportions should be the same, the particular speakers that will be driven) are
but perhaps with forces of 2.5 kg for muscle 1 and certainly not "on" the record, because the record
5 kg for muscle 2. The ratio remains 1:2, or 0.50. can be played on any audio system. In the next
This feature (relative force) of the movement section, we discuss some of these more superficial
sequence would seem to remain invariant for the features of movements. These aspects of move-
ment are considered the parameters of the GMP.
patterns of handwriting in the examples in figure
6.18. We can see this in two ways. First, in this
Parameters of Generalized
kind of model, the height of a given letter is deter-
mined in part by the amount of force applied to Motor Programs
the limb during the impulse applied by the motor Motor program theorists have argued that there
program. But the heights of the letters remain in are a limited number of parameters that can be
almost constant proportion to each other as the applied to a GMP. The evidence is strongest for
various letters in a given sentence are considered. parameters that are responsible for establishing
For both subjects.in Lashley's example (figure the overall duration, the overall force, and muscle
6.18), the tis always about twice the height of the selection. We discuss these next.
o that follows it. The forces that produced these
letter heights may have been in constant propor- Overall Duration Parameter
tion in the sequence as well. The basic idea of an overall duration parameter
is that, while the motor program contains rela-
The Phonograph Record Analogy tive timing and sequencing information, it can
It is sometimes helpful in understanding motor be executed slowly or rapidly depending on
control theories to consider a model that has the overall duration parameter assigned, just as
many of the same features as the theory. A good increasing the speed of the phonograph turntable
model for the GMP is the phonograph record. speeds up the entire sequence of sounds as a unit.
On the record, structured as invariant features, Initial evidence for an overall duration
are three things. First is the order of the events, parameter appeared in an unpublished study by
for example, specifying that a drumbeat comes Armstrong (1970b). Subjects were asked to learn
214 Motor Control and Learning

to move a lever through a particular spatial- a task that required precise spatial-temporal
temporal pattern. Figure 6.21 shows a tracing of patterning of pronation/ supination of the wrist.
the position of the lever as a function of time in Thus, to be successful the subjects had to make
the 4 s movement. Armstrong noticed that when a series of actions with criteria defined in both
the subject made the movement too rapidly, the space and time. The temporal structure of the
entire sequence was made too rapidly, as if the action for Shapiro's (1977) study is shown in
entire movement record was "compressed," figure 6.22, here called the "test" trials. The
with all parts of the movement shortened by the proportion of the total MT (which was 1,600 ms)
same relative amount, or in the same proportion. occupied by each of the nine wrist-twist segments
Although Armstrong did not compute the pro- is plotted as the line marked with open squares.
portions suggested in figure 6.21, a critical test of After considerable practice, Shapiro asked her
the idea is that the time interval between some subjects to speed up the movements but to keep
peak a and some other peak b divided by the the pattern the same; the pattern of proportions
time for the entire movement is about the same for these "compressed" trials is shown as the
in the two movements shown in the figure. Such line with filled circles in figure 6.22. Notice that
findings gave initial insight into the possibility the proportions of time for the various segments
of an underlying GMP, with an overall speed were almost exactly the same for the "test" trials
parameter that retained the invariant relative and the "compressed" trials, but that the MT
timing in the movement pattern, which could in the latter was decreased to 1,300 ms, on the
be parameterized to allow the total duration of average. Essentially, Shapiro showed that the
the movement to be made longer or shorter (see subjects could decrease the time of this well-
Pew, 1974a, for an early discussion of this work). learned movement sequence as a unit, keeping
Following Armstrong's (1970b) and Pew's the relative timing in the movement (defined by
(1974a) suggestions, Summers (1975) and Shapiro the proportions) constant. Again, these findings
(1977, 1978) examined similar questions in tasks support the view that a movement-duration
in which the experimenter could instruct the parameter can be applied to some fundamental
subject to change the overall speed intentionally program so that the given pattern can be sped up
(remember that movement durations made by or slowed down as a unit.
subjects in Armstrong's study changed acciden- Even more remarkable was another finding
tally). Shapiro's paradigm involved practice at that both Summers (1975) and Shapiro (1977,

Left

Q .. 15

-"'
ca
0
Cl>
10
..r::
I.)

:S 5
c:
0
:,;:::; 0
"iii
0
0..
.:.:: 5
I.)
:,;:::;

-..
0"'
c:
0
(.)
10

15

Right 1.72 1.95 2.28 3.59

Elapsed time (s)

FIGURE 6.21 The position-time record of an arm movement task, showing the correct move and a move in which the
overall movement time (MT) was too short.
Reprinted from T.R. Armstrong, 1970, "Training for the production of memorized movement patterns,'' Technical Report 26: 35.
 Central Contributions to Motor Control 11 215

0.18

0.16

~ 0.14
:;::
0
s 0.12
t:
&.
e
0.. 0.10

0.08

2 3 4 5 6 7 8 9
Segment number in sequence

FIGURE 6.22 Proportion of total movement time (MT) required to traverse each segment in a wrist-twist movement.
(Normal trials had a goal of 1,600 ms; compressed trials were sped up using the same phasing; speeded-up trials were sped
up while subjects attempted to ignore the earlier-learned phasing.)
Adapted from Shapiro 1977.

1978) obtained. They asked their subjects to make the longest MT (1177 ms) is at the bottom. Notice
the movement as rapidly as possible and to ignore that the onset times of the various letters "line up"
the relative timing that they had learned in the on the sloped lines, as if the longest trials were
earlier practice trials. In figure 6.22, the line with simply "stretched" versions of the shortest ones.
open circles represents these "sped-up" trials; Notice also that the relative time of occurrence
again, the pattern of proportions was almost iden- of a given letter in the word "enclosed" is almost
tical to that for the normal trials. Subjects were constant from attempt to attempt.
able to speed up the movements, but they were Similar findings have been produced by
apparently unable (or at least unwilling) to do so Shaffer (1980, 1984) in a study of typing and
with a different relative timing (see also Carter & piano playing, by Roth (1988) using an overarm
Shapiro, 1984; Verwey & Dronkert, 1996). throwing movement, and by Heuer, Schmidt,
There are many other examples of absolute and Ghodsian (1995) using a bimanual task in
and relative timing in the literature. Terzuolo and which each arm produces a different pattern, with
Viviani (1979, 1980) studied the typing of various the two arms being coordinated. All these data
wordS, examining the relative timing characteris- support the notion that a given overall sequence
tics. Figure 6.23 has a diagram showing various can be sped up or slowed down as a unit while
temporal records in typing the word "enclosed," the relative timing in the sequence is maintained
when the word was embedded in various places (i.e., relative timing was invariant). Again, these
in standard prose. In figure 6.23a, the time of data are consistent with the view that the same
occurrence of each of the letters is plotted for 42 GMP produced the different instances of typing
different trials. Each horizontal row of dots repre- the word "enclosed" in figure 6.23 by applying
sents the time of occurrence for each letter for one a slightly different duration parameter each time
trial. The trials are presented in the same order in the word was typed.
which they occurred in the experimental session, One more type of research paradigm has pro-
and no recognizable pattern of relative timing vided evidence that is important to consider. A
appears. In figure 6.23b, though, the trials have series of studies by Wulf, Shea, and colleagues
been reordered so that the trial with the shortest used a research strategy in which variables
overall MT (843 Iris) is at the top, and the trial with known to affect learning produced different
216 Motor Control and Learning

experimental variations of practice have the effect

of increasing the accuracy of relative timing (as
measured in retention and transfer tests) (Lai &
Shea, 1998; Lai, Shea, Wul( & Wright, 2000; Shea,
Lai, Wright, Immink, & Black, 2001; Wright &
Shea, 2001; Wulf & Schmidt, 1989; Wul( Lee, &
Schmidt, 1994; Wul( Schmidt, & Deubel, 1993).
At the same time, these studies also showed that
those same experimental variables degraded the
a scaling of absolute-duration performances (see
also Wulf & Schmidt, 1996), which is a measure
of the accuracy of parameterization. The overall
findings appear to show some inconsistencies
(cf. Wulf & Lee, 1993; Wulf & Schmidt, 1994b,
1997), however, and more work seems needed
to clarify these effects. But overall, there is ample
evidence that some factors have different effects
on the learning of relative timing versus the
parameterization of absolute timing, justifying
their theoretical separation (see Shea & Wul(
b
2005, for a review).
These results are examples of double dissocia-
tions as we discussed in chapter 5. Such double
dissociations provide very strong evidence that
two proposed theoretical structures are actually
different empirically. The interpretation is that
variables affecting the learning of program vari-
ables (such as relative timing, relative force, and
sequencing) are not the same variables as those
that affect the learning of parameter-assignment
c processes. Hence, these kinds of results strongly
support the theory that program development
FIGURE 6.23 Temporal structure in typing the word "enclosed." and parameter assignment are theoretically (and
(a) Letters are shown in the same order in which they were origi- actually) separate processes.
nally typed; (b) the same letters are ordered in terms of their overall
movement time (MT); (c) the letter durations are expressed as pro- Overall force Parameter
portions of overall MT.
Reprinted from Neuroscience, Vol. 5, C.A. Terzuolo and P. Viviani, "Determinants and
A second parameter proposed for implementing a
characteristics of motor patterns used for typing," pg. 1092, Copyright 1980, with permission GMP is an overall force parameter that modulates
of Elsevier.
the amounts of force produced by the participat-
ing muscles. The force parameter is involved with
effects depending on what was learned or mea- determining how forcefully the relevant muscles
sured (much more will be described about learn- will contract when they are recruited by the pro-
ing variables in chapters 11and12). This strategy gram. The evidence for the existence for such a
attempts to look for patterns of dissociations in parameter is weak, but logically a force parameter
learning (see also the same idea in the study of is included in the model (see also Schmidt, 2003).
the two visual systems, discussed in chapter 5), Pew (1974a) described, as an example, a post
such that a particular learning variable has dif- office in which a conveyer belt carried small pack-
ferent effects on the learning of relative timing ages to be sorted by an employee. The person
(a measure of GMP performance) as compared picked up the package and, with a" set shot" that
to its effects on overall duration (a measure of might be considered good form for a basketball
parameterization) (DeJaeger & Proteau, 2003; player, tossed the package into one of about 15
Shea & Wul( 2005). According to some studies, equidistant bins for later delivery to different
 Central Contributions to Motor Control im 217

postal zones. This package-sorting "system" increased (scaled) while the original temporal
required a number of processes on the part of pattern was maintained. Of course, increasing
the performer. First, because the bins were equal the force leads to increased distance that the pen
distances from the person, the final velocity (as travels; hence, the word is larger with the same
the package left the hand) of each package needed spatial-temporal pattern. Similar interpretations
to be approximately the same in order for each can be made from a study of handwriting by
package to reach its bin, regardless of its weight. Denier van der Gon and Thuring (1965), who
But a package with a larger mass will require showed that when the friction of the pen on
the application of more force at a given duration the writing surface was increased, a systematic
in order to achieve the desired terminal velocity. decrease in the writing size resulted but with no
Thus, the performer must choose a force param- change in the pattern of letters produced.
eter that can be applied to the generalized "set In the examples just cited, the overall force
shot" program. Presumably, the person would parameter applies to all of the participating
pick up the package, heft it to determine its mass, muscles proportionally, keeping the relative forces
and then select a force parameter for the general- applied to the limb proportional. This concept is
ized program that would achieve the proper goal. very much like the overall-duration parameter,
The program could be run when the force and which is applied to the sequence as a whole. A less
duration parameters had been selected. restrictive view is that the force parameter can be
Another example that supports the concept applied to various actions in the sequence without
of an overall force parameter comes from Hol- affecting other actions in the body. For example,
lerbach (1978). Figure 6.24 shows the acceleration walking with a heavy backpack would seem to
tracings from a subject writing the word "hell" require that more force be applied to the muscles
two times, one word being twice the size of the that operate against gravity in walking, but the
other. The accelerations are, of course, directly muscles that cause the foot to move through the
proportional to the forces that the muscles are air in the swing phase would not need to have
producing during the action. The tracings have extra force applied to them. Perhaps a force
the same temporal pattern, yet the accelerations parameter is selected that applies only to those
in the tracing for the larger word are uniformly aspects of the program that require extra force.
larger than those for the smaller word. It appears However, this idea has the disadvantage of requir-
that the forces applied to the pen were simply ing the motor system to do more "computing" in

Away

c:
0
~Q)
a;
u
u
<(

Toward

FIGURE 6.24 Vertical accelerations produced in writing the word "hell;' with one word having twice the amplitude of the other.
(The tracings show a remarkable degree of temporal agreement, with systematic differences in amplitude of acceleration.)
Reprinted, by permission, from J.M. Hollerbach, 1978, A study of human motor control through analysis and synthesis of handwriting. Unpublished doctoral dissertation
(Cambridge, MA: Massachusetts Institute of Technology).
218 Motor Control and Learning

order to move. Consequently, there appears to be Muscle-Selection Parameter

less overall support for the separation of absolute In the analysis of the handwriting examples
and relative forces than has been seen for the other shown in figure 6.18 (from Lashley, 1942), we
features of the GMP theory. argued that the muscles for the particular action
Interaction of Duration and Force could not be stored "in" the GMP, because the
Parameters same program produced movements in entirely
There is a further argument with respect to the different limbs. Thus, the sequential ordering
necessity for a force parameter, but it is less obvi- embedded in the motor program is considered
ous than the one just given. Consider a movement to be abstract, with the specific joints and muscles
in which you begin with your elbow straight, flex added during the preparation (and parameteriza-
the elbow to 90°, and then extend it to the straight tion) of the program. In this case, it is reason-
position again, completing all of the movements able to think of the specification of muscles
in an overall MT of 300 ms. The motor program (or joints) as another parameter of the motor
presumably determines the relative timing of program.
the biceps, the cessation of the biceps and the Additional evidence for this view comes from
initiation of triceps (for the reversal), and then numerous experiments using a bilateral-transfer
the contraction of the biceps again to "clamp" the paradigm. For example, Shapiro (1977) used a
limb in the final target position. Now consider wrist-twist task similar to that described earlier,
what would happen if you simply decreased having subjects practice this sequence with the
the duration parameter of the program without right hand for five days. Then she asked the
changing a force parameter. Selecting a shorter subjects to make the same movements with the
duration parameter would cause the program to left hand, which had never been used for this
move through the biceps-triceps-biceps sequence pattern before. She found a pattern of activity
more rapidly while keeping the forces produced shown in figure 6.25, in which the well-practiced
by these muscles constant. What would happen right-hand pattern is indicated by the open
to the movement? Because the impulses will be circles and the novel left-hand pattern is indi-
shorter in time, the impulses will be smaller, cated by the closed circles. The two patterns are
and the limb will not have moved as far in the nearly identical, and the case can be made that
time allowed for biceps activity; thus, the move- the program that was generated by practice with
ment will reverse itself short of the 90° position. the right hand could be produced with the left
Decreasing a duration parameter while holding hand. Further evidence for the preservation of
a force parameter constant would result in an sequence learning during transfer to different
inappropriate movement in terms of its extent. effectors has been shown by Keele, Jennings,
One possible remedy is to choose the duration Jones, Caulton, and Cohen (1995; see also Jordan,
parameter so that the overall MT is correct, and 1995; Keele et al., 1990).
then to choose an overall force parameter that will In another variant of the bilateral-transfer
be sufficient for the limb to move to 90° before paradigm, Park and Shea (2002) found that rela-
reversing itself (Schmidt, Zelaznik, Hawkins, tive force and time transferred from one limb to
Frank, & Quinn, 1979). If the force parameter is the other, but absolute force and time did not.
too large, the movement will go too far in the Therefore, this could be interpreted to suggest
proper amount of time; if the force parameter is that positive bilateral transfer of the invariant
too small, the movement will not go far enough. features occurred, not the movement parameters.
Thus, with this view, movement distance for Stated another way, since bilateral transfer of
a given program is determined by a complex the properties of the GMP was high this finding
combination of duration and force parameters. supports effector independence of the GMP, but
Clearly, duration and force parameters must since bilateral transfer of the parameters was
complement each other. The selections of the force low, this finding suggests effector specificity in
and duration parameters are not independent, bilateral transfer of movement parameters (see
as the particular value of the force parameter also Park & Shea, 2003, 2005; Wilde & Shea, 2006).
will depend heavily on the chosen duration These findings are reminiscent of Lashley' s (1942)
parameter. handwriting examples shown in figure 6.18: The
 Central Contributions to Motor Control 111 219

0.20

G)
.§
.... 0.15
0
c
0
t:
g_
e
c..
0.10

2 3 4 5 6 7
Segment number in sequence

FIGURE 6.25 Proportions of total movement time (MT) required to traverse various movement segments in a wrist-twist
task. (The pattern is similar for the practiced right hand and for the unpracticed left hand.)
Adapted from Shapiro 1977.

essential features of the handwriting were main- effector, in which the same pattern is produced
tained when limbs were switched, although the in two different sets of "muscles."
"smoothness" was affected dramatically.
Changing Parameters and Programs
Parameters in the Phonograph Record Additional evidence supporting the GMP comes
Analogy from experiments in which some aspect of the
Earlier we presented the analogy between a GMP movement has to be changed during the move-
and a phonograph record, with information about ment. For example, Quinn and Sherwood (1983)
order, relative timing, and relative force struc- had subjects make elbow flexion or extension
tured "in" the motor program ("on" the record) movements of 400 ms. Occasionally an audi-
to define a given pattern. To complete the analogy, tory signal, administered in different blocks of
add the ideas about parameters just discussed. trials, would instruct the subject to either (a)
The overall duration parameter is analogous move faster or (b) reverse the movement. The
to the speed of the turntable. When the record findings, similar to those from earlier studies in
turns more rapidly, the overall duration of the this same general paradigm (Gottsdanker, 1973;
record's activity decreases, but the relative timing Vince & Welford, 1967), showed that the latency
of sounds remains invariant. Next, the overall of the corrections (the interval from the auditory
force parameter can be thought of as the volume stimulus until the first EMG change) was 100 ms
control, whereby the same pattern of action can shorter when the movement had to be sped up
be produced either loudly or softly. This is very than when it had to be reversed. Theoretically,
much like writing in small or large letters with with a reversal, the subject has to stop running
the pattern of the writing remaining the same. a given program and select, parameterize, and
Muscle-selection parameters are analogous to the initiate a different one that will reverse the move-
operation of speakers. If you have an expensive ment. However, when the movement is sped up,
set of speakers in the living room and a cheaper the existing program can be retained, and only
set in the kitchen, choosing one of these will affect a reparameterization is necessary (e.g., with
the sound quality (and, of course, the location) of adjusted overall duration and force parameters);
the music you are hearing, but will not change the the stages involved in program selection and
record itself. Here, the speaker is analogous to an initiation can be bypassed.
220 Motor Control and Learning

Roth (1988) has shown that these principles can be tolerated before we reject a conclusion that
hold for sport skills studied in the laboratory. For the data reveal invariance?
example, the RT to change a tennis ground stroke One solution to this debate was provided by
to a lob (presumably requiring a different program Gentner (1987). He proposed two statistical meth-
and different parameters) was estimated to be ods for assessing relative invariance in a set of
about 600 ms, whereas the RTs to change the direc- data. One method, called the constant-proportion
tion or length of the ground stroke (presumably test, uses statistical regression to assess whether
requiring only new parameters) were estimated or not the relationship between MT and relative
to be about 200 ms less. Analogous results were timing has a slope that deviates from zero (see
provided for table tennis and volleyball skills, chapter 2 for discussion about regression meth-
suggesting that the difference between program ods). If relative timing is invariant, then the slope
plus parameter selection versus only parameter of the regression line will not deviate from zero,
selection is general across a variety of movement either positively or negatively. Gentner's (1987)
behaviors (see also Obhi, Matkovich, & Gilbert, analysis provided an objective, statistical solution
2009; Wright, Black, Park, & Shea, 2001). to the problem of assessing invariance. Using
these methods, Gentner reanalyzed some previ-
Concerns About Generalized ously published data sets and found that, while
Motor Programs some studies found support for invariant relative
timing, others did not. More recent experiments,
The GMP theory was first proposed more than 35 using the methods suggested by Gentner, have
years ago, and there have been numerous empiri- also produced evidence that is weighted against
cal and theoretical examinations of its predictions perfect statistical invariance (Burgess-Limerick,
since then. In general, the theory has held up Neal, & Abernethy, 1992; Maraj, Elliott, Lee, &
well. However, as with all theories, some data Pollock, 1993; Wann & Nimmo-Smith, 1990; but
and analyses do not provide support. see also Franks & Stanley, 1991).
The most contentious issue with regard to Are statistical tests appropriate to assess the
GMPs has been the concept of invariance, espe- invariance predictions of relative timing in GMP
cially as it relates to relative timing. We argued theory? The answer is unclear. Several questions
earlier that in order for a timed segment to be con- can be raised from a statistical point of view, such
sidered invariant, its duration, relative to the total as (a) the appropriateness of accepting the null
duration of the activity, must be constant over hypothesis when significant effects are not found
a series of separate executions of the program. (which would be evidence in support of invari-
But have another look at the relative timings for ance) and (b) the level at which to set the cutoff
each letter of the word "enclosed" (figure 6.23). point for rejection of the null hypothesis. Gentner
Although the relative durations for each letter suggested that a level of a= 0.05 is appropriate;
show rather consistent relative timings, there are however, a case could be made for more or less
still some deviations. The questions that arise are stringent levels.
these: Are these deviations meaningful? And how
do you decide what is a meaningful deviation? Central Versus Peripheral Invariance
Heuer (1988, 1991) has raised another important
Statistical Invariance issue. He suggested that, even in the absence of
A qualitative answer to these questions is to draw measured invariance, there might still be central
a straight line through the center of a set of data invariance. Heuer's argument uses the Wing
points (as was done with the absolute timing data and Kristofferson (1973a, 1973b) distinction
in figure 6.23b). If the data were perfectly invari- between central and peripheral timing as a basis
ant, then all the individual data points would fall (see chapter 7 for details). The idea is that the
exactly on vertical lines-the more data points timing observed at the output or peripheral level
that are observed off the line, the weaker is the is a combination of a central mechanism that
evidence for invariance. In reality, there is very periodically triggers an effector into action and
little chance that motor behavior will ever show the (somewhat variable) motor delays (such as
true, perfect invariance. Therefore, the question neural delays and muscle recruitment time) that
becomes, How much deviation from perfection occur following a central trigger. Heuer (1988)
 Central Contributions to Motor Control Iii 221

demonstrated that a variable motor delay could might a decentralized theory of motor control (i.e.,
result in an absence of invariance in observed a theory that does not involve a central motor
performance even though the relative timing of program) explain the findings of Wadman and col-
a central timing signal had no variance at all. leagues (1979; discussed earlier here), in which the
Thus, perhaps because of complexities in the complex coordination of agonist and antagonist
muscle properties in fast movements (e.g., Heuer muscle firings was preserved when the limb was
& Schmidt, 1988; Gielen, van den Oosten, & ter blocked in position and prevented from moving
Gunne, 1985; Zelaznik, Schmidt, & Gielen, 1986), at all? If the movement is shaped by the ongoing
it is possible that one might not detect invariance dynamics, then how could the "normal" pattern
at the level of the GMP by searching for invari- of muscle activations occur in the absence of any
ances in motor output. Perhaps this issue will movement at all? Similarly, how could the find-
be resolved only by future research analyzing ings of Valls-Sole and colleagues (1999), in which
the brain potentials of action prior to move- an acoustic-startle stimulus "released" a prepared
ment output. We will return to the discussion of movement almost 100 ms earlier than for a normal
invariant relative timing when we discuss how RT, be reconciled by a theory in which plans for
the system regulates the coordination of two or action do not have a prominent role?
more activities at the same time (in chapter 8). Regarding the divide that separates different
theoretical perspectives on how movements are
Doubts About Motor Programs
regulated, Steve Keele said: "To a large degree we
A broader question that has been asked, not just are simply talking about different kinds of phe-
specific to GMP theory, is whether or not the motor nomena, and those phenomena have generated
program concept remains a relevant or viable very different databases that are not addressed
theoretical construct at all (Morris, Summers, by both classes of theory. It is difficult to pit two
Matyas, & Iansek, 1994; Summers &Anson, 2009). 'theories' or perspectives against each other
A number of alternative motor control positions if they do not speak to the same phenomena"
have been described in the past 30+ years, and (Keele, 1998, p. 404). As we see it, the criticisms
these have received considerable theoretical and of the motor program concept lose strength in
empirical support. Some of these views suggest argument when a decentralized concept must
that the motor program view ascribes too much come to terms with how rapid, ballistic-type tasks
"intelligence" to a central command center-that are controlled. At the very least, for actions such
much of the evolution of a movement as it unfolds as these, we feel that the motor program concept
in the course of an action occurs as a natural con- provides the most complete and parsimonious
sequence of the dynamic interactions of the central account of the research evidence.
nervous system, the effectors, the environment,
and the task at hand. In other words, the argument
is that a complex plan of action that serves to con- Summary
trol movement need not be stored in memory and
recalled prior to initiating movement. The response-chaining hypothesis proposed
We share some of these concerns about motor by James (1890) was the first open-loop theory
programs, especially as they relate to the ongoing for motor control. It held that each action in a
regulation of coordination skills. In chapter 8 we sequence is triggered by the movement-produced
discuss some theoretical alternatives regarding feedback from the immediately preceding action.
how these skills might be controlled (including a Research on the role of feedback in movement
GMP account). However, we also note that many performance under various deafferentation con-
researchers who express doubts about motor ditions has tended to show that sensation from
programs usually study motor tasks that involve the moving limb is not essential for motor perfor-
continuous skills, often rhythmical or cyclical (or mance, although it contributes to the smooth con-
both) with relatively long durations, for which trol of many actions. Thus, the response-chaining
ongoing regulation, using feedback from the hypothesis cannot be universally correct.
environment and body, would seem critical to Motor control scientists have three reasons
successful performance. Of these critics of the for believing that movements are controlled by
motor program concept we ask how, for example, programs: (a) the slowness of the information-
222 Motor Control and Learning

processing stages, (b) the evidence for planning trates the difference between errors of
movements in advance, and (c) the findings that selection and errors of execution.
deafferented animals and humans can produce 2. Find a research article that uses relative
skilled actions without feedback. This is not to say timing as the main dependent measure.
that feedback is not used in movement. Feedback What theoretical issues are investigated
is used (a) before the movement as information with use of this measure in your article?
about initial position, or perhaps to tune the
spinal apparatus; (b) during the movement, when
it is either "monitored" for the presence of error Web Resources
or used directly in the modulation of movements
reflexively; and (c) after the movement to deter- From the National Transportation Safety Board,
mine the success of the response and contribute featuring unintended acceleration and other
to motor learning. examples of movement errors:
An early definition of motor programs as struc- www.ntsb.gov/events/boardmeeting.htm
tures that carry out movements in the absence of
Recent editorial about the role of human error in
feedback was found to be inadequate to account
Toyota automobiles:
for the evidence about feedback utilization during
movement. Also, problems were associated with www.nytimes.com/2010/03/11/opinion/
the requirement for storage of many different 1 lschmidt.html
motor programs (the storage problem), as well as Video of an actual case of unintended acceleration:
with the means by which the motor program
www.youtube.com/watch?v=ihDK8ny4ouc
could create a novel action (the novelty problem).
For these reasons, the motor program is thought
of as generalized-containing an abstract code Notes
about the order of events, the relative timing (or
temporal structure) of the events, and the relative 1 Four of Taub' s monkeys were reexamined 12 years after

force with which the events are to be produced. their surgery, and all revealed considerable functional reor-
These generalized motor programs (GMPs) ganization of the brain structures responsible for sensory
require parameters in order to specify how the representation (Pons et al., 1991). Thus, it seems that motor
and sensory systems may have both short- and long-term
movement is to be expressed in the environment.
methods for adapting to the loss of sensory feedback.
Such parameters are the overall duration of the
2 This view could also be related to the reflex-chaining
movement, the overall force of the contractions,
hypothesis. The difference is that the closed-loop model
and the muscle (or limb) that is used to make the would have the feedback evaluated against a reference
movements. With such a model, many different of correctness, whereas the reflex-chaining view would
movements can be made with the same program have the feedback from the movement trigger the next
(reducing the storage problem), and novel move- action directly.
ments can be produced through selection of 3 Note that Gray (2002, 2009a) did not manipulate the

parameters that have not been used previously vertical deviation of the pitch from the strike zone in
(reducing the novelty problem). these batting simulations (i.e., "inside" or "outside" of
the strike zone).
4 The generalizations that errors in execution can be corr-

Student Assignments ected (a) without interference from other similar correc-
tions and (b) with latencies unaffected by the number of
possible corrections have not been studied carefully and
1. Prepare to answer the following questions should be considered with caution.
during class discussion: 5 Lashley had a good reason for choosing these particu-

a. Describe the open-loop operation of lar words to be written; the term motor equivalence refers
any human-made device of daily living to the idea that different effectors can be used to achieve
(e.g., a microwave oven). the same goal.
6 Not long ago, a teenaged niece of R.A.S. asked him,
b. Use an example from sport that illus-
"What's a phonograph?" Although the question raises
trates the concept of point of no return.
some doubt about the longevity of this analogy, we note
c. Provide an example from food prepara- that phonograph records have made a resurgence in
tion activities in the kitchen that illus- popularity in recent years.
 CHAPTER 7

ne of the most common errors of move- Whereas a neat set of simple, elegant principles
ment control occurs when we try to can be stated for various branches of the physi-
perform a task faster than "normal." Attempts cal sciences, we should not expect the same for
to thread a needle rapidly, to type an e-mail in the behavioral sciences, or for motor control in
a hurry, or to pour a glass of milk quickly often particular. For a number of reasons, in motor
result in an increased number of aiming errors. control we find far fewer statements possessing
These are errors based often on a lack of precision, sufficient generality and supporting evidence to
and they are usually (but not always) studied in have attained the status of a "law." One reason
tasks involving arm and hand movements. is that the motor control principles have been
In this chapter we focus on the fundamental far more difficult to discover, based as they are
principles that pertain to various movement on data from biological systems that are more
variables and some theoretical ideas that have variable ("noisy") and complex than the physical
emerged from them. Such principles are critical systems. Often the relationships are not obvious
to any science, as they describe the relationships and must be "teased apart" from the background
among measures of the objects under study. As noise or variability in order to be observed. The
such, many of the basic laws of motor behavior situation is complicated further when differ-
may be seen as analogous to the fundamental ent principles are observed for different people
principles of physics. The laws relating the mass, (termed individual differences, chapter 9), which
velocity, and acceleration of objects when forces seems to suggest that different laws should exist
are applied to them (the principles of mechan- for different individuals or for different classifica-
ics), for example, have served as the cornerstone tions of people (children vs. adults, for example).
of the physical sciences and therefore deserve a Even so, these motor control principles hold well
special status. In the same way, the field of motor in many separate sets of data and generalize to a
behavior has analogous principles that are some- variety of practical situations, and thus represent
how fundamental to all the rest: principles that statements of fundamental knowledge about
describe, for example, the relationship between skilled performance.
the speed at which a limb moves and the resulting One of the most common occurrences in daily
accuracy, or the relationship between movement activities is known as the speed-accuracy trade-off
distance and movement time. Common sense tells us that, as we move more

223
224 Motor Control and Learning

rapidly, we become more inaccurate in terms of yet precise measurements to answer the research
the goal we are trying to achieve. For example, questions. The tasks used by Woodworth involved
trying to type too fast or pour a glass of milk too simple repetitive line-drawing movements to a
quickly generates annoying mistakes; the adage target. Movement speed was varied by changing
"haste makes waste" has been a long-standing the frequency of a pacing metronome. Studies
viewpoint about motor skills. As we will see, were done with various distances, with the right
however, the speed-accuracy trade-off exists in and left hands, and with the eyes closed and open,
different forms for different types of movement in an attempt to uncover some of the fundamental
tasks; the principles of speed-accuracy trade-offs relationships between speed and accuracy. Gener-
are specific to the goal and nature of the move- ally, Woodworth found that accuracy (measured
ment tasks. as absolute error) decreased as the movement
Three specific types of trade-offs are presented speed increased, that the left hand was less accu-
later in this chapter. These relate to situations in rate than the right hand (in right-handed subjects),
which spatial or temporal accuracy, or both, are and that the decrease in accuracy with increased
the primary demands of the movement. Theories speed was greater when the eyes were open than
of movement control that relate to these trade-offs when they were closed (see "R.S. Woodworth on
are presented later in the chapter. Manual Aiming," p. 143, and figure 5.5, p. 144).
Most of his results have not, in general terms at
least, been contradicted in more than a century of
Fitts' Law: The Logarithmic research conducted since Woodworth published
Speed-Accuracy Trade-Off his results in 1899. Fifty-five years later, the nature
of this speed-accuracy relationship was described
as a formal mathematical law by Paul Fitts (1954;
The first class of speed-accuracy trade-off func-
"Paul Fitts" on p. 11).
tions occurs in situations in which the goal is to
move a limb (or some other "effector") as quickly Fitts' Research
as possible to reach or touch a target, doing so
with a minimal amount of error. Such is the goal In 1954, Fitts published a systematic analysis of
in tasks like typing; moving a mouse-driven the relationship between speed and accuracy that
cursor to a desktop icon; moving the foot to the has become one of the landmark publications in
brake pedal in driving; and numerous other the history of motor behavior research (Kelso,
activities that require rapid movements to push, 1992). In the typical Fitts paradigm (or Fitts task),
touch, grasp, intercept, or displace an object. a subject is to tap a handheld stylus (a pencil-
shaped object) alternately between two target
Early Research plates as rapidly as possible for a predetermined
duration (e.g., 20 s). The two targets are usually
The first major attempt to study scientifically the rectangular and oriented as shown in figure 7.1.
relationship between the speed of a movement Both the width of the targets (W) and the ampli-
and its resultant accuracy probably came from tude of the movement between them (A) are
Woodworth (1899; see chapter 5 and "R.S. Wood- under the control of the experimenter and there-
worth on Manual Aiming" on p. 143). This work fore can be altered to produce a large number of
was far ahead of its time in terms of both the ideas combinations of A and W. The task is scored as
examined in experiments and the techniques used. the number of taps (regardless of whether they
Woodworth proposed that aiming movements are are correct) in 20 s, but subjects are cautioned to
made up of an initial-adjustment phase that propels limit errors (missed targets) to no more than about
the limb toward the target in an open-loop mode 5% of their movements. A more general view of
of control (chapter 6) and a current-control phase the experimental setup is shown in figure 7.2.
based on visual feedback that causes the limb to
"home in" on the target (chapter 5). Fitts' Law Defined
At the time Woodworth's experiments were Fitts found that the relationship between the
conducted, sophisticated electronic recording amplitude (A) of the movement, the target width
techniques were not yet available; so clever (W), and the resulting average movement time
methods were devised that provided practical (MT) was given by the following equation:
 Principles of Speed and Accuracy 111 225

FIGURE 7.1 The Fitts paradigm. (The performer taps a stylus alternately between two targets of width W separated by a
distance A)

MT = a + b[Log2 (2A/W)] (7.1) The Fitts equation has the general form of a
where MT is the average movement time for a linear equation (Y =a+ bX), where Y is average
series of taps, computed as the trial duration (e.g., MT, Xis represented by the term Log2 (2A/W),
20 s) divided by the number of taps completed and a and b are the empirical constants of a linear
in that time. For example, a 20 s trial duration equation. (See also the section on linear empiri-
divided by 50 taps in the trial yields 20 I 50 = 0.4 cal equations in chapter 2 and the discussion of
s/tap, or 400 ms/tap as the average MT. Hick's law in chapter 3.) Therefore, a graphical

Targets

FIGURE 7.2 A subject performing the Fitts tapping task.

Reprinted from Categories of Human Learning, A.W. Melton (Ed.}, P.M. Fitts, Perceptual-motor skills learning, categories of human learning, pg. 258. Copyright ©1964, with
permission from Elsevier.
226 Motor Control and Learning

plot of average MT against Log2 (2A/W) should three data points (representing movements with
be linear (a straight line). very short MTs). Notice, for example, that two
The data from one of Fitts' original experi- conditions for which the Log2 (2A/W) = 6.0 had
ments are presented in figure 7.3. The values virtually identical MTs (representing two tasks,
of A and W were varied experimentally by one with A= 16 in., W = 1/2 in., and another
changing the arrangement of the target board task with A= 8 in., W = 1/ 4 in.). There are simi-
(as in figures 7.1 and 7.2) for different blocks lar situations with the other data points plotted
of trials, and the resulting MTs were measured for a given value of Log2 (2A/W), such as 3, 4,
after subjects had received some practice at ands.
these tasks. Figure 7.3 illustrates the average
MTs as a function of Log2 (2A/W), where each Interpreting Fitts' Equation
of the latter values is computed by taking the What does it mean that the Log2 (2A/W) plots
values of A and W, dividing them, and looking linearly with the average MT in the Fitts task? A
up the value of the Log2 (2A/W) in the table of closer look at each of the components makes Fitts'
logarithms (see appendix). For example, one of equation easier to understand.
the data points in figure 7.3 is associated with a The Index of Difficulty First, notice that
value of 5 on the x-axis. The task values for this the value of Log 2 (2A/W) seems to determine
particular data point had a target amplitude how much time was required for each of these
(A) of 4 in. (10.2 cm) and a target width (W) of movements, so this value seems to be related
1/4 in. (0.64 cm). Thus, the value 2A/W = 2(4) in some way to how "difficult" the particular
I 0.25 = 32. Now, from consulting the table in the combination of A and W was for the subject.
appendix, the Log2 (32) = 5.0. (More simply, the For this reason, Fitts called this value the index
Log2 of a number is the power to which the base of difficulty (ID), which is expressed in terms of
2 must be raised in order to reach that number; bits, taken from information theory (see later
i.e., 2 5 = 32.) here, and chapter 3 for a fuller discussion). Thus,
The data in figure 7.3 indicate that, for the in Fitts' terms, the "difficulty" of a movement
various combinations of A and W used in Fitts' task was related jointly to the distance that the
experiment, the average MTs lie almost perfectly limb moved and to the narrowness of the target
on a straight line, except perhaps for the leftmost at which it was aimed. In fact, the relationship
is even more restrictive than this, as the "dif-
ficulty" of the movement is theoretically the
0.8 same for any combination of A and W that has
the same ratio. Doubling A and doubling W at
0.7 the same time result in a value of 2A/W that
is the same, and hence the same value of Log2
0.6 (2A/W) and the same predicted average MT.
~
c Another way to say this is that the average MT
Cl) 0.5
E
Cl)
is linearly related to the ID, where ID = Log 2
> (2A/W). Thus, Fitts' law could also be written as
0 0.4
E
~
MT = a + b(ID).
~ 0.3 The Empirical Constants (a and b) Next, the
E
j:: values a and bare the empirical constants-they
0.2
are required in order for the mathematical equa-
0.1 tion of the line to fit the observed data from
the experimental setting. The constant a is the
y-intercept, referring to the value of MT where
2 3 4 5 6 7
Log2 (2AIW)
the line of best fit crosses the MT axis. Specifi-
cally, the constant a is the value of MT when the
FIGURE 7.3 Average movement time (MT) as a func- ID is zero. But what does it mean to say that a
tion of the index of difficulty (ID) (Log 2 [2A/W]). movement has "zero difficulty"? This has been
Data from Fitts 1954. a serious problem for the understanding of the
 Principles of Speed and Accuracy 111 227

ideas surrounding the Fitts task (see Welford, any one of these. That is, the average MT for any
1968, for a more thorough discussion). For our given limb still plots linearly with the ID, but
purposes, a movement with "zero difficulty" is with different values of a and b.
one with a ratio of 2A/W of 1.0 (because Log2 The Log Term Next, why is a Log term used
[1] = 0). Therefore, the intercept refers to the in the equation, and why is the Log given to the
situation in which the amplitude is one-half base 2? When the idea was originally published,
the target width, which results in targets that the theoretical rationale and interpretation was
overlap, so that the subject's task of tapping based on a dominant theme of the day-the
"alternately" from one target to another actually information theory of communication (Shannon &
involves tapping up and down as quickly as Weaver, 1949; see "Uncertainty and Information
possible, with essentially no accuracy require- Theory" on p. 67). Recall the discussion of Hick's
ment. law of choice reaction time (RT) in chapter 3; the
The constant b is the slope, and it is far more equation for that relationship also had a Log 2
straightforward in its interpretation. Here, the term. The Log2 (N), in which N was the number
slope refers to the added MT caused by increas- of equally likely stimulus-response alternatives,
ing the ID by one unit, which is one bit (binary was a measure of the amount of information (in
digit; refer to chapter 3). In this sense, the slope bits) required to resolve the uncertainty about
refers to the sensitivity of the effector to changes N alternatives. The Log2 term in Fitts' law can
in the ID. An example of this comes from Lan- be seen in a similar way: 2A/W is related to the
golf, Chaffin, and Foulke (1976), where the number of possible movements, and the Log2
results from a number of different movement (2A/W) is the information required (in bits) to
situations using the Fitts task were compared. resolve the uncertainty among them.
Langolf et al. found that the slope increased
progressively as the limb used was changed The Speed-Accuracy Trade-Off
from the finger to the wrist to the arm. The data
Fitts' law implies an inverse relationship between
from Langolf and colleagues suggest that the
the "difficulty" of a movement and the speed
larger and more cumbersome limbs (the arms)
(i.e., the time) with which it can be performed.
are more sensitive to the changes in the ID than
Increasing the ID decreases the speed (i.e.,
are the fingers, which can be controlled more increases the MT). One way to think about this
precisely. is that the individual in some way "trades off"
Differences in the slopes of the Fitts equation speed against accuracy, and this trade-off is
have also been shown to be sensitive to many done so that the rate of information processing
other factors. For example, higher slopes in is held constant. In addition to this strict view in
the Fitts equation are typically found in older terms of the constancy of information processing,
adults, suggesting that the impact on average people presumably have some control over their
MT of higher IDs is greater for older adults than strategy in moving. They can move very quickly
for younger adults (Goggin & Meeuwsen, 1992; at the "expense" of being less accurate, or they
Pohl, Winstein, & Fisher, 1996; Walker, Philbin, & can move very accurately at the "expense" of
Fisk, 1997; Welford, Norris, & Shock, 1969). The being slower. In this way, Fitts' law has been fun-
slope of the Fitts equation also can be reduced damental in describing one particular aspect of
considerably with practice: B. Kelso (1984) the speed-accuracy trade-off, or the performer's
found that the slope was reduced to nearly zero capability to change the control processes so that
following 40,000 trials conducted over 20 days speed and accuracy are kept in some balance. In
of practice. Thus, the slope of the Fitts equation the case of Fitts' law, the subjects were told to
represents an interaction of ID with the" control- "emphasize accuracy rather than speed," and
lability" inherent in the system (e.g., the effector error rates averaged less than 5%. Thus, Fitts'
characteristics, age of the subject, skill level). law describes how MT must be traded off, under
However, it is important to keep in mind that different values of ID, in order to maintain accu-
even though the slopes and the MT values may racy. In a later section, we will see how errors
be different under various conditions or with are traded off in order to determine the speed of
different individuals, Fitts' law still holds for limb movement.
228 Motor Control and Learning

Subsequent Research on Fitts' law 90% of the variance in many data sets. Thus, the
improvement in accuracy gained by changes to
It may not be obvious why so much attention Fitts' equation is relatively small. Fitts' original
has been paid to a single principle of motor per- equation for ID has been very useful in explain-
formance. There appear to be several reasons. ing the performance of both one-dimensional
First, as we have seen so often, human motor tasks (figures 7.1 and 7.2), and two-dimensional
behavior is complex and challenging to under- tasks (such as the angular moves of a cursor on a
stand. Because of this, it is very difficult to computer screen; Murata, 1999). For three-dimen-
provide precise mathematical descriptions of sional movements, however, Murata and Iwase
behavior that have general applications. Yet Fitts' (2001) found that Fitts' original definition of ID
law provides just that, and Fitts created it when explained only 56% of the variance in the data.
almost no precise mathematical work was being They found a much better fit of the experimental
done in motor behavior. Second, the law appears data with an additional term in the definition of
to relate to many different situations and to a ID, compared to Fitts' equation. The inclusion
number of variations of the original Fitts task. of an additional term that considered the angle
Thus it appears to represent some fundamen- of deviation (from vertical) in the z-dimension
tal relationship that governs many examples toward the target [sin(0)] improved the amount
of motor behavior. Third, since the publication of explained variance to 73%.
of Fitts' law, no investigations have shown it
to be fundamentally incorrect, although there The Discrete Fitts Task
now exist alternative ways of dealing with the Can the fundamental principle of Fitts' law be
relation between speed and accuracy. (One way applied to other, more "natural" movement
is to change the temporal and spatial accuracy situations? Fitts and Peterson (1964) showed that
demands of the task, which is described later the principle could be applied to a single-aiming
in this chapter.) In the next few sections we task in which a stylus is aimed at a target in a
document some diverse research directions and single, discrete move. The subject's task was to
applications that have been inspired by Fitts' law make a single move as quickly and accurately as
during the past 50+ years. possible to a target of fixed amplitude (A) from
Modifications to the Fitts Equation a starting position and with a given target width
(W). Consequently, Fitts and Peterson (1964) sug-
Modifications to the basic Fitts equation have gested that people trade off speed for accuracy in
been made in an attempt to achieve a better discrete tasks in much the same way they do for
fit-which refers to the amount of variance in continuous, cyclical movements. However, recent
the experimental data that can be explained by research suggests that the issue is much more
the equation. Modifications to the definition of complicated than this (Guiard, 1993, 1997; Schaal,
the ID by Welford (1968; Welford et al., 1969) and Sternad, Osu, & Kawato, 2004; Smits-Engelsman,
MacKenzie (1989) provided slightly better fit, as Van Galen, & Duysens, 2002). Other research since
too did a consideration of the width of the effec- the time of Fitts' original work has extended his
tor (e.g., a dowel inserted into a target hole) that ideas in a number of different ways.
was being moved into the target area (Hoffmann
& Sheikh, 1991). In contrast to the logarithmic Generality of Fitts' Law
relationship between A and W proposed by Since the publication of Fitts' law, investigators
Fitts, an exponential relationship described in have studied it in a variety of contexts, revealing
terms of a power law was suggested by Kvalseth that the principle shows remarkable generality
(1980), where and very few exceptions (see "An Exception to
MT= a(A/W)b (7.2)
Fitts' Law?" on p. 230). For example, in addition
to describing the aiming-accuracy principle for
Numerous other modifications to the Fitts young adults, the Fitts relationship holds well
equation were reviewed by Plamondon and Alimi for children (Hay, 1981; Schellekens, Kalverboer,
(1997). And, while many of these have served to & Scholten, 1984) and for older adults (Goggin &
improve the overall fit, it should be kept in mind Meeuwsen, 1992; Pohl et al., 1996; Walker et al.,
that the Fitts equation often explains well over 1997; Welford et al., 1969). Although Fitts' law
 Principles of Speed and Accuracy ff!! 229

was initially based on movements of the upper ticular configuration of alternate tapping is not
limbs, the principle has been found to hold when very representative of many real-life tasks (e.g.,
different effectors are compared, such as the foot, Schmidt, Zelaznik, & Frank, 1978). Despite its
arm, hand, and fingers (Drury & Woolley, 1995; peculiarities, however, the Fitts task (or slight
Langolf et al., 1976); when movements are con- variations of it) have long been found useful as
ducted underwater (R. Kerr, 1973, 1978) and in diagnostic measures in therapeutic situations.
space flight (Fowler, Meehan, & Singhal, 2008); For example, a common diagnostic tool used by
and when the movements required are so small neurologists in clinical examinations is called the
that they must be viewed under magnification "finger-nose test" (Desrosiers, Hebert, Bravo,
(Langolf et al., 1976). Fitts' law even applies and Dutil, 1995). In this test the patient is asked
when tasks are imagined and the movement is to touch his nose, then extend the arm to touch
not actually produced (Cerritelli, Maruff, Wilson, a small target such as the tip of the neurologist's
& Currie, 2000; Decety & Jeannerod, 1996), and finger, touch the nose, and so on, repeating these
when the difficulty of a Fitts task is only perceived alternations as rapidly as possible in a given
(Augustyn & Rosenbaum, 2005; Grosjean, Shif- time frame (e.g., 20 s ). The number of successful
frar, & Knoblich, 2007). cycles is counted, and the result is compared to
Fitts' law also applies in the context of every- established age-related normative data (Desro-
day activities. In Fitts' (1954) original work, two siers et al., 1995). Sometimes this task is used
other experiments were reported in which sub- to compare performances of the two upper
jects either placed discs over pegs or inserted pegs limbs within an individual who might have
into holes, where A was the distance between neurological damage to one side. This "Fitts-
the targets and W was the amount of tolerance like" task has proven to be a useful, reliable,
between the size of the hole and the diameter and valid diagnostic tool (Gagnon, Mathieu,
of the peg. With ID defined in this way, the Fitts & Desrosiers, 2004; see "Neurological Disor-
equation predicted MT very well. Movement- ders and Problems in Movement Control" on
time effects also follow Fitts' law when one p. 320).
compares tasks in which subjects point at, reach
for, and grasp objects of different sizes (Bootsma, Importance of Fitts' Law
Marteniuk, MacKenzie, & Zaal, 1994; Marteniuk, Lastly, one of the most powerful motivations
MacKenzie, Jeannerod, Athenes, & Dugas, 1987). for scientists working toward an explanation
As well, Fitts' law describes well the position- is the existence of a well-established principle
ing movements of various computer-input or law-one that has survived the test of time
devices such as keys (Drury & Hoffmann, 1992; and has been shown to be applicable to a wide
Hoffmann, Tsang, & Mu, 1995), joysticks (Card, variety of situations or people. Fitts' law cer-
English, & Burr, 1978), computer "mice" (Card et tainly meets these criteria. Thus, one natural
al., 1978; MacKenzie & Buxton, 1994; Trankle & outgrowth of this work was an attempt to
Deutschmann, 1991), and head-pointing devices understand the movement control ptbcesses that
(Andres & Hartung, 1989; Jagacinski & Monk, produced the particular relations described by
1985). Other ergonomics-related situations also Fitts' law. That is, investigators began theorizing
appear to have benefited from the study of Fitts' about why Fitts' law occurs. We have already
law, such as the length of screwdrivers and mentioned viewpoints based on information
crowbars (Baird, Hoffmann, & Drury, 2002) and theory, whereby the ID was taken by Fitts (1954)
alterations in the sizes and distances of monitor to be a measure of the amount of information
icons during computer interactions (Balakrish- needed to resolve the uncertainty about the
nan, 2004). The review by Plamondon and Alimi movement. However, dissatisfaction with this
(1997) documents many 'other demonstrations theoretical perspective led researchers to pro-
and applications of Fitts' law. pose alternative ways of explaining the kinds of
speed-accuracy trade-offs that Fitts and others
Applications of the Fitts Task had observed in their data. These theoretical per-
The original Fitts task is a rather strange move- spectives are dealt with in later sections of this
ment situation, and some have felt that the par- chapter.
 An Exception to Fitts' Law?
In the many hundreds of studies that have examined Fitts' law, only a very few exceptions have been
found to the general principle that.increases in the ID result in longer MTs. One of those exceptions
has been found recently. Adam, Mol, Pratt, and Fischer (2006) had subjects make discrete, aimed
movements to one of seven target locations on a computer touch screen. Because the targets
always remained the same size, changes in ID were produced by changes in the distance from
the home position to the target. For each trial, the specific target to which the subject moved was a
green square that appeared on the screen after the subject touched the home position. In one set
of conditions, the green target square appeared alone (figure 7.4a, line 2). In another condition, the
seven possible target locations used were outlined in black and remained on screen at all times.
The green target used for any given trial was achieved by coloring one of the target outlines when
the subject touched the home position, leaving the other nontargets as unfilled outlines as shown
in figure 7.4a (line 1).
T_he results of the study are illustrated in figure 7.4b. For trials in which the targets were presented
alone (without the seven boxes), a linear increase in MT accompanied increases in ID, as predicted
by Fitts' law. This was true also for the condition with the nontarget boxes; but, peculiarly, only for
targets 1 through 6. An exception to Fitts' law was found for the most distant (seventh) target, as
the MT for this target was shorter than that for the previous (sixth) target (which had a smaller ID).
In a follow-up study, Pratt, Adam, and Fischer (2007) used outlined target contexts of three boxes,
rather than seven, but adjusted the target by coloring the leftmost outlined box in one condition, the
middle box in another condition, and the rightmost box in a third condition. The results replicated the

Home Target

Outlined
targets shown

• DDDDDDD
a
Targets
presented

540
alone
• D
I 500
GI

:!.... 460
c
GI
E 420
~
0
:;E 380

4.4 4.8 5.2

b Index of difficulty

FIGURE 7.4 (a) An apparent exception to Fitts' law occurred when subjects moved
from the home position to the target located in the farthest right position in the lower panel
(line 1, containing the outlined target boxes), compared to movements made when targets
were presented alone (line 2). (b) The exception is seen as a reduction in movement time
to the seventh target position compared to the target with a smaller ID to its left (filled
circles in the figure).
Adapted from Adam et al. 2006.

230
 Principles of Speed and Accuracy m 231

findings from Adam ana colleqgu~s~~~~~~

target IDs used, the rightmostotthe(')uttineci
target located to the left . ·•··• · •. •. . ·. •·• .••.·. • •.·•· •·•·•·········· > ••·
So-called exceptions toaweW~st~bti
question about (a) whether Fitts' l~w
conditions somehow fall outside}if th~.'dortiain
the nearly ubiquitous findingth~t MT al
in which a target is presented had.q!Jl()Ql
boundary that somehow sped the t:Jelivery of.
findings here, such as the factthatthe ertors ...
on the sixth target, which occurred in botti·~r~s¢;
but this secondary finding might give b'ints ~bo~i·~ri .
also have important consequencesfor· .·.. · ·.~tk~~ot
designs in which the various.11onfar9el. 6os'~r&onJ .•.
case, these results should provide much fq~per. f6r.qi~§:l!J5Siofl•etrii:f fl:irt~
aiming and Fitts' law, as such "exceptions"usuafo/do.<. ·

The Linear Speed-Accuracy too fast or slow. Only those movements that
conformed to the goal MT (i.e., movements that
Trade-Off were within ±10% of the MT goal) were used for
analysis.
As noted in the previous section, the Fitts para- One experiment used a factorial combination
digm involves an unusual movement situation of three movement amplitudes (10, 20, or 30 cm)
that is not typical of many everyday tasks and and three goal MTs (140, 170, and 200 ms), result-
that produces one specific type of speed-accuracy ing in nine different combinations of A and MT,
trade-off. In an alternate approach, the Fitts para- performed by subjects in separate sessions. Errors
digm was changed to examine the speed-accuracy were measured as the within-subject standard
effects in tasks that require a single, aimed move- deviation (SD) of the movement amplitudes,
ment-requiring mainly preprogrammed actions. which defined the "spread" or inconsistency of
movements aimed at the target (see chapter 2). In
The Single-Aiming, Constrained keeping with the Fitts tradition, these errors are
Movement-Time Paradigm termed effective target width (W), which defines
the effective size of the "target the subject is
The revised paradigm used rapid single-aiming using" when moving with a particular MT and
movements of a stylus from a starting position A. Notice also that this paradigm is different from
to a target 10 to 60 cm away (Schmidt et al., the Fitts paradigm in that We is the dependent
1978; Schmidt, Zelaznik, Hawkins, Frank, & variable, with A and MT being the independent
Quinn, 1979). But rather than moving as quickly variables; in the Fitts paradigm, MT is the depen-
as possible, as in the Fitts and Peterson (1964) dent variable, and A and W are the independent
experiment, the subjects were asked to complete variables.
the movement in a particular goal movement time Figure 7.5 shows a plot of We for the various
(MT) that was specified by the experimenter. As combinations of A and MT. There was a clear
well, the movements were aimed at a thin target increase in variability as A increased; the effect
line that did not change in width (W). Thus, both was almost linear for the various MTs studied.
timing accuracy and amplitude (A) accuracy were Also, We increased systematically for any given
required of subjects. Performance of the goal MT movement distance as the MT decreased. Both
was achieved with augmented feedback after increasing A with MT fixed, and decreasing MT
each trial about whether the movement was with A fixed, imply increases in average velocity
232 Motor Control and Learning

6
'E
s 5
~
4

10 15 20 25 30
Movement amplitude (cm)

FIGURE 7.5 Effective target width (WJ in a rapid single-aiming task as a function of the movement time (MT) and move-
ment distance.
Reprinted, by permission, from R.A. Schmidt et al., 1979, "Motor-output variability: A theory for the accuracy of rapid motor acts;' Psychological Review 86: 425. Copyright©
1979 by the American Psychological Association.

(in cm/ s ). Thus, we can see these data as another is stable enough for some to name it a law of rapid
example of the principle that increasing move- actions (e.g., Flach, Guisinger, & Robison, 1996;
ment velocity decreases spatial accuracy (as in Jagacinski & Flach, 2003; Keele, 1986; Sanders &
Fitts' law), but for very rapid movements in the McCormick, 1993). 1
case of these movements.
This speed-accuracy trade-off can be seen
somewhat more easily in figure 7.6, where we
in figure 7.5 has been plotted as a function of 9
the average velocity, defined as A/MT. Now the 8
relationship between We and A /MT is nearly linear
across a range of movement velocities, and most of 7
the individual data points fall reasonably close to 6
a line of best fit. Thus, the data in figure 7.6 can be e
expressed in terms of a linear equation in which s 5

W, = a + b (A I MT) (7.3) 4

For the data in figure 7.6, the value of a (the 3

intercept) was 2.12 mm, and the value of b (the
2
slope) was 0.033 mm for one-unit increases in
cm/ s. Notice that for very different amplitudes
and movement times, but with the ratio of A
and MT about the same, the We was also about 50 100 150 200
the same. Movement velocity (cm/s)
These effects have been produced in a number
of separate experiments using various tasks- FIGURE 7.6 Effective target width (W0 ) as a function of
even using eye movements (Abrams, Meyer, & the average velocity (A I MT).
Reprinted, by permission, from R.A. Schmidt et al., 1979, "Motor-output variability:
Kornblum, 1989; Patla, Frank, Allard, & Thomas, A theory for the accuracy of rapid motor acts," Psychological Review 86: 425.
1985). This linear speed-accuracy trade-off effect Copyright© 1979 by the American Psychological Association.
 Principles of Speed and Accuracy im 233

Relationship to Fitts' Law trolled MTs (i.e., MT goals that are longer than
the subjects' minimum MT), whereas in the Fitts
The variables in the single-aiming paradigm are paradigm the MT goal is to be as fast as pos-
essentially the same as those in the Fitts paradigm sible while maintaining a high accuracy rate.
(but used slightly differently, as we have seen); Some have suggested that requiring the subject
and yet the speed-accuracy trade-off is consider- to achieve a particular MT goa.l, per se, causes
ably different-logarithmic in the Fitts paradigm, the shift to a linear speed-accuracy trade-off
and linear in the single-aiming paradigm. What (Meyer, Smith, & Wright, 1982; Wright & Meyer,
are the crucial conditions responsible for pro- 1983). Some evidence favoring this view has
ducing these two different trade-off functions? been reported by Zelaznik, Mone, McCabe, and
Two hypotheses have been suggested, which are Thaman (1988), who found that a relaxation of
related to the key differences between the two the precision demands in matching the MT goal
paradigms. diminished the strength of the linear speed-accu-
Feedback Hypothesis racy relationship.
Remember that the linear speed-accuracy trade- What effects might the MT goal have on the
off has been found using single-aiming move- motor control processes in these two paradigms?
ment tasks in which the total MT is very brief- Although the control mechanisms will be the
probably less than the amount of time required focus of discussion later in this chapter, it is
to detect that an error has been made and to important to note at this point that the single-
issue a correction (e.g., 200 ms or less in the data aiming, MT goal paradigm encourages subjects
from Schmidt et al., 1979, in figures 7.5 and 7.6). to adopt a noncorrected, single-impulse control
One obvious hypothesis is that the logarithmic strategy, whereas the ballistic (minimized MT)
trade-off occurs for movements that are at least goal requirements of the Fitts task often result
partially governed by feedback-based corrections in one or more corrective actions (Jagacinski &
(e.g., the Fitts task), whereas the linear trade-off Flach, 2003; Meyer, Abrams, Kornblum, Wright,
occurs for tasks that are entirely preprogrammed. & Smith, 1988; Meyer, Smith, Kornblum, Abrams,
In this view, the two trade-off functions are not & Wright, 1990).
in "competition," but rather are functions that Impressive support for this hypothesis was
describe different emphases on movement con- provided in a clever experiment by Carlton (1994)
trol-open-loop versus closed-loop emphases. in which corrective submovements were exam-
This does not mean, however, that Fitts' law ined in two movement tasks. Subjects produced
holds for movements with long MTs and that a 400 ms goal MT in one task (a timed task), and
the linear trade-off holds for movements with moved as rapidly as possible in the other task (a
short MTs. Whereas MT is certainly a factor in ballistic task). For each subject, the dispersion of
determining whether a movement will or will 95% of the aimed movements in the timed task
not be preprogrammed, the hypothesis is that was used to manufacture the target plate for the
MT provides a lower limit for feedback control; ballistic task; that is, a within-subject measure of
that is, movements with very short MTs cannot variability in one task (W) was used to determine
be under closed-loop control. However, move- the size (W) of that subject's target in the other
ments with long MTs could be under open-loop task. With the spatial accuracy demands of each
control provided that the environment is stable, task now closely equated, Carlton found that
the task is well learned, errors are not too "costly," the ballistic task condition resulted in corrective
and so on. For example, Zelaznik, Shapiro, and submovements on 93% of the trials, whereas
McColsky (1981) found that 500 ms movements corrections occurred on less than 20% of trials in
showed a linear trade-off. the timed task. This difference in the frequency
of corrective submovements between the two
Movement-Time Goal Hypothesis tasks is even more impressive when one consid-
Another obvious difference between the para- ers that, in the timing task, MT was about 90 ms
digms that produce the logarithmic and the longer than in the ballistic task: If MT were the
linear speed-accuracy trade-offs is the intended sole determinant of corrective submovements
goal MT. The single-aiming paradigm uses con- here, then presumably there should have been
234 111 Motor Control and Learning

more time available for a correction to occur in

the timed task.
Thus, it appears that the feedback and the
MT goal hypotheses converge on a similar issue
regarding the difference between the linear and
logarithmic spe~d-accuracy trade-offs. A linear
trade-off appears to occur in movement tasks
that encourage a preprogrammed, open-loop
control process; a logarithmic trade-off occurs in
the performance of tasks that encourage closed-
loop, corrective processes. The nature of these
control strategies is described in more detail later
in this chapter.
Slide

The Temporal
FIGURE 7.7 A coincident-timing task (top view). (A
Speed-Accuracy Trade-Off target on a belt moves directly toward the subject, who
attempts to move the pointer so that the two "coincide" at
the coincidence point.)
Certainly the view is widespread that when we Reprinted, by permission, from R.A. Schmidt, 1967, Motor factors in coincident
do things faster, we do them less precisely; and, timing. Unpublished doctoral dissertation (Urbana, IL: University of Illinois).

as we have seen in the previous sections, there is

considerable evidence to support it. However, the In table 7.1, the absolute errors in timing are
evidence that we have considered so far describes given as a function of four movement distances
the trade-off that occurs in reference to spatial (15, 30, 45, and 60 cm) and two movement-speed
accuracy. What happens when the focus is on tem- instructions (to move either at a maximal speed
poral accuracy? We deal with these situations next. or at a moderate speed). Notice that the absolute
error in timing (for any movement distance) was
Discrete Anticipation-Timing Tasks
uniformly smaller (20% on the average) for the
In tasks requiring anticipation and timing, such "maximal" instruction than for the "moderate"
as hitting a baseball, the individual must monitor instruction. Thus, when the person was perform-
the environmental situation (the flight of the ball) ing the task more "violently," with a smaller
and decide when to swing so that the bat arrives MT and larger movement velocity, the timing
at the plate at nearly the same time as the ball. In accuracy in these conditions actually improved.
chapter 3 we mentioned that these tasks require Such findings seem to contradict the notion of the
both receptor anticipation (of the ball flight) and speed-accuracy trade-off as discussed previously
effector anticipation (of one's internal movement in this chapter.
processes). Errors in timing result if the bat arrives At least two explanations for these effects
earlier or later than the ball. What is the effect on are possible. First, when the person moved the
errors in timing of increasing the speed (decreas- handle more rapidly, the movement was initiated
ing the MT) of the limb or bat? when the target was closer to the coincidence
Early research on this topic (Schmidt, 1967, point than was the case when the movement was
1969b) required subjects to move a slide along a slower. That is, the subject waited longer before
trackway so that an attached pointer would "hit" initiating the rapid movement. This extra time
a target moving rapidly at right angles to the to view the stimulus may have provided a more
trackway (see figure 7.7), with a follow-through accurate receptor anticipation (the estimate of
permitted. Accuracy in this task was measured as when the target would arrive at the coincidence
errors in time-either early or late with respect to point), which would permit a more precise
the target arrival. Subjects were asked to make a estimate of when the person should initiate the
movement that was of "maximal" or "moderate" movement.
speed, and four movement distances (15, 30, 45, A second explanation is that the rapid move-
and 60 cm) were used. ments themselves were more consistent than the
 Principles of Speed and Accuracy 235

Note. AE = absolute error; MT VE =variable error of MT; all measured in milliseconds.

Adapted from Schmidt 1967, 1969c.

slower moves. The variable error (the within- started a timer, and passing a switch along the
subject SD as discussed in chapter 2) of the trackway stopped the timer. The subject's pri-
MTs (see table 7.1) was about 44% smaller for mary goal was to produce a specific goal MT.
the rapid movements than for the slower ones. Newell and colleagues used a number of MTs
Thus, a second feature of fast movements is (ranging from 100 to 1,000 ms) combined with
more temporal stability from trial to trial than for a number of movement distances (ranging from
slower movements. This can be seen as an effector 0.75 to 15 cm), as can be seen in table 7.2.
anticipation advantage: The person can predict The primary dependent measure of interest
with greater accuracy when the limb will arrive was the variable error in timing (VE 1), defined
at the target if the movement is rapid because the as the within-subject SD of the MTs about the
trial-to-trial variability in the movement's travel subject's own mean, which represents a mea-
time is smaller. sure of an individual's inconsistency in timing.
As shown in table 7.2, for the 5 cm movements,
Discrete Movement-Timing Tasks the inconsistency in movement timing increased
markedly as the MT increased from 100 to 1,000
Newell, Hoshizaki, Carlton, and Halbert (1979) ms (VE 1 increased from 10.8 to 125.7 ms). The
performed a number of experiments on the effect was similar for the 15 cm movements,
temporal consistency of movement, and they with VE 1 increasing from 9.0 to 91.2 ms. Thus, it
provided perhaps the best documentation of the appears that this effect is like that seen in antici-
effects of MT on discrete timing. They used a pation timing (Schmidt, 1967, 1969b ); a shorter
ballistic-timing task in which the subject moved MT, given the same distance, produces increased
a slide along a trackway. The initial movement movement-timing consistency.

From Newell et al. 1979.

236 Motor Control and Learning

An even more interesting feature of these racy in timing, whereas earlier we showed that
findings is that the timing error is proportional to increasing the speed resulted in diminished spatial
the MT. In the fourth line of table 7.2 are the VE 1 accuracy. The finding that variability in timing is
values divided by the MTs, multiplied by 100 to proportional to the duration of the interval to be
convert them to percentage values. If the VE 1 is timed is consistent with a long-held finding in the
proportional to MT, then the VE/MT% values psychology literature called Weber's law.
will be similar for any MT from 100 to 1,000 ms.
Indeed, this is essentially what happened-the Temporal Consistency
VE/MT% values ranged between 8.6% and 14.9% and Movement Velocity
for all of the combinations of MTs and distances.
This effect of MT can perhaps be thought of The MT is not the only factor that strongly affects
as one that follows from the generally held view the VE 1-the movement's average velocity (i.e.,
that short intervals of time are" easier" to estimate the movement distance divided by the MT,
or produce than are long intervals of time. To usually in cm/s) has a strong influence as well.
illustrate, do this simple experiment. First, take Refer again to table 7.2 (Newell et al., 1979) and
a stopwatch and, without looking, estimate 1 s consider the VE 1 for the various values of move-
(1,000 ms) by pressing the button and releasing it ment velocity (second line in the table). Notice
at the appropriate time. Record the actual time for that for a given MT (e.g., 100 ms), the movement
each of 10 trials. Now, do the same task again but with the smaller movement distance, and hence
use a target interval of 2 s (2,000 ms). You should the lower movement velocity, has a slightly
find that the shorter interval is much "easier" to higher VE1 (10.8 ms for the 5 cm movement, 9.0
produce accurately; you should be much closer, ms for the 15 cm movement). This effect is even
on the average, to the target interval with the stronger for the 500 ms and 1,000 ms movements
1,000 ms task than with the 2,000 ms task. And, in the same table; the movement with the higher
if you had calculated VEs for your performances, velocity had a smaller timing error, even when
your VE for the 2,000 ms task should be roughly the MT was held constant.
twice that for the 1,000 ms task. The processes To see this effect expressed another way,
responsible for determining the duration of the examine the VEJMT% values presented on the
intervals are variable, and they seem to be vari- bottom line of table 7.2. Because the division by
able in direct proportion to the amount of time MT theoretically "cancels out" the effects of MT
that is to be produced. Because the movements in on the timing error (since these two variables are
the experiments of Newell and colleagues (1979) nearly proportional), any changes in VE/MT%
were, in effect, based on processes that take time, as a function of movement velocity must be due
it is reasonable that they should have been vari- to something other than MT. Here, the movements
able in time in nearly direct proportion to the with the longer movement distance (and hence
amount of time that they occupied. the greater movement velocity) have smaller VE1
There is considerable evidence that, gener- /MT% values. For the three MTs, these values
ally speaking, the VE1 (or inconsistency) in the were about 9% for the 15 cm movements and from
production of some interval of time tends to be a 10% to 14% for the 5 cm movements. Increasing
nearly constant proportion of the amount of time movement velocity made the movements more
to be produced, at least within broad limits. For consistent in terms of time.
example, Michon (1967) found essentially this Newell, Carlton, Carlton, and Halbert (1980)
effect with rhythmic tapping at different rates; studied these velocity effects in experiments that
Gottsdanker (1970) found the effect for RT (with were more thorough than those reported in the
subjects who had long RTs having greater within- earlier paper (see Newell, 1980, for a review). Vari-
subject VE 1 of their own RTs); and Schmidt and ous movement distances (ranging from 1.5 to 67.5
colleagues (1979) found these effects for aiming cm) and MTs (ranging from 100 to 600 ms) were
tasks in which the MTs were controlled. This well- used in combination, producing a set of veloc-
documented finding is an apparent contradiction ity values that ranged from 5 cm/ s to 225 cm/ s.
to the speed-accuracy principles described previ- The timing inconsistency (VE 1) was studied as a
ously in this chapter. Here, increased speed (by function of these variations in velocity; we have
decreasing the MT) produces increases in accu- converted these data to the VE/MT% measure
 Principles of Speed and Accuracy w 237

here so that they can be compared to the findings a specific goal MT. A similar finding also exists
in the previous section (see table 7.2). The 1980 for tasks in which the subjects make a repetitive
data of Newell and colleagues (experiment 3) are series of timed movements. For example, Schmidt
shown in figure 7.8, where the VE/MT% is plot- and colleagues (1978, 1979) studied this problem
ted against movement velocity. As the velocity by having subjects make rhythmic back-and-forth
increased, the errors in timing decreased markedly movements of a lever in time to a metronome. The
at low velocities, and decreased more gradually subject exerted timed forces with a strain gauge
with further increases in velocity. Similar effects on the handle. There were four different goal
have been seen in experiments involving wider. MTs (200, 300, 400, and 500 ms), and the major
ranges in velocities and distances. (Newell, Carl- concern was the variability in the duration of the
ton, Kim, & Chung, 1993; Jasiewicz & Simmons, force bursts (i.e., impulses) produced by muscular
1996). This velocity effect is yet another example of action during these movements. The results of this
the temporal speed-accuracy trade-off. experiment are illustrated in figure 7.9, where the
An interesting application of these findings within-subject variability in impulse duration is
relates to baseball: As a batter swings "harder" plotted against the MT imposed by the metro-
(with a smaller MT, a larger movement distance, nome. The strong, nearly proportional relation
or both), the errors in timing should tend to found between these two variables is certainly in
decrease, not increase. Note that this prediction keeping with the data presented in the previous
is for errors in timing and does not relate to errors section about movement durations as a whole.
in spatial accuracy; we consider these issues in Synchronization and Continuation Tasks
more detail later in the chapter.
The back-and-forth, force-burst task used in the
studies by Schmidt and colleagues (1978, 1979)
Repetitive Movement-Timing Tasks
represents a variation of repetitive timing tasks
The finding that movement duration becomes that have been studied for over a century (e.g.,
more consistent in time as the MT decreases has Bartlett & Bartlett, 1959; Dunlap, 1910; Stevens,
been shown in tasks in which the subject makes a 1886; Woodrow, 1932). There are two fundamen-
discrete movement with the intention of achieving tal versions of this task. The subject's goal in the

20

15

~
.....
~ 10
w
>

0 25 75 150 225
Velocity (cm/s)

FIGURE 7.8 VE 1 (expressed as VE1 I MT%) as a function of the movement velocity. (Dividing VE 1 by MT theoretically "can-
cels out" the effect of MT on errors.)
Adapted from Newell et al. 1980.
238 Motor Control and Learning

30

U)
.§.. 25
c:
.!2
'lii
:i 20
"C
Cl>
(/)
:;
c.
.§ 15

c~ 10
~
:.0
-~ 5
m
>

Average MT (ms)

FIGURE 7.9 Variability in impulse duration as a function of the movement time (MT).
Reprinted, by permission, from R.A. Schmidt, H.N. Zelaznik, and J.S. Frank, 1978, Sources of inaccuracy in rapid movement. Information processing in motor control and
learning, edited by G.E. Stelmach (Champaign, IL: Human Kinetics), 197.

synchronization task is to produce a motor action consequences to be perceived as synchronous.

(such as a finger movement) resulting in a "tap" Alternative, viable hypotheses exist, however,
that coincides with an external perceptible event and the negative-asynchrony effect continues to
(such as the auditory beeps of a metronome). stimulate a vibrant research area (e.g., Mates &
In the continuation task (Stevens, 1886), a typi- Aschersleben, 2000; Repp, 2001, 2005).
cal trial begins with a series of synchronization In some respects the continuation task may be
taps; then at some point the experimenter turns considered a simpler version of timing than the
off the pacing stimulus (e.g., metronome), and synchronization task, in that there is no objec-
the subject's goal is to continue tapping at the tive, pacing stimulus to anticipate. For example,
same pace. Both the synchronization task and Balasubramaniam, Wing, and Daffertshofer
the continuation task have been used in numer- (2004) found that movement trajectories were
ous investigations of human motor performance. quite "smooth" when subjects produced finger
One of the curious findings with the syn- taps in the absence of a pacing signal (see top
chronization task is a negative asynchrony-that illustrations in figure 7.10). In contrast, producing
is, subjects tend to produce taps prior to the finger taps in time with a metronome, either "on
pacing tones. Aschersleben' s (2002) review of the the beat" (synchronizing) or "off the beat" (syn-
evidence shows that the asynchrony is larger in copation), resulted in considerable asymmetry
musically untrained than in trained subjects and in terms of the kinematics. As seen in the middle
that the amount of asynchrony can be reduced and bottom illustrations in figure 7.10, the down-
with practice and augmented feedback (see chap- ward deflections of the finger to produce a tap,
ter 12). The suggestion is that negative asynchro- regardless of whether they were made on or off
nies arise due to a closed-loop process in which the beat, were produced much faster than upward
feedback from the motor output is compared to movements. These asymmetries were associated
the auditory stream from the pacing stimulus. with timing accuracy, possibly indicating the use
If the feedback information is processed more of velocity changes as an error correction strategy
slowly than the pacing-stimulus information, (Balasubramaniam, 2006).
a tap must be produced sooner in order for However, in other respects the continuation
the central representations of the two auditory task is more complex than the synchronization
 Principles of Speed and Accuracy 239

-0
Q)
()
ctl
Cl.
c
:::::i

c:
Q)
E
Q)
(.)
z
cu Q
ii
c"'
b

~
LL
LL
·c:;
0
Q
~
c
Time Displacement

FIGURE 7.10 Finger trajectories in three timing tasks. (a) Top figures illustrate a typical position-time graph (left side) and
velocity-position graph (right side) for an unpaced tapping condition; (b) middle figures illustrate tapping on the beat; and (c)
bottom figures illustrate tapping off the beat. The top figures illustrate trajectory symmetry, while the middle and bottom fig-
ures illustrate much faster velocities in the tap portion of the movement cycle.
With kind permission from Springer Science+Business Media: Experimental Brain Research, "Keeping with the beat: Movement trajectories contribute to movement timing;' Vol.
159, 2004, pgs. 129-134, R. Balasubramaniam, A.M. Wing, and A. Daffertshofer.

task. In the absence of the pacing stimulus, the words, the adjacent intertap intervals were found
subject must establish a central representation of to vary inversely with each other (i.e., they dem-
time and adjust performance based on a memory onstrated a negative covariance or correlation),
for the pacing rate. Much of the research in the although intervals that were separated by two
past four decades has been based on an open- or more intervening (i.e., not adjacent) intervals
loop, hierarchical model of timing performance were independent of each other (i.e., they had a
using the continuation task, which we consider covariance of zero).
next. To account for these and other observations
in continuation timing, Wing and Kristofferson
The Wing-Kristofferson Hierarchical
(1973b) developed an open-loop, hierarchical
Timing Model model of performance, illustrated in figure 7.11.
Recall that in the continuation task, a subject There are two levels in the hierarchy: a central
synchronizes a particular movement (e.g., a tap) timekeeper level and a motor implementation level.
with a series of paced tones. After 20 or so tones The model states that, in the absence of a pacing
have been presented, the metronome is stopped stimulus, a centrally generated "internal clock"
and the subject continues to produce 30 or so sends pulses via the central nervous system that
more taps. Performance of these latter taps is of result in the execution of a movement (e.g., a tap).
primary interest. As we have noted several times The internal clock-which can be thought of as
previously in the chapter, variability in timing is similar to the timing generated by the motor pro-
related closely to the MTs of these intertap inter- gram (chapter 6)-is responsible for generating
vals. As the mean intertap interval decreases, the the desired goal durations (C;); the resulting times
overall consistency increases. However, some- between successive taps are the intertap intervals
thing else goes on here as well. In Stevens' (1886) (ITI). Any variability in the rate at which the clock
original experiment, a slower intertap interval times the pulses (or intervals) will result in ITI
tended to be followed by one that was faster, and variance that has a central source. There is a delay,
vice versa (e.g., see description in Wing, 1980, however, from the time a central command is
and figure 4.2 in Vorberg & Wing, 1996). In other issued until a tap is produced by the motor system.
240 Motor Control and Learning

Central timekeeper intervals (Cj)

Central timekeeper
pulses (Pi)

Delays in motor
implementation (Mj)

Taps (Tj) T5

lntertap intervals (ITlj)

FIGURE 7.11 The Wing-Kristofferson hierarchical model of timing. A high-level, central timekeeper produces pulses (e.g.,
Pi, Pi+P etc.) that generate a time interval (C;). The duration of each observed intertap interval (ITli) is influenced by the Ci and
the delays in implementing the motor activity that initiate (Mi) and terminate (Mi+ 1) the ITI. Measures of variability in timing are
observed and estimated using statistical techniques.
Adapted from Tutorials in motor behavior, A.M. Wing, pgs. 469-486, The long and short of timing in response sequences, edited by G.E. Stelmach and J. Requin, Copyright 1980.
By permission of A.M. Wing.

In fact, each ITI is affected by two motor implemen- clock variance and motor implementation vari-
tation delays, because one tap defines the initiation ance. As the figure illustrates, the variability due
of the ITI (M) and the next tap defines the termi- to motor implementation processes contributes
nation of the ITI (Mi+ 1). Viewed another way, any very little to the total, and the effect is almost
one motor delay (e.g., M2) will have an influence entirely due to variability in central (clock) pro-
on both the preceding interval (ITI1) and the suc- cesses.2
ceeding interval (ITI 2). Thus, a short M2will tend Wing (2002) has summarized a large number
to shorten the preceding ITI and lengthen the next of studies in which processes affecting clock
ITI. So, a second source of variation at the motor and motor implementation variances were
delay level can add variability to the observed ITis separated. For example, factors such as the age
that is independent of the clock variability. Wing of the subject (Greene & Williams, 1993) and
and Kristofferson developed statistical methods increased cognitive demands due to a secondary
for estimating the value of the clock variance and task (Sergent, Hellige, & Cherry, 1993) greatly
the motor delay variance. influenced clock variance but had no effect on
motor variance. Compared to unimanual tap-
Factors Affecting Clock and Motor ping performance, adding a second limb so that
Variance both limbs are tapping together in continuation
Figure 7.12 presents data from Wing (1980), in has the effect of reducing the ITI variance, almost
which the variance associated with the duration all of it due to the reduction in clock variance
of the ITI has been decomposed into measures of (Drewing & Aschersleben, 2003; Helmuth & Ivry,
 Principles of Speed and Accuracy mm 241

Two Types of Timing?

450 Most of the studies discussed in this section have
400
considered timing tasks involving the tapping of
a finger in time to a metronome that either is cur-
~ 350 rently present or was present in the recent past (i.e.,
§.
GI
300
Stevens' and Wing-Kristofferson's synchronization
()
c
Ill
and continuation tasks). Timing has been consid-
·;::
Ill 250 ered to be explicit in these tasks in the sense that the
>
'ti
200
nature of the task and the subject's internal timing
~ processes represent compatible goals-the "envi-
E
:;:::
UI
150 ronment" specifies a time goal that the subject tries
w
100
to match with specifically timed movements. In
studies of individual differences in timing ability,
50 Robertson and colleagues (1999) found that people
who are consistent timers for slow tapping rates
(e.g., 800 ms per tap) are also consistent timers
for fast rates (e.g., 400 ms per tap). That is, the
within-trial variabilities in tapping are similar
FIGURE 7.12 The effects of the intertap interval (ITI) among individuals for the two tasks, resulting in
duration on clock and motor variance. high correlations between people-timing appears
Adapted from Tutorials in motor behaviour, A.M. Wing, pgs. 469-486, The long and
short of timing in response sequences, edited by G.E. Stelmach and J. Requin,
to be a type of general ability (see also the section
Copyright 1980. By permission of A.M. Wing. "Exceptions to Henry's Specificity Hypothesis" on
p. 305). In contrast to this tapping timing task is a
circle-drawing timing task (e.g., Robertson et al.,
1996). Interestingly, though, if hand tapping is
1999), in which subjects produce circles continu-
paired with foot tapping, then clock variance is
ously with a defined completion cycle time (e.g.,
reduced while motor variance actually increases
400 or 800 ms for each revolution). Again, high
(Ivry, Richardson, & Helmuth, 2002). Additional
interindividual correlations exist for the variances
practice does reduce motor variance, but not
in producing timed circles at different cycling
clock variance (Drewing & Aschersleben, 2003).
rates. However, there is no correlation between
If the length of one of the "effectors" is altered in
timed taps and timed circles, suggesting that these
bimanual timing, then only motor variability is
two tasks represent different fundamental timing
affected (Turvey, Schmidt, & Rosenblum, 1989).
processes. This suggestion is supported by studies
The Wing-Kristofferson model has also been involving patients with cerebellar damage, who
useful in contrasting the influence of neural revealed deficits in performing tapping timing
disorders on timing in various patient groups. tasks but not circle timing tasks (Spencer, Zelaznik,
For example, Ivry, Keele, and Diener (1988) Diedrichsen, & Ivry, 2003). These and related find-
found that patients with one type of lesion in ings support a dissociation of processes involving
the cerebellum showed increased clock variance these two types of timing tasks (e.g., Studenka &
whereas another group, with lesions in a differ- Zelaznik, 2008; Torre & Balasubramaniam, 2009;
ent part of the cerebellum, showed elevation Zelaznik et al., 2005).
only in motor variance.
The Wing-Kristofferson model has proven
useful in identifying dissociations in motor Central Contributions
timing, which is a very important contribution to the Spatial
to understanding processes in motor control.
Although various alterations to the model have Speed-Accuracy Trade-Off
been proposed over the years (e.g., Collier &
Ogden, 2004; Vorberg & Wing, 1996), the original To this point in the chapter we have presented
Wing-Kristofferson model remains a powerful three relatively different ways in which speed
and important research tool in studies of motor and accuracy are related. Two of these pertain to
timing. spatial trade-offs, and one pertains to a temporal
242 !!ii Motor Control and Learning

trade-off. Some theoretical issues related to timing impulse is a critical determiner of movement, and
were presented in the last section, and others will variability in impulses is a critical determiner of
be discussed in more detail in chapter 8. In this the variability in movement, then an analysis of
section, we concentrate on the spatial trade-off. the variability of the components of the impulse
Two general perspectives have dominated much (variability in force and variability in duration)
of the thinking; these perspectives correspond in should allow considerable insight into the sources
general to the topics discussed in chapters 5 and of spatial errors in movement control-at least in
6. With regard to the generation of an impulse to relatively quick actions where feedback processes
propel a limb toward a target, one perspective do not play a major role. Schmidt and colleagues
has been to consider the importance of central provided the early modeling of these phenomena
contributions. The other perspective has been to (Schmidt et al., 1978, 1979), and revisions of the
examine how corrective processes contribute to same idea were provided later by Meyer and
the speed-accuracy trade-off. In the following colleagues (Meyer et al., 1982, 1988).
sections we discuss the major ideas that have
Force-Variability Principles
been suggested within these general theoretical
approaches. In this section we consider factors that produce
Keele (1981, 1986) suggested that central pro- variability in the amount of force generated by
cessing of rapid aiming movements can be classi- the activation of the motor program. This is an
fied under two broad categories. In one category, important issue for understanding the processes
the thinking is that central commands specify the underlying skillful behavior. All that the muscles
distance that must be traveled to reach a target. can do to bones is to exert force on them, with this
Distance programming is assumed by models application being adjustable in terms of amount
of impulse variability, whereby the agonist and of force or in terms of the temporal onset and
antagonist muscles produce bursts of force (open duration of that force. Complex patterns of force
loop) that propel a limb a specific goal distance. produced during a particular contraction are
The other general class of models assumes a pro- presumably also under the control of the motor
gramming of commands that specifies a target's system. If the activation sent to the muscles is
location. Location programming is assumed by a preprogrammed, then any factors causing the
class of models that specify equilibrium-tension amount of force to deviate from the intended
ratios between agonist and antagonist muscle amount of force will cause the movement to devi-
ate from its intended path or fail to meet its goal.
groups.
Put simply, muscular forces produce movements, and
Impulse-Variability Theory variability in muscular forces produces variability in
movements.
A number of the principles related to the linear Schmidt and colleagues (1978, 1979) began
speed-accuracy trade-off form the basis for a series of studies on the relationship between
impulse-variability models of movement control. forces involved in quick, preprogrammed move-
Two principles are of critical importance: (a) The ments and their within-subject variability, and
variability in the duration of a group of muscular these findings are described next. In relation to
contractions is directly proportional to the mean slower movements, there has been some inter-
duration; and (b) the variability in force applied in est in these questions for over a century (e.g.,
a group of attempts is an increasing function of the Fullerton & Cattell, 1892; see Newell, Carlton,
mean force to approximately 65% of maximum, & Hancock, 1984, for a review), although the
with a leveling off or slight decrease thereafter. theoretical orientations of this earlier work were
The reason these principles have such importance considerably different. The initial issue addressed
is that they define variability in the two dimen- in the studies by Schmidt and colleagues con-
sions of the impulse-the primary determinant of cerned the relationship between the amount of
what the limb will do when muscles attached to force produced and the resulting within-subject
it are activated. The notion of the impulse, that variability in that force. For example, why did
is, the forces produced over time, was discussed an aiming movement with twice the amplitude
in chapter 6 (section on the impulse-timing (MT constant) produce approximately twice the
hypothesis). So if, as we have argued earlier, the error inhitting a target (e.g., Schmidt, et al., 1979;
 Principles of Speed and Accuracy 111 243

Woodworth, 1899)? According to the generalized

motor program (GMP) idea (chapter 6), in order 0.75
to produce a movement of twice the amplitude,
the program would remain the same, but the
overall force parameter would be increased so 0.60
that the limb would travel twice the distance in
the same MT. Could it be that when the forces in ~
the limb are increased so that the limb can travel ~ 0.45
farther, the variability in this force is increased as .S!
.lO:
well, making the output of the movement more m
~
variable? This was the hypothesis. Ci 0.30
c
Moderate Force Levels The subject's task was (/)

to produce quick "shots" of force against an 0.15

immovable handle. Attached to the handle was
a strain gauge that measured the amount of force
applied; this information was sent to an oscil-
2 4 6 8 10 12
loscope screen so that the subject could see the
Amount of force (kg)
force levels. A zero-force level was indicated at the
bottom of the screen, and increasing force applied
FIGURE 7.13 Variability in peak static force as a func-
to the handle would move the dot upward. Sub- tion of the amount of force to be produced.
jects attempted to produce ballistic "shots" of Reprinted, by permission, from R.A. Schmidt, H.N. Zelaznik, and J.S. Frank, 1978,
force that would peak exactly at a target location Sources of inaccuracy in rapid movement. Information processing in motor control
and learning, edited by G.E. Stelmach (Champaign, IL: Human Kinetics), 19Z
on the screen, the goal being to produce the same
peak force on each trial. The peak force and the
within-subject variability (SD) of the peak forces the speculation that, as the amount of force was
were measured; subjects were asked to produce increased (in order, for example, to move farther
different amounts of force during separate series in the same MT), the variability in force would
of trials. increase in nearly direct proportion, leading to
In figure 7.13, the SD in peak force is plotted increased variability in the movement's spatial
as a function of the amount of force the subjects outcome-in this case, in hitting a target accu-
were asked to produce. A clear linear relationship rately.
between force and its variability can be seen. In Near-Maximal Force Levels Will the linear
a number of other experiments, force and force relationship between force and force variability
variability were found to be linearly related in continue to hold as the forces are increased even
situations in which the forces were much smaller further? This is an important question, as many
than indicated in figure 7.13 (40 g elbow flexion of the skills performed by athletes have very high
measured at the wrist) and also when they were force requirements, such as those for the forces in
as large as 65% of the subject's maximum (Newell the quadriceps muscles generated during a foot-
& Carlton, 1985; Schmidt et al., 1978, 1979; Sher- ball punt for distance. With such large forces, one
wood & Schmidt, 1980). might expect that there would be no "room" for
These results were from static (isometric) con- variability near the performer's maximum. If the
tractions, though, and it was not clear that such subject actually did produce the theoretical maxi-
relationships would occur in actual movements, in mum on every trial, there would be no variability
which the muscle shortens as it is producing force (because the SD of a set of constant values is zero).
against the bone. Sherwood and Schmidt (1980) Thus, as the maximum force is approached, one
examined this issue in an experiment in which could imagine a point of maximum force vari-
forces were recorded during a simple ballistic- ability (perhaps near the middle of the range in
timing movement. Again, a linear relationship force capability) and a gradual decline in force
between force and force variability was found, variability (i.e., more consistency) as the forces
suggesting that this principle had a strong role are increased further. With this reasoning, an
in movements. These data certainly supported inverted-U function should exist between force
244 mi Motor Control and Learning

variability and force across the total range of ated with a longer time to achieve peak force,
human force capability. and they argued that this should have the effect
Figure 7.14 presents the data from another of producing an artificial inverted U at high force
experiment by Sherwood and Schmidt (1980), in levels. In a subsequent experiment in which the
which the forces were increased to very near the time to peak force was held strictly constant,
subject's maximum. The figure shows the vari- Sherwood, Schmidt, and Walter (1988) showed
ability in peak force as a function of the level of that the inverted-U function was no longer as pro-
force produced, as before. But in this case a maxi- nounced as indicated in figure 7.14. Thus, perhaps
mum point occurred in force variability, with a rather than an inverted-U function, the relation
strong tendency for decreasing force variability as between force and force variability follows a
the forces were increased further. The peak force curvilinear, negatively accelerated function. Beyond
variability occurred, on the average, at about 65% about 65% of maximum force, force variability
of maximum force for these subjects; the largest tends to increase at a decreasing rate, leveling
force examined was about 92% of maximum. off somewhere near the maximum force capabili-
Inverted-U or Negatively Accelerating Func- ties (or perhaps decreasing very slightly-Sher-
tions? Some controversy has developed with wood et al., 1988). Theoretical modeling of force
regard to the inverted-U function shown in figure summation characteristics by Ulrich and Wing
7.14. Newell and Carlton (1985, 1988; Carlton & (1991) suggests that both the inverted-U and the
Newell, 1988; Newell et al., 1984) have criticized negatively accelerated functions may be correct
the procedures used to obtain this inverted-U in certain situations.
result on the grounds that the time to peak force Modeling the Initial Impulse
was not controlled in these experiments. They
showed that larger peak forces tend to be associ- It is easy to understand the general idea of
impulse-variability principles if one considers
only the initial impulse for acceleration in a rapid
movement. A good example is the ballistic-timing
task, in which the subject moves a slide along a
trackway, past a switch at the end, and attempts to
1.25
achieve a goal MT (initial movement until arrival
at the switch). Here the subject is accelerating for
Ci 1.15
the entire (measured) movement, so the move-
~ ment is governed by the initial impulse for accel-
CD
~
eration. When the experimenter changes either
0
:;;; 1.05 the movement amplitude (A) or the goal MT, the
subject's movement is assumed to be handled
& by changes in the parameters of the GMP for
0
~ 0.95 this rapid action (see chapter 6). Let us examine
the effect of variables like movement amplitude
and movement time-critical participants in
0.85 the speed-accuracy trade-off-on the nature of
impulse variability in this task. 3
The Effect of Movement Amplitude Consider
what happens to the impulse as the person is
5 10 15 20 asked to move twice as far in the same MT. In
Amount of force (kg)
this case the movement-duration parameter of
the GMP must be fixed, so the duration of the
FIGURE 7.14 Variability in peak dynamic force as a
impulse will also be constant. However, the
function of force produced, including near-maximal force
values. overall force parameter must be doubled so that
Adapted from D.E. Sherwood and R.A. Schmidt, 1980, ''The relationship between the force produced will be twice as large as for
force and force variability in minimal and near-maximal static and dynamic
contractions," Journal of Motor Behavior 12: 75·89, adapted with permission of
the shorter movement. From the earlier section
the publisher (Taylor & Francis Ltd, http://www.informaworld.com). on force variability (figure 7.13), we know that as
 Principles of Speed and Accuracy Ill! 245

the amount of force is doubled, the variability is component of impulse variability is inversely
approximately doubled as well. Therefore, with a related to the squared MT, or
doubled force but a constant duration, the entire Force variability= k3 X (1 I MT2) (7.6)
impulse will have twice the variability, with all
of the increase in variability occurring because We see that the overall impulse variability is
of the changes in the amplitude of the impulse. related both (a) directly to the MT in the tempo-
The overall result is that the impulse variability ral dimension and (b) directly to 1 /MT2 in the
is linearly related to the movement amplitude: force dimension. Combining these two effects
of impulse variability (equations 7.5 and 7.6)
Impulse variability = k1 X (A) (7.4)
produces the generalization that
where k1 is an empirical constant, and A is the Impulse variability = k4 X (MT / MT2 )
movement amplitude (the intercept is zero = k4 (1 I MT) (7.7)
because proportionality is assumed; see the sec-
tion on empirical equations in chapter 2). That is, total impulse variability is proportional
to l/MT.
The Effect of Movement Time Next, consider
what happens to the impulse as the MT is halved Amplitude and Movement Time When we
with a constant movement amplitude. From the combine the effects of A and the effects of MT from
notions of GMPs, we know that the overall dura- equations 7.4 and 7.7, we obtain the relation that
tion parameter will be halved as well, so that all Impulse variability = k X (A I MT) (7.8)
of the impulses in the action will be half their
original duration. We know from figure 7.10 that where the total variability in an impulse for
the variability in the duration of an interval is accelerating a limb is directly related to the
directly proportional to its duration, so halving amplitude of the movement (A) and inversely
the duration of the impulse should reduce its vari- related to the duration of the movement (MT).
ability in time by half also. Generally, the temporal Because the velocity of a movement after an
variability of an impulse is linearly related to MT: impulse is directly proportional to the size (area)
of the impulse, and because the variability in the
Temporal variability = k2 X (MT) (7.5) impulse leads directly to variability in velocity,
where k2 is another proportionality constant this relation implies that the variability in the
(again, the intercept is zero). This is interest- velocity of a movement when an impulse has
ing (and somewhat counterintuitive) because stopped acting on it will be directly proportional
the variability of one component (temporal) of to A/MT as well. This is a key feature in impulse-
the impulse becomes more consistent as the MT variability modeling, and many other interesting
is shortened, while the variability of another predictions emerge from it.
component (force) becomes less consistent at the Impulse-Variability Principles
same time. in Movement Control
But along with this shortening of impulse
duration as the MT is halved, the impulse must The original impulse-variability theory (Schmidt
increase in amplitude (force), so that the increased et al., 1978, 1979), which concerned the effects of
velocity needed for the shorter MT can be various movement variables on initial-impulse
achieved. From physics we know that for a move- variability, seems to account relatively well for
ment with twice the velocity (i.e., half the MT), the the behavior of single impulses in a number of
area of the impulse must be twice as large. But the rapid movement tasks (see Schmidt, Sherwood,
duration is half as long, so the amplitude must be Zelaznik, & Leikind, 1985, for a review). As a
four times as large (i.e., four times the amplitude result, the model accounts fairly well for accuracy
together with half the duration yields twice the in tasks in which only a single impulse is acting,
area). Then, because of force-variability principles such as the ballistic-timing tasks described at the
(figure 7.14), the force variability is increased by a beginning of this section.
factor of four as well, so that halving the MT pro- Errors in Ballistic Timing For ballistic-timing
duces a fourfold increase in the force component tasks, we have already mentioned that the vari-
of impulse variability. More generally, the force able error in timing (VE1) has been shown to be
246 !!JI Motor Control and Learning

nearly proportional to the MT (tables 7.1 and 7.2, which pulls obliquely on it, their combined action
figure 7.8). The model also accounts for spatial being to move the limb toward the ball. What will
errors-the error in position at the moment that be the effect on spatial accuracy of increasing the
the goal MT has elapsed, or We. In this case the mass of the bat, with MT constant?
model predicts that we should be independent If muscle groups A and B are contracting with
of MT, and several experiments show this to be less than 65% of maximum force, then increasing
essentially so (e.g., Schmidt, 1994; Schmidt et al., the mass of the bat will require increased force
1979, 1985). Therefore, changing the MT affects output in each of them, and the force variability
mainly timing errors in this timing task, not spatial will increase in both muscle groups at the same
errors. The principle is just the opposite when A is time (e.g., figure 7.15). This should result in pro-
manipulated in this task. Here, increasing A causes gressively larger variations in the direction of bat
nearly proportional increases in spatial errors but travel and in progressively lower probability that
causes almost no effect in the timing errors. 4 This the bat will hit the ball. But what if the forces in
seemingly curious set of findings is derivable from the two muscle groups are larger, perhaps greater
the model, and is related to the tight connection than 65% of maximum? Here, with increased
between space and time in such situations involv- mass of the bat, there would be (from data in
ing tasks in which time is the dependent variable figure 7.14) a decrease (or at least a much smaller
(see, e.g., Newell, 1980; Newell et al., 1984). increase) in the force variability as the forces are
Spatial Errors in Very Rapid Movements Some increased, resulting in increased consistency of bat
applications of impulse-variability notions can be travel as the mass is increased.
seen in a task like hitting a baseball, where the limb Schmidt and Sherwood (1982) conducted an
seems to be driven by a single impulse, with no experiment that seems to show just that. Subjects
deceleration until after the bat passes over the plate made horizontal forward arm swings to attempt
and the ball is struck. Here, in addition to being to hit a target, with a MT that was to be as close to
temporally accurate, the batter must be spatially 200 ms as possible. Figure 7.16 shows the spatial
accurate so that the swing will not miss the ball. We accuracy in hitting the target as a function of the
can think of such a movement as shown in figure mass added to the handle, measured as VE and
7.15, where the limb (or limb plus bat) is moving total variability (E). As mass was added to the
horizontally toward a ball. Assume that the limb is unloaded handle, there was first an increase in
controlled by two muscle groups, A and B, each of spatial errors; but the function soon peaked, so

Muscle
group A

Muscle
group B

FIGURE 7.15 The limb conceptualized as a mass moved by two forces (muscles) operating in different directions.
 Principles of Speed and Accuracy 247

18

17

16
'E
E
....... 15
~
1i
CIS
·;: 14
CIS
>
13

12

11

Added mass

FIGURE 7.16 The effect of added mass on the spatial accuracy of a 200 ms arm swing movement. The values above the
x-axis are the percentages of maximum force produced with that added mass.
Reprinted, by permission, from R.A. Schmidt and D.E. Sherwood, 1982, 'Jll.n inverted-U relation between spatial error and force requirements in rapid limb movements:
Further evidence for the impulse-variability model," Journal of Experimental Psychology: Human Perception and Performance 8: 165. Copyright© 1982 by the American
Psychological Association.

that the errors decreased with further increases I force-variability function seen in figure 7.15. It is
in mass. Thus, there was an inverted-U relation interesting to note that when the movement was
between spatial accuracy and mass for this task, very rapid and required 84% of maximum force,
with the peak at approximately the mass that the amount of error was nearly the same as that
resulted in about 60% to 70% of maximum force. in a very slow (130 ms) movement.
This corresponds surprisingly well with the peak Both of these results provide exceptions to the
in the force/force-variability function found in speed-accuracy trade-off ideas presented earlier,
figure 7.14. as increasing speed (or increasing mass) resulted
A similar experiment was conducted with in increased accuracy when the movements were
variations in MT, this time with the mass held very violent. This supports indirectly the impulse-
constant. If decreased MT results in increases in variability model, in which the variability in forces
force beyond about 65% of maximum, then we is expected to be a major determiner of the spatial
should find that making the movement faster accuracy in these rapid actions. There are also
results in increased accuracy, not less as would int~resting practical implications. Perhaps telling
be expected from the speed-accuracy trade-off a baseball player to swing slowly to "make con-
effect. Figure 7.17 illustrates the results. As the tact" would actually result in less spatial accuracy
average MT decreased from 158 ms to about than giving an instruction to swing "hard" (see
102 ms, there was a progressive increase in the "Principles of Motor Control Applied to Baseball
spatial errors, as one would expect. But when Batting," where we develop this idea more fully).
the MT was reduced further to 80 ms, requiring
approximately 84% of maximum force, there was Limitations of the Impulse-Variability
a decrease in the spatial errors, with the func- Theory
tion showing a peak error somewhere near 50% One limitation of impulse-variability theory is
of maximum contraction (102 ms MT). Again, that it does not account for movement accuracy
this is compatible with the peak in the force in aiming tasks (e.g., the linear speed-accuracy
 12

11

10

e
s> 9
~
:.cIll 8
·;:
Ill
>
7

FIGURE 7.17 The effect of movement time (MT) on the spatial accuracy in a horizontal arm swing movement. The values
above the x-axis are the percentages of maximum force produced with that movement time.
Reprinted, by permission, from RA Schmidt and D.E. Sherwood, 1982, ')\n inverted-U relation between spatial error and force requirements in rapid limb movements:
Further evidence for the impulse-variability model;' Journal of Experimental Psychology: Human Perception and Performance 8: 165. Copyright© 1982 by the American
Psychological Association.

Principles of Motor Control Applied

··to Baseball Batting
We now attempt to use a number of the principles discussed in this section to understand a com-
plex skill like bcatting a pitched baseball. Let us review some basic facts about hitting a baseball.
In figure}.18are.datafrom a number of studies referred to earlier, as well as from Hubbard and
Seng (1.954). Essentially, it takes about 460 ms for a ball (moving at 89 mph) to travel from the
pitchertd the plate. Hubbard and Seng found that the MTof the bat (the time from the first move-
me~t of the bat until it reached the plate) was about 160 ms; this is equivalent to about 21 ft (6.4
m} ot·.ba.ILtravel.
Whatwouldhappen if the batterwere to reduce the MT of the swing by 20 ms, from 160 ms
to 140 ms? Of course, if contact with the ball is made,· the ball travels farther and faster, which
is usually beneficial in baseball. But, what effect does speeding up the swing have in terms of
acctJracy?
First, in terms of the decision processes prior to swinging, as the person watches the ball, the
most important aspects of the ball's flight are assumedly those closest to the batter.But, by speeding
upthe swing to 140 ms (seefigure?.18), the batter can view the ball for 20 ms longer. More impor-
tantly, those 20 ms occur during the time that the ball is as close as possible to the batter before
the ultimate decision has to be made to trigger the action. In this way, the batter should be able to
make more accurate predictions about the spatial relation of the ball to the batter when it arrives in
the hitting zone, and therefore more effective choices regarding the swing.

248
 What does swinging faster do to the timing of the bat swing initiation? Fr0m the discussions
about anticipation in chapter 3, in order for the bat to arrive at the. plate at the proper time, the
batter must anticipate the processes between the final (internal) "go" signal and the end of the
movement. This is called effector anticipation. Here, this interval begins at the point marked
"Decisions end" and ends when the bat reaches the plate. When the MT is 160 ms, this interval
is 328 ms, and when the MT is 140 ms it is 308 ms. Because the accuracy in timing an interval
is diminished as the interval length increases, more accurate effector anticipation occurs when
the swing is 140 ms versus 160 ms, and less variability in the time of swing initiation should
result. Schmidt (1969c), using a laboratory task analogous to batting (figure 7.7), found that
when MT was decreased, the variability in when the bat swing was initiated was decreased
as well. A third advantage of swinging more rapidly is that the movement. itself is more tem-
porally consistent as the MT is shorter. A number of studies (e.g., Newell et al., 1979, 1980;
Schmidt et al., 1979) show decreased MT variability as the MT is decreased. Newell and col-
leagues (1979, 1980) found a stabilizing effect of movement velocity essentially independent
of MT. So, when the batter swings faster, the movements themselves become more temporally
consistent.
The combination of increased consistency in the time of swing initiation and in the duration of
the swing itself should lead to increased consistency in the time at which the bat arrives at the
plate. In the laboratory task mentioned previously (figure 7.7), Schmidt {1969c) found that the
temporal accuracy in meeting the "ball" was greater when the MT was shorter (see table 7.1 ).
If the level of muscle contraction in the bat swing is above about 65% of the batter's maximum,
then increased swing speed, or increased bat mass with a constant swing speed, should produce
increases in the spatial accuracy of the swing (see figures 7.16 and 7.17). This is so because with
increased levels of contraction there is increased consistency in force production, which results
in increased consistency in where the limbs go during the action. These kinds of effe.cts have
been shown for laboratory tasks (Schmidt & Sherwood, 1982), as well as in sport-related tasks
(Zernicke & Roberts, 1978).

FIGURE 7.18 A time line showing the critical events in hitting a pitched baseball. (The top example has a 160
ms movement time [MT], and the bottom example has a 140 ms MT.)

Iii 249
250 m Motor Control and Learning

trade-off) in which a number of impulses are pre- emphasized more recently is that the contractile
sumably combined to produce a complete action portion of muscle has elasticity as well, such that
(e.g., as in figures 7.5 and 7.6). So, the impulse- the entire muscle-tendon unit is analogous to a
variability model falls considerably short of the "complicated spring." This concept has influ-
goal of explaining speed-accuracy phenomena enced thinking about what muscles do when they
in general. A modification of the theory, using move bones in skilled actions (see Partridge &
similar assumptions about GMPs, force and time Benton, 1981, for a review).
variability, and so on, was provided by Meyer One way to study the properties of muscles
and colleagues (1982). This view does represent is to describe them in terms of a length-tension
an improvement in certain ways, but it suffers curve-that is, the relation between a muscle's
from a number of other problems that seem length and the tension that it is capable of pro-
to argue against its adequacy as an account of ducing at that length under a given level of con-
speed-accuracy effects (see Schmidt et al., 1985; traction. In an anesthetized animal, the length
Zelaznik, Schmidt, & Gielen, 1986). of the muscle can be predetermined; the nerve
Overall, the impulse-variability theory is stimulated artificially so that the level of acti-
accounts reasonably well for certain types of bal- vation to the muscle can be controlled, and the
listic actions that do not require feedback. Thus, resulting tension is measured. Such a procedure
the theory provides an important description of can produce a family of length-tension diagrams,
some of the centrally generated errors that result with one curve for each of the levels of contrac-
in speed-accuracy trade-offs. Later in this chapter tion that the experimenter uses. Some of these
we describe a more recent model of Fitts-type curves from Rack and Westbury (1969) are shown
tasks by Meyer and colleagues (1988, 1990) that in figure 7.19, in which five levels of activation
describes how impulse-variability principles are were used (defined in terms of the number of
involved in both the initial and corrective por- impulses of electrical stimulus given per second).
tions of aiming movements. Notice that at all levels, a generally increasing
relationship was found between the length of
Equilibrium-Point Theories the muscle and the tension it developed. This
Alternatives to theories suggesting that move- relationship is roughly what we would expect
ment distances are controlled on the basis of if the muscle were a spring attached to a lever
programmed impulses are theories suggesting (the bone). A fundamental mechanical principle
that the movement end point is programmed and about springs-Hooke's law-is that the tension
that muscle and mechanical properties deter- produced is directly proportional to the amount
mine the trajectory. The relation of these theo- the spring is stretched (i.e., proportional to the
ries to speed-accuracy trade-offs per se has not spring's length). Figure 7.20 is a hypothetical
been fully investigated (Latash, 1993; Latash & length-tension diagram for springs. The four
Gutman, 1993). However, these ideas represent curves represent four different springs, each
a major theoretical advance regarding the way with a different stiffness. Stiffness is the force
limb positioning movements are controlled by required to lengthen the spring by one unit (i.e.,
the central nervous system, so it is important to the change in tension divided by the resulting
consider them here. change in length), represented as the slope of the
length-tension curve.
The Length-Tension Diagram
Muscles, and tendons that connect muscles to Mass-Spring Mechanisms
bones, have a certain amount of compliance, The realization by Asatryan and Feldman (1965;
or springiness. In an older view of muscle (e.g., Feldman, 1966a, 1966b)5 that muscles could, in
Huxley, 1969), the notion was that the contractile certain gross ways, behave something like com-
component of muscle was responsible for the plex springs has revealed a possible mechanism
production of forces and that a series elastic com- for movement control known as mass-spring
ponent (in the muscular connective tissue and in control. Consider a lever, pivoted near one end,
the tendons) provided elasticity. Although the with two springs attached on either side. This
concept has been known for more than a century setup is shown in figure 7.2la. Think of the
(Weber, 1846; see Partridge, 1983), what has been lever as the bone in the forearm; the pivot is the
 Principles of Speed and Accuracy II!! 251

2.5

2.0

~ 1.5
c
0
"iii
c
{!!. 1.0

0.5

150° 120° goo 60° 30°

Angle of ankle

FIGURE 7.19 Tension produced by muscle as a function of the level of activation (impulses per second) and the length of
the muscle.
Reprinted, by permission, from P.M.H. Rack and D.R. Westbury, 1969, ''The effects of length and stimulus rate on tension in the isometric cat soleus muscle;' Journal of
Physiology 204: 443-460.

elbow joint, and the two springs are the groups to the "hand" extends the elbow. First consider
of muscles that span the joint-the flexors and the curve labeled "Flexors" in figure 7.21b. Here,
the extensors. as the elbow angle is increased from 30° to 180°
In figure 7.21b are the hypothetical length- (i.e., the elbow is extended), a progressively
tension curves for these two springs, assuming increased tension is produced in the flexors
a constant level of motor activation. Consider because they are being lengthened (stretched).
what happens when some external force applied Also, the curve labeled "Extensors" represents
the tension in the extensor muscles. As the elbow
extends, the tension in the extensors decreases
because the length of the extensors decreases,
and the tensions they produce are related to
their lengths.
What happens when this force moving the
"hand" is suddenly removed? For a given amount
of stiffness in the springs, the lever would move to
c ("spring to") an equilibrium position (or equilibrium
0
"iii
c
point) in its range, finally stabilizing at a point
{!!. where it would be held by the opposing actions
of the two springs. The equilibrium point is rep-
resented by the elbow angle in which the tension
(or, more properly, torque at the elbow) from the
flexor group is equal and opposite to the tension
from the extensor group. In the diagram, the two
tensions are equal at only one elbow angle: the
Length elbow angle at which the two length-tension
diagrams cross each other, at about 95°. When
FIGURE 7.20 Idealized length-tension curves as would the system is perturbed, the mass-spring system
be produced from four springs, each with a different stiffness. will tend to move the limb back to the equilibrium
252 111 Motor Control and Learning

Flexion

30° 60° goo 120° 150° 180°

Extension Flexion Extension
a b Elbow angle (a)

FIGURE 7.21 The mass-spring or equilibrium-point model. (a) Muscles seen as springs; (b) the length-tension diagrams
for the flexors and extensors plotted for different elbow angles, with the intersection being the equilibrium point where the ten-
sions (more properly, torques) in the two muscle groups are equal and opposite.

point, regardless of the amount or direction of the activities just begin to cause the muscle to con-
original deflection. This view helps to explain tract, with tension increasing progressively as the
how limbs can be stabilized in one position, as in muscle is lengthened further. Second, the stiffness
the maintenance of posture (see also figure 5.19). of the muscle is increased as well-seen as the
steeper slope of the length-tension function for
Equilibrium-Point Principles Flexors2 as compared to Flexors1 •
(a and A Models) These shifts in the length-tension function
The mass-spring view perhaps explains some result in a change in the equilibrium point from
simple things about how we can hold a limb in 11oc to soc. But the limb is still at 11oc, and not
one place and how the limb is stabilized after in equilibrium, because the flexors are exerting a
an external perturbation. But how can it explain units of torque, whereas the extensors are exerting
movement? Closely related models have been b units; the difference in torque (a - b) begins to
proposed by Feldman (1966a, 1966b, 19S6) and move the limb into flexion. This movement con-
Polit and Bizzi (197S, 1979), which differ in the tinues until the limb is in equilibrium again, this
ways the processes are thought to work. In the time at the new equilibrium point of soc, where
A. model (Feldman), illustrated in figure 7.22, the two muscle groups are producing torques
assume that the limb is initially at a particular that are again equal and opposite (c units). The
equilibrium position (llOc) defined by the two analogous process can also occur in the extensor
length-tension functions for flexors and exten- group, or probably in both groups at the same
sors (what Feldman called the invariant charac- time in most normal movements. So, through
teristic). When the flexor group is activated, there appropriate selection of the activation to the
is a shift from one length-tension function to muscle groups spanning a joint, and hence a new
another (from Flexors1 to Flexors2 in the figure), equilibrium point, the joint can be moved to any
which occurs because of two distinctly different position within its anatomical range. The view-
changes. First, the function is shifted to the left point has been referred to as an equilibrium-point
through a change in the threshold length (which model because the limb moves to the mechanically
Feldman labeled A.). This is the muscle length defined equilibrium point.
at which the length-tension curve crosses the At this point, the Polit-Bizzi and Feldman
zero-tension axis (here shifted leftward from A.1 models become slightly different. According
to A.2); it is the muscle length at which the reflex to the a model (Polit & Bizzi, 197S, 1979), the
 Principles of Speed and Accuracy 253

80 110 180
Flexion Extension
Elbow angle (a)

FIGURE 7.22 Length-tension diagrams for the extensors and flexors for various elbow angles at times 1 and 2. (The equi-
librium point is shifted from 110° to 80° by increased activation in the flexors to produce a new length-tension relation.)

mechanical characteristics of the muscles enable on the issue (see Bizzi, Hogan, Mussa-Ivaldi, &
the limbs to move to the new equilibrium posi- Giszter, 1992; Balasubramaniam & Turvey, 2005;
tion without any feedback (see the next section Berkinblit, Feldman, & Fukson, 1986; Feldman &
for evidence). On the other hand, the A. model Latash, 2005; Feldman & Levin, 1995; Feldman,
(Feldman, 1966a, 1966b, 1986; Feldman & Levin, Ostry, Levin, Gribble, & Mitnitski, 1998; Gottlieb,
1995; Latash, 1993) holds that the feedback 1998, 2000; Gottlieb, Corcos, & Agarwal, 1989;
from the spindle activities is an essential part of Jaric & Latash, 2000; Latash, 1993). For our pur-
the process, perhaps to ensure that the muscle poses, however, the equilibrium-point models
maintains approximately constant stiffness as provide an important contrast to the impulse-
Nichols and Houk (1976) suggested. In any timing model of movement control discussed
case, no feedback to either higher centers or the earlier in this chapter. In that model, the critical
stages of information processing is needed to determiner of the limb's action and trajectory was
move accurately; errors are not detected and the amount of force programmed and the timing
corrected as in a closed-loop system. Another and duration of this force. With the equilibrium-
difference concerns what the motor system is point model, the muscle activation is simply
thought to control. In the a model, the motor changed to a new level, and no preestablished
system activates only the alpha motor neurons to timing of the onsets and offsets of the muscular
change the length-tension function; the muscle impulses is involved.
spindles and the gamma system are uninvolved. Finally, an important distinction is that, in the
In the A. model, the alpha and gamma systems equilibrium-point theory, the motor system does
are controlled together, with the muscle spindles not have to "know" where the limb is starting
being involved in the control of stiffness. from in order to move it to a new location. As
The controversy surrounding the ability of the figures 7.21 and 7.22 illustrate, the equilibrium
a and the A. models to explain certain features point can be achieved mechanically regardless of
of limb control has existed for many years and the starting position. Thus, the equilibrium-point
continues to draw diverse opinions. Numer- models are somewhat simpler than the impulse-
ous commentaries have been published, and timing view because only two levels of activation
a number of references represent a good start are specified to the muscles. With the impulse-
for the interested reader wishing to gain an timing view, the system must know where the
appreciation of the differing opinions that exist limb is at the beginning of the movement and
254 llM Motor Control and Learning

then must specify the appropriate durations and and the animal looked at it, the experimenters
intensities of the muscular impulses. would unexpectedly shift the initial limb posi-
tion. Sometimes a mass would be applied to the
Evidence for Equilibrium-Point Control
lever unexpectedly, or a brief pulse of force was
Scientists have long known that muscles act like applied that restrained or aided the movement
complicated springs," but the Russian physiolo-
11
temporarily.
gist Feldman (1966a, 1966b; Asatryan & Feldman, Typical records of these arm movements are
1965) was probably the first to describe relation- shown in figure 7.23. The top three are from the
ships between position and torque in human monkeys prior to deafferentation. An unresisted
movements, and to propose how mechanical move is shown in figure 7.23a, and a perturbation
properties might be used in movement control is applied (as indicated by the horizontal bar) to
(see footnote 5 at the end of this chapter). Feld- aid the movement in figure 7.23b and to resist the
man's work was not widely known in the West movement in figure 7.23c. The same monkeys
until it was popularized by Turvey (1977). At then performed the movements after recovery
about the same time, Polit and Bizzi's (1978, from surgical deafferentation. The unresisted
1979) work in the United States with deafferented move in figure 7.23a appears to be quite like the
monkeys independently supported similar ideas, move before the deafferentation, except that it is
and some of this evidence is given next. slower. When the perturbation was applied to
Experiments With Deafferented Mon- aid the movement (figure 7.23b) or to resist the
keys Polit and Bizzi used monkeys with deaf- movement (figure 7.23c), the movement end point
ferented forelimbs that were trained to point was achieved regardless of the direction of the
the hand and an attached lever to a target light. perturbation. Sometimes when the target location
With the hand actions, the monkeys could not, was close to the starting position, the perturbation
of course, feel the limb move, and they could not in limb position would take the hand past the
see the limb either, as the movements were made target. In these cases the limb moved "backward"
in the dark. The major dependent variable was toward the target and achieved nearly the same
the terminal location of the movement, and Polit position as when it was unresisted (see also Bizzi,
and Bizzi studied these moves when the limb was Polit, & Morasso, 1976).
perturbed prior to or during the movement. For These findings raise some interesting ques-
example, when the stimulus light was turned on tions. First, the monkeys tended to move directly

a b c

FIGURE 7.23 Elbow movements in normal (top) and deafferented (bottom) monkeys. (The end point is achieved even if a
perturbation, indicated by a horizontal bar, is applied during the movement.)
Reprinted with permission from A. Polit and E. Bizzi, 1978, "Processes controlling arm movements in monkeys;• Science 201: 1236. Reprinted with permission from AAAS.
 Principles of Speed and Accuracy 1111 255

to the target regardless of the mass that was mechanism itself would support the weight.
applied to the limb and regardless of shifts in The experimenters were interested in the con-
initial location. All this was accomplished without stant errors (CEs) in achieving the target on the
feedback from limb position to the spinal cord. "normal trials" (with the expected weight) and
The hypothesis that the monkey felt" the change
/1
on the "switch trials," for which the weight was
in position, or that reflex processes were involved, either added or subtracted unexpectedly.
cannot explain these findings. From table 7.3 (top), when the mass was sud-
The results also tend to argue against an denly increased, the movement end point (the
impulse-timing hypothesis. According to this CE, or constant error) was nearly unaffected. The
view, the monkey first determines where the same was true in the mass-subtracted portion
limb is, and the movement is then controlled of the experiment. However, the MTs shifted
by the GMP to produce the required impulses. considerably, being far longer when the mass
First, if the initial position is shifted, under the was suddenly added and far shorter when the
impulse-timing view the animal would have to mass was suddenly subtracted. These results are
take this into account, because the program (if consistent with the equilibrium-point view, as
unaltered) would cause under- or overshoot- the movements arrived at the target even when
ing of the target. Yet, the deafferented animal the inertial characteristics of the lever were unex-
could not detect this shift, and moved directly pectedly changed, with only the rate of approach
to the target. Second, when the monkey's limb to the target position being affected by the
was shifted past the target position before the load.
movement, the limb moved "backward" toward A second experiment had the lever movements
the target. The impulse-timing hypothesis has in the vertical plane rather than in the horizontal
the limb moved by a contraction of the agonist plane. The equilibrium-point model, in this case,
first, then of the antagonist, with this order of predicts that the movement end point should now
contraction being specified by the motor pro- be affected by the changed weight since the added
gram. If so, then the initial movement should weight will bias the equilibrium point downward
have been away from the target, not "backward" because of gravity; a weight subtracted unex-
toward it. pectedly will tend to shift the equilibrium point
Experiments With Humans The studies just upward. Thus, the prediction is that the limb will
described generated skepticism for a number of undershoot the target when the weight is added
reasons. First, it is not certain that these processes and overshoot it when the weight is subtracted,
found in animals also operate in humans. Second, which is different from the prediction of no
it is never perfectly clear that deafferentation change in the horizontal-movement experiment
keeps all sensory movement information from where gravity is not involved.
reaching the brain, as bone-conducted vibrations When the weight was added, from table 7.3
from the movement can be sensed by parts of we see that the movement in the switch trials
the body that are not deafferented. Finally, and was about 5° shorter than in the normal trials.
perhaps most importantly, the monkeys may When the weight was subtracted, the limb move-
have adopted an equilibrium-point mode of ment was approximately 8° longer than in the
control simply because Polit and Bizzi deprived normal trials. Large shifts in MT also occurred,
them of all their usual movement mechanisms. with the added weight slowing the movement
Do these experiments have relevance for human and the subtracted weight speeding it. The
movement control? Probably, as seen in various prediction of an impulse-timing view would be
experiments with humans. that the limb would come to a stop in the cor-
In experiments with humans, Schmidt (1980; rect time. Interestingly, a reflexive closed-loop
Schmidt & McGown, 1980) had subjects produce model would predict that the movement should
rapid elbow flexion movements of a lever to a achieve its desired end point here, because the
target. Occasionally, the load on the lever was limb system would simply move to the position
unpredictably changed before the movement, that it "recognizes" to be correct, and added
and the subject moved with the changed load weight should have no effect on the terminal
conditions. In the first of these experiments, the location of the limb-so closed-loop models are
lever movement was horizontal, and the lever contradicted as well.
256 Motor Control and Learning

Mass added CE +6.36° +6.81°

MT 187 ms 278 ms
Mass subtracted CE +5.78° +6.28°
MT 214 ms 180 ms

Mass added CE +15.82° +10.40°

MT 202 ms 243 ms
Mass subtracted CE +7.83° +15.79°
MT 196 ms 155 ms
From Schmidt and McGown 1980.

Extensions of the Equilibrium-Point Models separate equilibrium points is specified, so the

Berkinblit and colleagues (1986) proposed a model tends to be a compromise between the
very interesting model for sequential actions on equilibrium-point models and the motor program
the basis of the wiping reflex in the spinal frog models (see Feldman & Levin, 1995; also Rosen-
(figure 6.8). They argue that this action is really a baum, Loukopoulos, Meulenbroek, Vaughan, &
series of approximately seven discrete positions. Engelbrecht, 1995, for a similar application).
The action is achieved by specifying a sequence Another application of the equilibrium-point
of equilibrium positions, and each of these posi- model was described by Hollerbach (1978, 1981),
tions is achieved exactly as in the equilibrium- who attempted to simulate handwriting through
point model. This model requires something like the use of various mechanical principles. He con-
a central program to control when each of the ceptualized the muscles that move the fingers as

Forward

l<
~<
~

\!\!\!\!\!\!\!\!\1 Right

<
>
>
s>>
t
Backward

FIGURE 7.24 A mass-spring model for handwriting. (The forward-backward and right-left spring pairs oscillate to produce
writing-like patterns of a pencil attached to the mass.)
Reprinted, by permission, from J.M. Hollerbach, 1978, A study of human motor control through analysis and synthesis of handwriting. Unpublished doctoral dissertation
(Cambridge, MA: Massachusetts Institute of Technology).
 Principles of Speed and Accuracy 111 257

springs that move the finger up (toward the top sal position) when the action requires a reversal
of the page placed horizontally), down, left, and that must be timed. The impulse-timing model
right, as shown in figure 7.24. One important point explains these results easily: The motor program
about the models of Rosenbaum and colleagues "told" the agonist when to tum off, and the move-
and Hollerbach is that treating the muscles as ment with added weight could not go as far in this
"complicated springs" allows an explanation of fixed amount of time, so the movement's reversal
how a trajectory is produced with a minimum of fell short of its goal.
complexity in the motor programming. It is fairly well accepted that the equilibrium-
point model is the best account of how a joint
Limitations of the Equilibrium-Point Models
achieves its terminal position. But there is con-
Most of the support for the equilibrium-point siderable doubt that this model can account for
model comes from only one kind of movement- the events that occur at the very beginning of
simple, unidirectional positioning tasks (but see the movement. Other evidence suggests that the
Feldman & Levin, 2009, for recent theoretical devel- initial trajectories are generated by processes
opments). To what extent is this kind of mechanism somewhat more complicated than this (Atkeson
involved in movements '.hat use more than one & Hollerbach, 1985; Hollerbach & Flash, 1982).
direction of a limb or in movements that involve Also, equilibrium-point models cannot account
more than one limb and that must be coordinated? for the results of Wadman, Denier van der Gon,
Schmidt and McGown (cited in Schmidt, 1980) Geuze, and Mol (1979), in whose experiment
and Schmidt, McGown, Quinn, and Hawkins the unexpected blocking of the limb resulted in
(1986) investigated this problem with various an unaffected agonist-antagonist-agonist EMG
kinds of movements. One study used the same pattern (see figure 6.9). It also cannot account for
kind of apparatus and experimental design as just Polit and Bizzi's (1979) experiments in which the
described (table 7.3) for the single-direction moves, monkey pointed at an unseen target by moving
but the horizontal movements now involved a the elbow joint: If the shoulder joint was moved
reversal in direction. The subject moved in flex- before the action, the overall pointing direction
ion, reversed the move, and extended the elbow was grossly incorrect. Overall, neither th~ equilib-
past the starting position so that the time from the rium-point model nor the impulse-timing model
beginning until the time the starting point was is capable of accounting for all of the findings.
again reached was 300 ms. Mass was unexpect-
edly added or subtracted from the lever exactly
as before. Of interest was where and when the Correction Models
movement reversal occurred under the normal of the Speed-Accuracy
and switch conditions. Was the reversal point
determined by equilibrium-point control, so that Trade-Off
it would be unbiased in location with added load?
The major results are shown in table 7.4. The To this point in our discussions about the theoreti-
reversal point was shorter when the mass was cal basis for rapidly aimed movements, we have
added and longer when the mass was subtracted ignored the problem of how errors are corrected.
unexpectedly. Thus, the equilibrium-point model However, from previous discussions we know
does not seem to account for the results (on rever- that errors (variability) in centrally generated

Adapted, by permission, from R.A. Schmidt, 1980, "Past and future issues in motor programming:' Research Quarterly for Exercise and Sport 5: 122-140.
258 Motor Control and Learning

signals are bound to occur. The motor system is The model, illustrated in figure 7.25, is based on
inherently "noisy," leading to variability in the the idea that a ballistic, distance-covering phase
generation of impulses or joint positions. Thus, would operate for a fixed period of time, moving
theories describing how these error corrections the limb a certain amplitude toward the target.
occur have been important for a more complete This initial phase would have a spatial inaccuracy
understanding of the speed-accuracy trade-off. proportional to the distance that had been moved
during that time. Then, feedback processes would
The Crossman-Goodeve Model evaluate the size and direction of the error and
At the 1963 meeting of the Experimental Psychol- initiate a second ballistic movement that would
ogy Society in England, Crossman and Goodeve serve as a correction. This second movement
presented a feedback-based explanation of Fitts' would also have an error proportional to its
law. They suggested that Fitts' law could be (much shorter) distance; its error would then
derived mathematically, on the basis of feedback be evaluated, another correction would be initi-
control in movement (with a number of assump- ated, and so on until the movement reached the
tions), without the need to resort to ideas about target. Thus, the model is based on rapid alterna-
information theory as Fitts (1954) had done. This tion between ballistic movement processes and
derivation with its associated argument was feedback-based corrective processes during the
described in a more accessible form by Keele action. The MT, which is the dependent variable
(1968) and was reprinted in its entirety in 1983. in the Fitts equation, was thought to be based on
Crossman and Goodeve (1963 /1983) assumed the number of these corrective processes that had
that movement toward a target is made up of two to be made to achieve the target.
kinds of processes-much as Woodworth (1899) Keele (1968) used Crossman and Goodeve's
had proposed 50 years earlier-except that the basic idea and added to it the assumptions that
ballistic, distance-covering phase and the feed- the time between corrections was fixed at 190 ms
back-controlled "homing-in" phase were thought (Keele & Posner, 1968; see chapter 5) and that the
to operate in rapid alternation during a movement error in each movement was about one-seventh of
to the target. This mode of control is termed the total distance moved in each correction. Keele
intermittent (or iterative); hence, their idea has argued that each correction was processed in the
often been termed the iterative-correction model. stages of information processing, requiring atten-
tion, and that aimed movements were made up of
a series of such corrections leading to the target.
With use of these assumptions, the Crossman-
Goodeve model showed a good quantitative fit
to some of the experimental data (Keele, 1981).
One major drawback to the Crossman-Goodeve
model of rapid aiming movements is related to
the speed with which humans can process visual
feedback. This was a major theme of chapter 5,
and it is one of the most fundamental reasons for
believing that very rapid movements must be
planned in advance. Another problem relates to
the psychological refractory period: Even though
it might be possible to make one such correction in
190 ms, it is doubtful that the second and third cor-
rections could also be made this quickly (review
discussion of psychological refractory period
A
effects, chapter 4). These and other criticisms of
Amplitude the Crossman-Goodeve theory have been dis-
cussed in more detail in several reviews (Schmidt
FIGURE 7.25 Crossman and Goodeve's (1963/1983) et al., 1978, 1979; Meyer et al., 1982, 1988).
iterative-correction model of aiming. Perhaps the most persuasive argument against
Adapted, by permission, from D.E. Meyer et al., 1988, "Optimality in human motor
performance: Ideal control of rapid aimed movements,'' Psychological Review
the Crossman-Goodeve theory was based on
95: 343. Copyright© 1988 by the American Psychological Association. kinematic records of subjects' movement trajec-
 Principles of Speed and Accuracy 111 259

tories (Langolf et al., 1976). Transitions between

one ballistic segment and the next could be seen
as sudden changes in the position or velocity
of the limb. Generally, most of the movements
studied had one correction (a very few had two),
although some had no visible corrections at all,
even with MTs of 700 ms. These findings failed
to support the hypothesis that a correction occurs
every 190 ms.
A side issue here is that Fitts' law, and the
speed-accuracy trade-off in general, had usu-
ally relied on closed-loop control processes for
explanation (e.g., the Crossman-Goodeve view).
However, note that one contribution of the
impulse-variability theory (Schmidt et al., 1979)
A
is that it provided evidence of speed-accuracy
trade-off even when the movements were far too Amplitude
short in time to be explained by feedback-based
processes. FIGURE 7.26 The optimized-submovement correction
model.
Optimized-Submovement Models Adapted, by permission, from D.E. Meyer et al., 1988, "Optimality in human motor
performance: Ideal control of rapid aimed movements," Psychological Review
95: 343. Copyright© 1988 by the American Psychological Association.
The failure of the Crossman-Goodeve theory to
handle these and other data was addressed by
Meyer and colleagues (1988, 1990) in what they shown as the dashed lines. Total MT in ail aiming
termed the optimized-submovement model. Meyer task is considered to reflect a strategy whereby a
and colleagues began with the idea that the initial subject attempts to trade off speed for accuracy by
segment of an aiming movement was handled optimizing the duration of both the initial impulse
by principles of impulse variability as described and, if necessary, the correction. Thus, move-
earlier in this chapter. But then they went on to ment accuracy is achieved in a minimum MT by
describe how corrections can be applied to the optimizing the control of the submovement(s)-a
trajectory after this initial movement to allow nice combination of some of the important open-
the limb to move accurately to a target. In effect, loop and closed-loop processes discussed in this
each correction is a new program whose accuracy chapter.
is governed by impulse-variability principles.
This concept is similar to the hypothesis that
Multiple-Submovement Model
Woodworth proposed in 1899, but it is made more In general, the dual-submovement model fits
concrete and experimentally testable by a number the experimental data quite well (Meyer et al.,
of additional assumptions concerning how many 1988). However, Meyer and colleagues also found
such corrections should be expected, and where that corrections occurred more frequently than
and under what conditions they should be found predicted under some combinations of target
in a movement as a function of various movement amplitude and width. In a revised version of
variables. the model, Meyer and colleagues (1990) made
additional assumptions to account for the fact
Dual-Submovement Model that more than one corrective impulse could
In most aiming tasks, the processes involved in be made during an aimed movement. This
bringing the limb to the target can be described multiple-submovement model was supported by
by two situations that are illustrated in figure 7.26. a reanalysis of data from Meyer and colleagues
The first situation (middle curve in figure 7.26) (1988) using a different kinematic analysis pro-
occurs when the initial action (termed the primary tocol (see also Jagacinski & Flach, 2003, chapter
submovement) requires no correction. The other 8). This reanalysis confirmed that about 80% of
situation occurs when the initial impulse either all aimed movements were composed of two or
undershoots or overshoots the target, requiring three submovements (i.e., one or two corrections;
a corrective impulse (or secondary submovement), 43% and 37%, respectively).
260 Motor Control and Learning

How Is a Correction Identified whereby movement-timing consistency is also

and Measured? increased as velocity is increased. These two
Although Meyer and colleagues' (1988, 1990) effects combined imply that increasing the speed
models provide a balanced, hybrid approach to the of movements in timing tasks leads to increased
issue of open- and closed-loop processes, experi- temporal consistency, contrary to ideas about
mental examination of their predictions places the spatial speed-accuracy trade-off. Timing
a critical emphasis on the quantitative analysis consistency has also been examined in repetitive
of corrective submovements. As we have seen movement tasks. A model of timing in which
before, motor output processes are rather noisy, variability is due to independent, central, and
and often it is not clear how evaluations of kine- peripheral sources has been a useful research tool.
matic data (e.g., position, velocity, acceleration, Many aiming movements are influenced by
or jerk) can discriminate between corrections and central (programmed) mechanisms, closed-loop,
motor noise. Various techniques have been used corrective actions (both reflexive and intentional),
to objectify kinematic data, but it is beyond the or both. Impulse-variability principles describe
scope of the present discussion to describe these how programmed impulses affect the trajectory
analytical techniques (for theoretical discussions of a rapid aimed limb movement. A different view
underlying these techniques, see Khan et al., 2006, is suggested by the equilibrium-point model, which
for a recent review and the following references: holds that the limb moves to a position defined by
Carlton, 1979; Chua & Elliott, 1993; Langolf et al., an equilibrium point between the forces (torques)
1976; Meyer et al., 1988; Pratt & Abrams, 1996). from the opposing muscles spanning a joint, and
that the movement to this position is dependent
on the mechanical, spring-like characteristics of
Summary the muscles. Such a model is different from an
impulse-timing mechanism whereby the amounts
Three types of speed-accuracy trade-offs were of force, as well as the times over which they are
described in this chapter. The earliest formal applied, are controlled by the motor program.
mathematical statement of this relationship was Early explanations for Fitts' law were based
proposed over 50 years ago by Paul Fitts, called on intermittent-control models, in which the com-
Fitts' law in his honor. It says that, in reciprocal mands for action were produced alternately with
movements, the average MT is linearly related to the analysis of feedback to determine movement
the Log2 of the ratio of the movement amplitude corrections. Several lines of evidence, however,
and the target width. This principle holds for a suggest that the intermittent-control model is
wide variety of movement situations and types of incorrect. More acceptable views indicate that
subjects and is considered one of the cornerstones the timing of an open-loop, distance-covering
of motor control. phase and later correction(s) is optimized in some
Force and force variability are strongly related way in order to idealize the inconsistency in the
to each other, with the relationship being essen- distance-covering phase as the forces applied to
tially linear until forces of about 65% of the the limb are increased.
subject's maximum are achieved. The force vari-
ability levels off or even decreases with further
increases in force toward the subject's maximum.
Student Assignments
The laws about MT variability and force vari-
ability have been combined into various impulse- 1. Prepare to answer the following questions
variability models for movement accuracy: The during class discussion:
prediction is that the variability of a movement's a. Calculate the Fitts index of difficulty for
initial impulse is linearly related to the ratio of the the following sport tasks: throwing a
movement amplitude (A) and the MT (A/MT), strike in baseball, scoring a basket from
or average velocity. the free-throw line, successfully scoring
An analysis of laws of movement timing shows a field goal from 35 yards in Ameri-
that, as the MT is decreased, an increase occurs can football, scoring a penalty kick
in movement-timing consistency. This occurs in soccer (assuming no goalkeeper),
in addition to the effect of movement velocity, holing a putt in golf from 30 feet, and
 Principles of Speed and Accuracy I'll 261

scoring a bull's-eye in darts. (As appro- ers & McCormick, 1993). However, given that most laws
priate, use regulation distances and are "awarded" posthumously, there is some reluctance
target sizes to calculate your answers.) in accepting such a tribute at this time.
2 The observant reader will notice that the measures
b. Discuss the open-loop and closed-loop
of variability in figure 7.12 are an order of magni-
processes involved in using a computer tude higher than those seen in figure 7.9. This occurs
mouse to open a desktop icon. because variance is the dependent measure in figure
c. Describe the playing of a musical instru- 7.12, whereas standard deviation is presented in figure
ment using the concepts of spatial and 7.9. Since the variance is equal to the squared value
temporal speed-accuracy trade-offs. of the standard deviation, the results presented in
figure 7.12 agree remarkably well with the values in
2. Find a research article that investigates or figure 7.9.
discusses the role of Fitts' law in ergonom- 3 The analysis of impulses described here is simplified
ics and human factors. considerably in order to give an overview of impulse-vari-
ability modeling. Space here does not permit a discussion
of the numerous assumptions underlying the analyses of
Web Resources impulse variability. For a more complete treatment, see
Schmidt and colleagues (1978, 1979, 1985).
More on Fitts' law: 4 These effects are not precisely as predicted by the

www.asktog.com/columns/022DesignedTo model, however. For example, changing A does change

GiveFitts.html VE1 slightly, and changing MT does change W, slightly,
in both cases probably because of the effects of velocity
http://msdn.microsoft.com/en-us/library/ discussed earlier (figure 7.8; Newell, Carlton, & Kim, 1994;
ms993291 Newell et al., 1979, 1980).
More on Weber's law: 5 Feldman (2009) and Latash (2008a) provide his-

torical reflections on the origins of the Asatryan and

www.cis.rit.edu/people/faculty/montag/ Feldman (1965) formulation. Related (though gener-
vandplite/pages/chap_3/ch3p 1.html ally unacknowledged) views were presented earlier by
Wachholder and Altenburger (1927; see Stemad, 2002)
and by Crossman and Goodeve (1963/1983). Perhaps
Notes these were generally unknown because Wachholder
and Altenburger wrote in German and the Crossman-
1 R.A.S. is very appreciative of the label "Schmidt's Goodeve model was presented at a conference, and not
Law," as some have called it (e.g., Flach, Guisinger, & widely circulated. These ideas were not readily available
Robison, 1996; Jagacinski & Flach, 2003; Keele, 1986; Sand- until many years later.
I n some instances, we seem to be able to produce
coordinated actions easily, almost trivially. We
engage in countless activities in which our limbs
or unimportant (rubbing and patting the head
and stomach).
We have recognized for several decades now
perform different actions at the same time, seem- that a major source of interference in interlimb
ingly without any interference at all (e.g., using a coordination is related to the temporal structure(s)
knife and fork; playing piano; walking and chew- of the actions being coordinated (for reviews,
ing gum). In tasks such as throwing and kicking, see Heuer, 1996; Klapp, 1979, 1981). Actions
our upper and lower limbs perform very different with the same temporal organization are easily
functions and movements, all without difficulty. coordinated, with a tight temporal relationship
Yet in other instances, there is substantial between the limbs, whereas activities with dif-
interference between the effectors when we try ferent temporal organizations are not easily
to perform some actions. Well-known cases of produced, if they can be produced at all. There is
interference include the task of patting your head clearly more to coordination than this, however,
while rubbing your stomach. Both hands seem to as will be seen in the variety of tasks and means
"want" to do the same thing, either to rub or to by which researchers have attempted to study
pat, but not to do the different actions as required. the fundamental problems of coordination. One
Another example comes from Peters (1977), who important factor in coordination is the duration
found that not a single subject, out of 100 who of the movements being controlled. As we have
were tested, could recite a nursery rhyme with noted in many of the chapters presented so far,
proper timing while concurrently tapping a dif- the control processes in discrete skills seem to be
ferent rhythm. Summers, Todd, and Kim (1993) very different (in terms of fundamental activities
provide further evidence about difficulties in tap- involved) from those underlying continuous
ping different rhythms with the two hands (pro- skills. This, among other things, has created a vig-
ducing so-called polyrhythms, discussed in more orous (usually healthy) debate between a group
detail later in this chapter). These and many other of scientists who think of movement control in
observations suggest the existence of interlimb terms of self-organizing systems versus a group
coordination processes that facilitate biologically that thinks in terms of motor program concepts.
important activities such as locomotion, but that Indeed, Keele (1998; see "Steve Keele" on p. 13)
tend to impede more arbitrary skills, which may held the view that this gulf between competing
be culturally important (piano playing, throwing) scientific "camps" "is due less to competing con-

Ill 263
264 Motor Control and Learning

ceptualizations of the same phenomena than to the Discrete Tasks

kinds of phenomena with which different groups
of investigators are concerned" (p. 403, emphases
added). Keele was referring to the tendency of The focus of discussion in chapter 7 was on the
(a) the self-organizing systems group to study principles by which limb movements, primarily
continuous, often rhythmical, skills of (usually) aiming movements, achieved a target with mini-
long duration, where processes involving feed- mum time and maximum accuracy. Although the
back and adjustment for environmental changes motor skills that led to the generation of these
can be studied most easily; and (b) the more principles involved relatively "simple" move-
cognitively based motor program group to study ments (such as repetitive tapping), in reality these
discrete skills of (usually) short duration, where movements are far from simple. The following
preplanning seems to be critically important and sections also describe relatively "simple" move-
feedback-based adjustments do not. ments, but here the focus instead is on the kinds
In chapter 2, we distinguished between of organization that must underlie the simulta-
discrete and continuous tasks in terms of their neous control of the different effectors-that is,
beginning and end points. Discrete tasks, such the coordination of the various parts of the body.
as turning on a light switch or swinging a golf
club, have definitive start and end points. In
Eye-Head-Hand Coordination
contrast, the start and end of continuous tasks, In chapters 5 and 7 we presented evidence that
such as running and driving, are rather arbitrary many tasks involving aiming, including reaching
or context dependent. Discrete tasks are usually and grasping actions, rely on visual feedback in
(but not always) performed quite rapidly, and order to maximize movement end-point accuracy.
considerable theoretical importance is placed on Visual feedback information is most precise for this
the motor program in preplanning and organiz- kind of task when the eyes can fixate on the target
ing the body's movements (see chapter 6). In or object for a significant length of time prior to
contrast, the role of premovement planning is manual contact (e.g., Abrams, Meyer, & Kornblum,
less important for continuous movements; and 1990). When a target occurs unpredictably, the
other factors such as feedback, error detection, localization of the target or object to be contacted
and error correction take on more important must be fixated as rapidly as possible in order to
roles in the ongoing regulation of control, chiefly provide sufficient time for processing the visual
because of the additional time allowed for them information. How is coordination among the eyes,
to occur. Thus, in keeping with the view of motor head, and hand achieved in this situation?
control-where discrete and continuous skills Early work with monkeys revealed the exis-
are performed fundamentally differently-we tence of a tight relationship between the head and
have organized this chapter by considering the eyes (Bizzi, Kalil, & Tagliasco, 1971). Figure 8.1
coordination of discrete and continuous tasks in illustrates a typical trial in which a monkey turns
separate sections. to look at an unexpected target that has appeared
Thus, the understanding of coordination seems in the periphery of the visual field. From a starting
to represent different challenges for discrete position in which the monkey's eyes and head
and continuous tasks. However, keep in mind are pointed straight ahead, the tracings in figure
that the problems encountered in coordinating 8.la reveal that, after a brief reaction-time (RT)
movements in discrete and continuous tasks period, the eyes initially make a rapid saccade to
share many similarities. The most prominent of fixate the target on the fovea. Movements of the
these is the degrees offreedom problem (Bernstein, head are initiated at the same time as those of
1967): Given that there are more independently the eyes but are much slower, as can be seen in
moving parts of the body than can be controlled figure 8. lb. As the head moves in the direction of
individually, how is the system able to perform the target, the eyes rotate in a direction opposite
actions that coordinate these parts so effectively? to the movement of the head. The timing of the
The study of coordination in both discrete and opposing motions of the eyes and head is tightly
continuous tasks points to some fundamental, yet coordinated, so that the fovea remains fixated on
distinct, principles regarding how the degrees of the target throughout the head movement. The
freedom problem is solved. existence of this coordination between the eyes
 Coordination 1m 265

At movement of the head or a decrement due to the

target
I immobilization of the head.
I
I The timing of limb movements is also well
coordinated with eye movements during the
c termination of movement toward a target. Using
Cll
E
Cll an eye-tracking system (see chapter 2), Helsen,
>-Cll
w > Starkes, and Buekers (1997) found, perhaps not
0
E too surprisingly, that an eye saccade always
a
reached and fixated on or near the target before
the hand reached the target. Of some surprise,
though, was the finding of an invariant relative
timing of the saccade and the manual response:
c
'tJ Cll Regardless of the amplitude of the movement
111 E
Cll ~ or the absolute movement time taken to reach
::c 0
E the target, the saccade brought the eye onto
b Time
the target at 50% of the total movement time
(see also Starkes, Helsen, & Elliott, 2002). Recall
FIGURE 8.1 Rotation of the (a) eyes and (b) head from chapter 6 that evidence of invariant relative
during rapid looking at a target.
timing supported the arguments for a generalized
Reprinted with permission from E. Bizzi, R.E. Kalil, and V. Tagliasco, 1971, "Eye-head
coordination in monkeys: evidence for centrally patterned organization," Science motor program (GMP). A similar argument could
173(3995): 452-454. Reprinted with permission of AAAS. http://www.aaas.org/
explain Helsen and colleagues' findings, such that
a GMP coordinates the relative timing of eye and
and head has been known for a long time and is hand movement. In the next section, we show
called the vestibulo-ocular reflex (VOR). In the case how GMP theory has been developed to explain
of rapid looking, the saccade facilitates a quick other types of coordinated discrete actions.
identification of the unexpected visual signal. The
counterrotation of the eyes maintains the fixation Units of Action
of the target while the head is turning. Discrete skills-especially rapid ones-seem to be
A similar relationship among the eyes, head, controlled by a pattern of neuromuscular activity
and hand appears to exist when an intentional that is largely organized in advance. Of course,
manual response is made to a stimulus. Biguer, this is the essential feature of the theory of motor
Jeannerod, and Prablanc (1982) found that the programs described in chapter 6. While many
onset of eye movements occurs almost simultane- have debated the relative merits of central versus
ously with initiation activity of electromyographic peripheral contributions to motor control, it seems
(EMG) signals in the arm and neck. This temporal incontrovertible that at least some fundamental
coordination among the initiation of eye, head, features of the action are specified in advance.
and limb movements is quite flexible, however, as The principles of speed and accuracy discussed
performance differences arise when instructions in chapter 7 involved actions in which the primary
do not constrain movement strategies (Abrams goal was to move an effector from one location
et al., 1990; Stahl, 2001) or when instructions to another. Pushing the keys on a telephone and
emphasize either movement speed or accuracy using a keyboard are examples of this type of
(Carnahan, 1992; Carnahan & Marteniuk, 1991). action. However, there are other, related actions
In addition to the temporal relationship among that have a more complex organization, involv-
eye, head, and limb movements, there is also a ing serial or parallel (or both serial and parallel)
spatial facilitation when these degrees of freedom coordination of different muscles and limbs. For
interact. Manual aiming is more accurate when example, a computer mouse enables the user to
the head is free to move than when head position move a cursor to a target, at which time a button
is fixed (Biguer, Prablanc, & Jeannerod, 1984; click (or double click) initiates the icon's function.
Vercher, Magenes, Prablanc, & Gauthier, 1994). It Here, one action must be completed before another
remains unclear, however, whether the improved action can be performed. If we consider an action
accuracy is due to facilitation from the allowed that is somewhat longer in duration and greater in
266 Motor Control and Learning

complexity, however, such as serving a tennis ball correlate with any of them. One interpretation
or shifting gears in a car, it is difficult to claim that would be that the first set of landmarks was
such longer sequences are controlled completely governed by a GMP (because relative timing was
by single programs. Nevertheless, it is possible approximately invariant there) but that the later
that such actions are composed of a string of set of landmarks was not (because the invariance
programs (each lasting part of a second or more). was no longer present). It could be the other way
Each of these programs might be thought of around too, with the group of highly related land-
as a unit of action-a "piece" of motor behavior marks being located at the end of the movement,
that can be utilized repeatedly in various actions, perhaps with all these landmarks being unrelated
producing essentially the same movements to a landmark at the start.
(but scaled to the environment) each time. For This provides the essential idea for the identi-
example, we might think of shifting gears in a fication of units. A unit is a sequence of behavior
car from second to third gear as being composed with essentially invariant relative timing-that
of three units: (1) gas up I clutch down; (2) shift is, it has high correlations among the times of
lever up-over-up; and (3) gas down/ clutch up. component landmarks. When later landmarks
(A racecar driver such as R.A.S. might perform of the action no longer share this invariance
this task as one unit.) And a given unit could be (i.e., they do not correlate with landmarks in the
observed in other actions; a clutch-down/ gas-up earlier portions), this indicates that the first unit
unit could be involved in both a gear change from has ended, or that there is some sort of boundary
first to second and a gear change from second to between it and the next unit.
third, while another unit (shift-lever movements)
Methods in Unit Identification
would be different in the two actions. In the next
section, we present one possible way to identify Schneider and Schmidt (1995; also Young &
such units of behavior, based on GMP concepts. Schmidt, 1990, 1991) asked subjects to learn a
coincident-timing task in which the timed motion
Identifying Units of Action of a right-hand-held lever was to coincide with
The underlying idea for identifying units of action the arrival of a virtual moving object (as in base-
is based on the notion of GMPs, discussed in ball or tennis). After the virtual object began
chapter 6 and in the previous section. To review moving, the subject made a small preparatory
(see also chapter 6), the GMP is a program with movement to the right, followed by a backswing
invariant (a) sequencing among muscles, (b) to the left, and finished with a striking movement
relative timing, and (c) relative forces among the back to the right (a kind of "backhand" action,
contractions. Superficial features such as overall with a follow-through), with the goal of "hitting"
movement duration and movement size are the virtual object with the lever during the second
controlled by parameters that scale (linearly, in rightward movement.
this simple view) the surface expression of the Kinematic analyses of each trial were used to
GMP's output, yet allow it to retain its invariant determine the timing of eight "landmarks" within
structure or pattern. each trial. These kinematic landmarks (shown as
An invariance in relative timing means that points A, B, C, ... , H in figure 8.2) were defined
the correlations (within subject, across trials) as maxima, minima, or zero crossings taken from
among the times of various temporal events (or the position-, velocity-, and acceleration-time
"landmarks") in an action should approach 1.0. records (see chapter 2 on kinematic methods).
Now, to turn the argument around, if an invari- Within-subject correlations were computed over
ance in relative timing happens to be found for a trials for all of the possible pairs of landmark times.
sequence of behavior (i.e., the correlations among The critical comparisons involved both forward
landmarks happen to be very close to 1.0), such and backward methods of computing the intercor-
evidence would be consistent with the hypothesis relations. The forward computation involved the
that a GMP produced that behavior. correlation of the first landmark with the second,
Now, consider a longer movement. Suppose first with the third, and so on (A-B, A-C, A-D, ...
that temporal occurrences of the first several , A-H). The backward method examined the cor-
landmarks in the movement intercorrelate highly, relations between the last landmark (H) and all
but that the time of a later landmark does not possible earlier landmarks (A-H, B-H, C-H, etc.).
 Coordination 111 267

150
cUI
0 CD
E! 110
Ill Cl
0 CD
D..~ 70
30

300

.......
>--!!! 0
13 xi
0 !
Qi g>
>~
-300

1000 2000
Time after target initiation (ms)

FIGURE 8.2 Kinematic landmarks (A, B, C, ... , H) taken from the position, velocity, and acceleration profiles of an action
are used to search for specific units of action.
Reprinted, by permission, from D.M. Schneider and R.A. Schmidt, 1995, "Units of action in motor control: Role of response complexity and target speed;' Human Performance 8:
27-39.

If the entire action is governed by a single traces in figure 8.4, where these traces for six
unit, then all of the correlations (using both the movements are overlaid. Notice that the traces
forward and backward methods) should be are all of the same general form (but adjusted
relatively high, as all intervals would be scaled with respect to each other in time) for the first
proportionally in time. The correlations are portions of the action (A-D, up to about 1,000
shown in figure 8.3. On the left, the correlations ms); the same is true for those in the last part of
from the forward method are high among the the action (F-H). However, the traces lose their
landmark pairs A-B, A-C, and A-D but are low common form at about 1,000 to 1,400 ms, with
for the later combinations with A. Using the the traces crossing over each other. This period
backward method, the correlations illustrated at of time, according to the analysis, is the transi-
the right side of figure 8.3 show that landmark tion between units. At this time, the subject has
His highly related to landmarks F and G, but is completed the first action (the initial backswing)
not related to any of the earlier landmarks. These and is preparing to time the initiation of the for-
data suggest that this action has two identifiable ward swing to strike the moving object. Notice
units of action, one containing landmarks A-D, also that the first unit (A through D) actually
and the other containing landmarks F-H, with contains two preparatory movements (right,
a boundary between them (between landmarks then left; see figure 8.4), indicating that this one
DandF). unit organizes at least two submovements and
The functioning of these units can perhaps involves a change in direction. These general
be seen more easily in the acceleration-time methods have been used in several real-world
268 Motor Control and Learning

1.0

0.8

5 0.6
~
~
0
0 0.4

0.2

A-B A-C A-D A-E A-F A-G A-H A-H B-H C-H D-H E-H F-H G-H
Landmark pair correlated

FIGURE 8.3 Correlations of the time of the first landmark (A) with the times of the later landmarks (left side); correlations
of the time of the last landmark (H) with the times of the earlier landmarks (right side).
Reprinted. by permission. from D.M. Schneider and R.A. Schmidt, 1995, "Units of action in motor control: Role of response complexity and target speed,'' Human Performance 8: 34.

4000

-4000
1000 2000
Time after target initiation (ms)

FIGURE 8.4 Visualizing units of action; unit 1 (ABCD) and unit 2 (FGH) are separated by nonsystematic crossovers after
point D.
Reprinted, by permission, from D.M. Schneider and R.A. Schmidt, 1995, "Units of action in motor control: Role of response complexity and target speed,'' Human Performance 8: 34.

tasks; for example, researchers in one study Reaching and Grasping

analyzed the unit structure involved in light-
ing a butane cigarette lighter, identifying three Another type of unimanual coordination has been
sequential elements (Schmidt, Wood, Young, & studied in situations when limb transport ends
Kelkar, 1996). in grasping an object. The composite photograph
 Coordination 1:1 269

FIGURE 8.5 Composite photograph of the reach/grasp action.

Reprinted from M.A. Goodale and P. Servos, 1996, Visual control of prehension. In Advances in motor control and learning, edited by H.N. Zelaznik (Champaign, IL: Human
Kinetics), 8Z By permission of M.A. Goodale.

in figure 8.5 illustrates how the reach and grasp Temporal Planning in Reaching
are coordinated in time and space. The functional and Grasping
aperture (separation of the thumb and fingers) of Jeannerod (1981, 1984) extended the concepts that
the grasping hand opens during the reach toward began with Woodworth (1899) to theorize about
the object. Although the hand could wait and open the coordination of reaching and grasping. Recall
once it reached the target, this would likely be from chapter 5 that Woodworth proposed a two-
an inefficient strategy in terms of economy of component model of manual aiming. A ballistic
effort and time. Typically, the hand reaches its
component brought the pencil near the target, and
maximum aperture well before object contact,
a closed-loop process provided fine adjustments
at a peak size that is larger than the width of the
to guide the pencil onto the target. According to
target, and then the aperture is fine-tuned to the
Jeannerod, reaching and grasping are organized
object's size just prior to contact.
into the same fundamental components: a trans-
Does this example reflect two separate and
independent actions of motor control, as implied port component and a grip-formation component.
by the phrase "reaching and (then) grasping," The composite photograph of reaching to grasp
or is the entire action controlled as one motor an object in figure 8.5 appears to illustrate that
program, which seems to be implied by the coor- the transport component moves the hand into
dinated relationship between the limb and hand the general vicinity of the target space so that
in figure 8.5? This question is at the heart of the the object may be grasped. The grip-formation
research controversy concerning how reaching component is responsible for preparing the hand
and grasping are controlled and has motivated to capture the object, and the person's intention
two very different theoretical positions about regarding what will be done with the object
movement planning, one based on temporal determines how the grip is formed (Jeannerod,
relationships and another that concerns spa- 1996; see review by Castiello, 2005). 1
tial relationships involved in acquiring an Jeannerod (1981, 1984) and Arbib (1981; Fagg
object. & Arbib, 1998) developed these ideas further by
270 Motor Control and Learning

providing a neurologically plausible rationale for found to be invariant. The occurrence of maxi-
dissociating the transport and grip components. mum aperture size was synchronized in time with
By this view, separate visuomotor channels in the peak deceleration of the wrist, regardless of
the central nervous system (CNS) are responsible the distance. An idealized trial representing the
for limb transport and grip formation. Jeannerod kinematics as reported by Jeannerod is illustrated
proposed that each channel independently in figure 8.6. As shown in the figure, the time to
processes different information in movement peak grip aperture coincided with the point of
planning. From a viewer-centered perspective, peak deceleration of the wrist, at approximately
an object's extrinsic properties (e.g., its distance 75% of the overall movement time (MT) (Jean-
from the body, its location in space, or both) are nerod, 1981).
used to plan the details of limb transport (e.g., Not all studies have supported the temporal-
the spatial coordinates of the wrist). Grip for- coordination view of reaching and grasping, how-
mation is determined from the object-centered ever, and the evidence for and against this view
perspective-the object's intrinsic properties (e.g., is mixed and complex (e.g., Bootsma, Marteniuk,
size, shape, texture) being used to determine the MacKenzie, & Zaal, 1994; Marteniuk, Leavitt,
specific details of hand shaping. Coordination of MacKenzie, & Athenes, 1990). As noted by Wing,
the reach and grasp was hypothesized to follow a Turton, and Fraser (1986), research is not necessary
temporal patterning-where the timing of various to show that arm movements and hand shaping
kinematic landmarks is scaled proportionally. can be performed independently, as each action
By independent processing, Jeannerod meant can be done in the absence of the other. When the
that changes in one type of visual information two are performed together, though, the evidence
(i.e., an intrinsic or an extrinsic property) would suggests that reaching and grasping are not inde-
affect one movement component but not the pendently controlled actions. A common research
other. In support of the theory, reaching for objects method that demonstrates this phenomenon
at various distances from the starting location involves a perturbation during the action, such as
(extrinsic information) was found to affect the an unexpected change in the location of the object
kinematics of the transport component only, but (Gentilucci, Chieffi, Scarpa, & Castiello, 1992;
not the grip-formation component (e.g., Jean- Paulignan, MacKenzie, Marteniuk, & Jeannerod,
nerod, 1981; Paulignan, MacKenzie, Marteniuk, & 1991), in the size of the object (Castiello, Bennett,
Jeannerod, 1991; but see also Jakobson & Goodale, & Stelmach, 1993; Paulignan, Jeannerod, MacK-
1991). Coordination of the two components was enzie, & Marteniuk, 1991), in the spatial orienta-

c:
.
C)
-s·
0 Q)
~ "O
CD
Q; 0 ~
Qi c:
(J
(J
~
<C 0
2.

0 0.25 0.50 0.75 1.00

Proportion of total MT

FIGURE 8.6 Illustration of relation between wrist transport kinematics and grip aperture size-idealized trial.
Based on Jeannerod 1981, 1984.
 Coordination 1111 271

tion of the object (Desmurget et al., 1996), or via a Brenner, & Biegstraaten, 2002; Smeets, Brenner,
mechanical perturbation of the moving limb itself & Martin, 2009) extended Wing's idea to suggest
(Haggard & Wing, 1995). The magnitude of effects that both the thumb and the finger(s) are aimed
observed in these studies was influenced most at specific locations on the object. Importantly,
when a change in the type of grip was required a change in either the position of the object (an
(Paulignan & Jeannerod, 1996). In general, these extrinsic property in Jeannerod's scheme) or the
studies demonstrate that any sudden change in size of the object (an intrinsic property) is com-
either the perceptual or effector attributes of the pensated for by simply changing the targeted
action will result in a reorganization of both the position for the thumb and finger(s).
reach and grasp components: The coordinated A related view by Rosenbaum, Meulenbroek,
component parts of the action are not reorganized Vaughan, and Jansen (2001) also considers reach-
independently (for further discussion see Goodale ing and grasping as a complex target-aiming task
& Servos, 1996; Jeannerod & Marteniuk, 1992; Pau- (see also Rosenbaum, Loukopolous, Meulen-
lignan & Jeannerod, 1996). In sum, this evidence broek, Vaughan, & Engelbrecht, 1995; and Rosen-
does not provide strong support for a temporal baum, 2009). In their model, a reach-and-grasp
view of coordination in the planning of reaching action occurs by means of a selection process in
and grasping actions. which various potential goal postures (a forward
model of the position adopted by the hand
Spatial Planning in Reaching and arm when the object has been grasped, see
and Grasping chapters 5 and 6) are compared with the current
An alternative view of movement planning posture of the hand and arm. The best candidate
considers reaching and grasping not as sepa- posture for the action is then selected, based on
rate components of an action, but instead as a a process in which certain movement constraints
coordinated aiming movement. From chapter must be met during the action (e.g., the resulting
7 we know that spatial variability increases if movement will avoid collisions, and effort is to
the same movement distance is completed in be economized). Models of the resultant kinemat-
progressively shorter MTs (Schmidt, Zelaznik, ics of this motion-planning view appear to agree
& Frank, 1978; Schmidt, Zelaznik, Hawkins, quite well with existing data and, together with
Frank, & Quinn, 1979). A similar effect appears the Smeets and Brenner theory, provide excit-
to apply in reaching for objects. In a study by ing and potentially significant advances in our
Wing and colleagues (1986), subjects were asked understanding of how reaching and grasping
to grasp an object by reaching for it as rapidly are coordinated.
as possible, or reaching for it at a comfortable
speed, with or without vision. Under full-vision Bimanual Coordination
conditions, the size of the grasp aperture widened Using the two hands together is often studied
as movement duration decreased (735 ms to 376 as a different type of coordination than the type
ms). Further widening of the aperture occurred we have just examined. Sometimes the hands
when the movements were made with the eyes perform similar actions (e.g., lifting a baby from
closed. These findings suggested that subjects the crib; steering a car), and sometimes the two
compensated for anticipated increased spatial actions are different (e.g., holding a bottle with
variability in limb transport by increasing the one hand and twisting the cap with the other;
size of the grasp, thereby maintaining a constant using a fork with one hand to stabilize a piece of
success rate by adjusting the tolerance for spatial food while using a knife to cut it with the other
error as needed (see also Wallace & Weeks, 1988; hand). How the control of one hand interacts
Wallace, Weeks, & Kelso, 1990). with the control of the other hand while both are
In contrast to a temporal-components view of moving at the same time represents a fundamen-
reaching and grasping, as we discussed previ- tal problem for scientists who study movement
ously, Wing (Wing & Fraser, 1983; Wing et al., coordination.
1986) suggested that a reach-and-grasp action
requires aiming the thumb at the object, then The Bimanual Fitts Paradigm
shaping the fingers as appropriate to acquire The movement amplitude and size of the target
the object. Smeets and Brenner (1999; Smeets, to which a limb is aimed are strong determiners
272 Motor Control and Learning

of MT-a relationship identified most commonly al., 1979; see also Corcos, 1984; Marteniuk, MacK-
with Fitts' law (chapter 7). However, this law was enzie, & Baba, 1984). Thus, it is probably more
based on experiments in which only one limb was appropriate to conclude that the limb that moved
moving. What happens to MT when two limbs to the more "difficult" task exerted a strong deter-
are moved? And more importantly, what hap- mining influence on the other limb (Riek, Tresilian,
pens to MT when the task indexes of difficulty Mon-Williams, Coppard, & Carson, 2003).
(IDs) assigned to the two limbs are different (i.e., This conclusion is supported by the results of
incongruent)? These questions were addressed in a variant of this rapid bimanual-task paradigm.
experiments by Robinson and Kavinsky (1976) Subjects moved both hands simultaneously the
and more thoroughly by Kelso, Southard, and same distance and to the same-sized target, but
Goodman (1979). Subjects in the study by Kelso with a cardboard hurdle placed between the home
and colleagues made bimanual aiming move- position and the target for one of the limbs only,
ments in which each limb moved to a separate so that one limb was forced to move with a higher
target as rapidly as possible. Two target IDs were trajectory than the other to get to the target (Good-
used: In an "easy" task, the subject moved a short man, Kobayashi, & Kelso, 1983; Kelso, Putnam, &
distance to a large target (ID= 0.8), and in a "dif- Goodman, 1983). In the study by Goodman and
ficult" task, the subject moved a long distance to colleagues, the height of the hurdle for one limb
a small target (ID= 3.7). was systematically varied from 0 cm (no hurdle) to
Kelso and colleagues (1979) found that the 40 cm, while the other limb always moved without
answer to the question of having both hands a hurdle. The MT findings are presented in figure
moving depended on which task was examined. 8.7. As expected, the MT of the limb going over
The effect of amplitude and target width held the hurdle increased markedly with increased
true for the limb assigned to the high-ID task. height of the hurdle. However, the limb that had
This target was reached in about the same MT no hurdle also revealed an increase in its MT as a
(average of 147 ms) regardless of whether the limb function of the hurdle height that the other limb
moved (a) alone (i.e., "unimanually"); (b) together was assigned to clear, showing the tendency for
("bimanually") with the other limb, which was the two hands to produce a similar pattern. Once
also moving to a high-ID task; or (c) bimanually again, it is important to note that the increased
but with the other limb moving to a low-ID task. MTs of the no-hurdle limb were not identical to
However, amplitude and target width did not the increased MTs of the limb that had to clear
always predict MT for the limb moving to a low-ID a hurdle; rather, the MTs of the two limbs were
task. The MT for this limb changed depending on influenced similarly by the hurdle. The size of the
the task conditions: (a) The MT was 98 ms when difference in MTs of the limbs with and without
performed unimanually; (b) the MT was slowed a hurdle increased with hurdle height, as can be
slightly (to 105 ms) when the limb was paired seen by comparing the slopes of the filled and
with a limb that also moved to a low-ID task; and open circle functions in figure 8.7. In the Goodman
(c) MT slowed much more (to 130 ms) when the and Kelso studies (Goodman et al., 1983; Kelso et
limb was paired with a limb moving to the high-ID al., 1983), the effect on the no-hurdle limb was also
task. In summary, incongruent bimanual-task ID seen in the kinematic analyses of the movement
conditions influenced the MT for a limb assigned trajectories. The trajectory required in order for
to an "easy" task, but did not influence MT for a the limb to clear the hurdle was accompanied by
limb assigned to a "difficult" task. a similar tendency in the no-hurdle limb (although
These findings were replicated by Fowler, the effect was not a mirror-image mapping of the
Duck, Mosher, and Mathieson (1991), who used two limbs). The bimanual aiming task demands
procedures identical to those of Kelso and col- influenced the control of the two limbs in terms
leagues (1979, experiment 3) but added another of both space and time.
aiming condition in which the task had an even A much stronger temporal interdependence
larger ID (of 5.2). However, in both of these stud- appears to occur with use of a bimanual reach-
ies, it is important to note that the incongruent ing/ grasping task. Subjects in studies by Jackson,
bimanual MTs for the low-ID task (130 ms in Kelso Jackson, and Kritikos (1999) reached both hands
et al., 1979) remained conspicuously smaller than forward to grasp small wooden dowels. In one
the MTs for the high-ID tasks (147 ms in Kelso et experiment, the two dowels were the same size
 Coordination 1!11 273

500

450

400

0
.§. 350
1-
:i:

300

250

10 20 30 40
Hurdle height (cm)

FIGURE 8.7 Movement times (MTs) for two limbs. The two limbs are moving simultaneously toward targets of equal dif-
ficulty, but only one limb (open circles) must go over a hurdle that varies in height. But, MT for the other limb, which does not
go over a hurdle, also increases (filled circles).
Adapted from Goodman, Kobayashi, and Kelso 1983.

but could be placed either at the same distance number of experimental paradigms. Instead of
from the home position (congruent bimanual varying target widths (as in the Fitts task), Mar-
task) or at different distances (incongruent task). teniuk and colleagues (1984) asked their subjects
In another experiment, the dowels were located to move as rapidly as possible and as closely as
at the same distance but were either the same possible to a small target point, and measured
size or different sizes. Regardless of the experi- the effective target width (We in Schmidt et al.,
mental manipulations, the effect on the temporal 1979; see chapter 7). Marteniuk and colleagues
outcomes was dramatic. When the limbs were found that one limb tended to overshoot a short
required to reach and grasp dowels in the incon- target when the other limb moved to a far target.
gruent conditions, all of the temporal measures of They also observed a small tendency for a limb
the reach-and-grasp actions were nearly identical to undershoot a far target when paired with a
for both limbs, and were very different compared limb moving to a short target. Experiments by
to the MTs in each task performed unimanually. Sherwood (1991, 1994) suggest that these spatial
Although more research needs to be done with assimilation effects occurred because the force
this paradigm (see also Jackson, German, & Pea- output of the limb moving the lesser distance was
cock, 2002), it appears that temporal coupling in some way biased by the greater force output
of the limbs is stronger when a reach-and-grasp required to move the other limb a farther distance.
action is the goal than in the bimanual Fitts In all these tasks, the goal was to produce
aiming task. Why there is a difference in these the same action in the two limbs, but with dif-
effects for reaching/ grasping versus the Fitts task ferent amounts of distance or slightly different
is not clear at this point. trajectories. The timing of the major force pulses
or reversals in direction was nearly identical in
Spatial Coordination the two limbs. In terms of motor programming
Spatial coordination can be observed when two views, the two limbs could have been controlled
or more body parts are assigned to perform dif- by the same GMP, but with different parameters
ferent actions, or arise as the result of a common assigned to each of the limbs. These results sug-
goal (see "Head-Arm Coordination in Golf gest an interlimb interference, not at the level
Putting"). These effects have been studied in a of the GMP, but rather in terms of the assigned
 Head-Arm Coordination in Golf Putting
Coordinating the motions of the arms and head is commonly emphasized in the act of putting in golf,
and most experts recommend that the head be kept completely motionless while the arms swing like a
pendulum during the putting stroke (e.g., Ni.cklaus & Bowden, 1974, p. 238; Woods, 2001, p. 37). However,
recent research suggests that this recommendation is violated not only by less skilled golfers, but also by
expert putters (Lee et al., 2008a, 2008b). Figure 8.8 illustrates the simultaneous velocity profiles of the
putter (held by both arms) and the head during the backswing and downstroke of the golf putt (lasting
about 1 s in duration). The typical less skilled golfer moved both the head and the putter away from the
direction of the putt during the backstroke, and then moved both the head and putter toward the direction
of the putt during the downstroke. All the pairs of velocity profiles from the 60 trials in the experiment
for one subject are overlaid in figure 8.8a, revealing a head-arm (putter) coordination pattern that was
repeated almost identically on each trial. This tight coupling of the head and arms in the same direction
during the putt resulted in high positive correlations (+0.78; correlations computed within each trial).

2000 150
2000 150
1500 100
~ :c ~ 1500 100 :c
(I)
E
(I) E Q)
Q)
a. a.
..§.1000 50 ..§. 1000 50 ~
> ~ >
.'!: 0()
;!:::
0() (.)
(.)
500 0 0 500 0
0
~ ~
Gi ~
-50 3
>
-50 3 1ii 0
~::l
0
ll. ~
3 =::l
a.
3
~
-500. -100 -500 -100

-1000 i:..;:;;:.:.r.;_""'-°"'-""'-"-'-°"'-°"'-""'-""'--'--'-.:.-l:J -150

-'-1000 -150 942
a Putt duration (ms) 908 b Putt duration (msec)
FIGURE 8.8 Velocity profiles of head and putter (arm) movements for (a) a typical less skilled golfer and (b) an expert
golfer. Each figure contains 60 pairs of traces, corresponding to each trial in the experiment.
Reprinted, by permission, from T.D. Lee.et al., 2008b, "Do expert golfers really keep their heads still while putting?" Annual Review of Golf Coaching 2: 135-143.

A typical set of trials performed by the expert golfers is illustrated in figure 8.8b. Once again, the
pairs of velocity profiles are nearly identical on the 60 trials in the experiment for this golfer. How-
ever, the major difference, compared to the less skilled golfer, was that the expert golfers moved
their head in a direction that was opposite to the motion of the arms during the putt. Experts moved
their heads toward the direction of the putt as the backswing moved away from the ball. Then, at the
same time the putter reversed its direction, the head reversed its direction too. This tight coupling
of the head and arms in the opposite direcUon during the putt resulted in high negative (-0.70)
correlations of the velocity profiles.
These findings have a number of important implications regarding coordination of degrees of
freedom in putting. One implication is that golfers of varying skill level reduce the complexity of the
putt by tightly.coordinating the motions of the arms and head to act as a GMP. Coordinating the inter-
dependence of the two limbs to act as a single unit of action-both for the less skilled and the expert
golfers-is consistent with the evidence that a single GMP governs the coordination of these degrees
of freedom (for details see the research of Heuer, Schmidt, & Ghodsian, 1995, discussed later in the
chapter).Another implication of these findings is that, during the motor learning process (presumably
with practice and augmented feedback), the GMP for the putt undergoes a fundamental change from
a same-direction to an opposite-direction coupling of the head and arms. However, it does not appear
that the skilled golfer keeps the head completely still during the putt, despite how it might appear.

274
 Coordination 275

parameters. One question that arises is whether movement could be detected in the action of
such assimilation would occur when the tasks the arm doing the simple movement, although
require different GMPs. the amount of interference was not sufficient to
Swinnen, Walter, and colleagues examined make the arms produce identical movements
interference in tasks that required the upper limbs (figure 8.9, band d). The patterns of forces in the
to perform two completely different patterns at two arms occurred at essentially the same time,
the same time (e.g., Swinnen, Walter, & Shapiro, as if the pattern of contractions in the complex-
1988; Walter & Swinnen, 1990)-that is, in tasks movement arm were somehow overflowing" to /1

for which a given program controlling the right the simpler arm action. Furthermore, the interfer-
arm could not have been merely scaled to pro- ence appeared to be greater when the left arm
duce the required action in the left arm. Walter performed the complex task and the right arm
and Swinnen (1990) had subjects produce an 800 produced the simple task; adding a load to the
ms movement with two reversals in direction complex-movement arm tended to increase this
with one arm and a unidirectional movement bias, and practice reduced the interference (see
in the other arm. When these actions were done Walter, Swinnen, & Franz, 1993, for a review).
separately, the typical position-time records These authors have discussed their results in
resembled the traces shown in figure 8.9, a and terms of what they call structural and metrical
c. But when the subjects were asked to produce coupling of the two limbs, with practice serving to
the two actions simultaneously, evidence of uncouple, or dissociate, the actions of the arms (e.g.,
the pattern of the arm performing the complex Swinnen, 1992; Walter & Swinnen, 1992). How this

Unimanual Bimanual

Ul 60 Ul 60
Q) Q)

~ 50
Cl
~ 50
Cl
Q) Q)
'C 'C
-.:: 40 -.:: 40
c: c:
Q)
Q)
E 30
Q)
E
Q)
30
(.) (.)

-a 20
t/)
-a 20
t/)

c 10 c 10

200 400 600 800 200 400 600 800

a Time (ms) b Time (ms)

Ul 60 Ul 60
Q) Q)

~ 50
Cl
~ 50
Cl
Q) Q)
'C 'C
-.:: 40 -.:: 40
c: c:
Q)
Q)
E 30
Q)
E
Q)
30
(.) (.)

-a 20
t/)
-a 20
t/)

c 10 c 10

200 400 600 800 200 400 600 800

c Time (ms) d Time (ms)

FIGURE 8.9 Bimanual coordination effects for limbs with different displacement-time goals; two-reversal task (a and b),
unidirectional movement (c and d}, unimanual performance (a and c), bimanual performance (band d).
Adapted from Brain and Cognition, Vol. 14, C.B. Walter and S.F' Swinnen, '1\symmetric interlimb interference during the performance of a dynamic bimanual task," pgs.185-200,
Copyright 1990, with permission from Elsevier.
276 Motor Control and Learning

uncoupling occurs, though, is not entirely clear. patterns. In fact, supporting this single-GMP
One possibility is that independent control of view, subjects became able to speed up or slow
each limb is acquired through the development of down this entire bimanual pattern on command,
separate motor programs. This seems unlikely for changing the left- and right-arm speeds together.
a number of reasons, most important of which is But, when instructed to produce the arm move-
the evidence presented in chapter 4 that the CNS ments at different speeds (e.g., speed the right
appears incapable of programming and initiat- arm and slow the left), subjects found this vari-
ing two different actions at exactly the same time. ant nearly impossible to do (Schmidt et al., 1998).
Another possibility is that some different form These data imply that the right and left arms
of interdependence emerges that allows the limbs were being scaled together in time so that rela-
to do slightly different things at the same time. tive timing (both within and between arms) was
Heuer, Schmidt, and Ghodsian (1995; also essentially invariant. In other experiments, where
Schmidt, Heuer, Ghodsian, & Young, 1998) the movements were done even more quickly, the
studied movement performances in a situation landmarks between the two arms were even more
where the two arms produced deliberately dif- strongly coordinated. Therefore, on the basis of
ferent rapid, discrete patterns. The seated sub- the logic of units of action presented earlier, the
jects moved two horizontal levers with elbow separate actions produced by the right and left
flexion-extension movements to produce specific arms were part of the same GMP. This result shows
space-time patterns for each limb. The right arm increased coupling between the limbs with prac-
produced a flexion-extension-flexion pattern tice-not decoupling as argued by Swinnen (1992;
through about 70°, while the left arm simulta- Walter & Swinnen, 1992), discussed earlier. This
neously produced a Hexion-extension pattern finding also provides a basis for the idea that com-
through about 60°, with MTs ranging from 400 plex, multilimb actions-in which all of the limbs
to 600 ms. Subjects had considerable difficulty at are doing different things, as in pole-vaulting,
first, but after several hundred trials they were or the golf-putting example given earlier-can
able to produce the actions very easily. probably be controlled by a single, learned GMP.
From the acceleration-time traces, Schmidt Apparently, when two limbs do different things
and colleagues (1998) defined 12 landmarks, at the same time, this does not necessarily mean
essentially as had been done in the analysis that they are operating independently. 2
of units described earlier (see figure 8.2). The
within-subject (across trials) correlations among End-State Coordination
all possible pairs of landmarks were then com- Recall from chapter 3 (see section on "Compatibil-
puted (see figure 8.3). The correlations among ity and Complex Actions" on p. 74) that, in some
the landmarks within each arm were generally situations, we choose rather awkward initial limb
very high, averaging 0.91 (right) and 0.94 (left), postures in order to maximize biomechanical
especially when the actions were done as quickly efficiency (or "end-state comfort") at movement
as possible (Schmidt et al., 1998). The relatively completion (Rosenbaum, 2009; Rosenbaum et al.,
high correlations among all of the landmarks 1990). In these cases, selecting a movement plan
suggested that each arm produced a one-unit to satisfy a specific (end) goal appears to be a
pattern, as discussed earlier. natural and spontaneous process. So, how would
The interesting point for coordination, how- two limbs be coordinated if each had a specific
ever, concerns the correlations among the land- but incompatible end goal? Would the end-state
marks between the arms. These correlations were comfort effect be traded off in order to achieve
also generally very high, especially when the a well-coordinated movement? Surprisingly,
actions were very rapid, and were nearly the same individuals do not prefer one coordination pat-
as those for the individual arms (mean = 0.91). tern to another-they tend to choose the pattern
These findings supported the view that the coor- most likely to satisfy a goal that maximizes end
dination between the two arms was controlled as comfort for both limbs, as described next.
a single unit (i.e., by a single GMP controlling both Kunde and Weigelt (2005) had subjects per-
hands at the same time, as discussed in an earlier form the tasks illustrated in figure 8.10, involv-
section in this chapter), even though the two limbs ing the raising of two blocks from a horizontal
produced completely different spatial-temporal to a vertical orientation. These actions could be
 Coordination 111 277

Ill
Symmetrical movements Asymmetrical movements
Continuous Tasks
8
Cl
'E A recurring issue throughout this book is the dif-
CD
:J
.... ference between discrete and continuous skills in
Cl
c
0 how motor control is achieved. Although some
0
commonalities exist (e.g., timing processes),
Ill major differences abound as well (e.g., role of
iii
0
Cl
feedback and motor programs). In this section, we
'E
CD
will see that many researchers who study coordi-
:J
....
Cl
nation in continuous tasks take a fundamentally
c different view regarding the process of movement
0
u
.5 control than researchers who study discrete tasks .
"Costs and Benefits of Coordinating Two Limbs"
(p. 279) presents two little experiments for you
FIGURE 8.10 Subjects performed bimanual movements
to raise two blocks, with colored ends, from horizontal to
to try, which represent examples of some funda-
vertical positions. Congruent goals were defined as raising mental processes when continuous movements of
the two colored ends to the same vertical position. Incon- two or more limbs are performed simultaneously.
gruent goals required one colored end to be on top and the These phenomena have interested scientists for
other on the bottom of the vertical position. Symmetrical
movements involved hand actions in which both raised over a century but have been the topic of intense
blocks were turned inward or outward. Asymmetrical move- experimental scrutiny only in recent years.
ments required both hands to move in the same direction.
Reprinted, by permission, from W. Kunde and M. Weigelt, 2005, "Goal congruency
in bimanual object manipulation," Journal of Experimental Psychology: Human
Early Research
Perception and Performance 31: 145-156.
Bimanual coordination of continuous tasks in
humans has interested researchers for many
performed with either symmetrical or asym- years (see "R.S. Woodworth on Coordination").
metrical hand movements, resulting in end For example, right-handed subjects in Langfeld's
goals that were either congruent or incongruent (1915) experiment tapped one finger as rapidly as
with respect to object orientation (see figure 8.10 possible for 30 s--either the index or middle finger
for details). The main finding was that subjects of the right or left hand. Between-finger differ-
showed no preference for symmetric over asym- ences on the same hand were small, but between-
metric hand movements. Rather, the achievement hand differences were large: Average MTs for the
of congruent goals with both hands resulted in right-hand taps were 181 ms compared with 222
responses performed with faster RTs and MTs ms for the left hand. This is not too surprising;
and fewer errors than incongruent goals, regard- asymmetries in performance are frequently found
less of how the actions were performed. Other when left- and right-hand performances of right-
experiments, in which subjects were allowed handed people are compared (e.g., Elliott & Roy,
to choose the orientation of the hands, revealed 1996; Sainburg, 2005). However, on some trials,
that even an uncomfortable coordination pattern subjects were asked to tap one finger on each hand
was preferred if it maximized end-state comfort such that the two taps were made simultaneously
(Fischman, Stodden,· & Lehman, 2003; Hughes and as fast as possible. When subjects tapped with
& Franz, 2008; Janssen, Beuting, Meulenbroek, the same finger on both hands (e.g., the right and
& Steenbergen, 2009; Weigelt, Kunde, & Prinz, left index finger at the same time) the average MT
2006). These findings are similar to the results was 208 ms, which represents a 27 ms decrement
seen in unimanual tasks (Rosenbaum, 2009). for the right finger (compared to 181 ms in uni-
These findings suggest that the nature of the manual tapping) but a 14 ms facilitation for the left
coordination pattern appears to be subservient hand (compared to 222 ms). Langfeld (1915) thus
to the desired end result. In the case of end-state showed that, in the process of becoming coordi-
comfort effects, biomechanical efficiency at move- nated, the temporal, "independent behavior" of
ment completion appears to take precedence over the two hands changed dramatically. The timing
how the limbs move there. of the two hands became much more interdepen-
278 Motor Control and Learning

R.S. WOODWORTH ON COORDINATION

Much of the currentworkon coordination ha:; i.ts roots in monographs that were not available in
English until quite sometime after their original publication: Some of these works were eventually
translated and had a very influential effect on English readers when they were published (e.g.,
Bernstein, 1967; von Holst, 1937/1973). In other cases, contemporary researchers have made
us aware of important untranslated works (e.g., Heuer, 1996; Latash & Zatsiorsky, 2001; Wor-
ringham, 1992).Another source of 19th-century work on coordination-a brief review published
in French by Woodworth (1903)-had not, to our knowledge, been translated into English prior
to the third edition of th.is book. A comparison of the following quoted passage to some of the
"modern" work on coordination reveals some remarkable early insights regarding spatial and
temporal coordination, which became a dormant issue in research for quite some time.
"It is common knowledge that one can execute with ease simultaneous, corresponding move-
ments with the right and left hand. One must, however, make an effort so that they correspond;
if one moves the right hand in mid-air, tracing any shape, one must devote attention to the left
hand simply for it to move, for it will trace the symmetrical corresponding shape. The connection
is more between the innervation of corresponding muscles on both sides of the body.
"Munsterberg proved that there was another relationship between movements of both sides.
When an arm is balanced in front, the other must naturally balance itself, not in front, but behind;
there are also other cases in which symmetrical movements on both sides must naturally alter-
nate with one another. It is not less true that it is easy to execute simultaneous movements that
correspond on each side.
"Ostermann found that this ease appeared only in bilateral symmetrical movements. If we
tried to make simultaneous movement, symmetrical in relation to a horizontal plane, the attempt
would result in inexactness and confusion. The attempt had not been "natural" and was executed
without confidence, whereas bilateral symmetrical movements were executed with ease and
confidence and some achieved exactness.
"Bloch discovered that in order for the symmetrical movements to be executed with ease
and precision, both arms have to move at the same time. If one moves an arm toward a certain
point, the effort that one will make to place the other arm to the corresponding point will result
in inexact movement" (Woodworth, 1903, pp. 97-98).

dent, resulting in a facilitation in performance a fast walking gait or a running gait. Each gait
for the finger on the nondominant hand and a has distinct relative timing and biomechanical
decrement for the finger on the dominant hand. characteristics. Within a certain range of speeds,
The struggle between the tendency of one oscil- either gait will get the job done. What process
lating limb to maintain an independent motor determines the gait that will be used? As it turns
behavior (termed the maintenance tendency by von out, this question is not a trivial one; and the
Holst [1937 /1973]) and the tendency of its behavior answer has provided considerable evidence to
to become coupled with another oscillating limb support an important view of motor control.
(which von Holst called the magnet effect) captures
a critical feature of temporal coordination in con- Animal Research
tinuous skills. As we will see in the next sections, In chapter 6 we discussed the mesencephalic
these strong natural tendencies play an important preparation in cats. Walking could be initiated in
role in the coordination of continuous actions. these spinalized cats on a treadmill; the provided
stimulation was not rhythmical and in time with
Gait Transitions the walking patterns, though. One important
Watch people as they hurry to catch a bus. They finding was that increases in the speed of the
usually choose one of two gaits to get to the bus: treadmill induced changes from a walk to a trot,
 Coordination !ill 279

Two quick and easy little expe~iments w1n~~ijibb~tr11~~

is already moving can h~we ~oth posi~v7 arr~t~eg
first experiment, using ablackboard and c~a:11<.,· try.t6
usual speed but backward (i.e., producin~a•mir~~rr •.
is very difficult to do and that fhe. patterrfis barely l'e
do it simultaneously with the dorrlir8.r;\~ hand ··· · ·
directions. You should find tnatthe pa!lern ·.........· ·.···
nondominant hand is much smoother andlthat .·
the two patterns are roughly similar (bi:itmifrb'r
images of each other). More importantfY;asyoti
perform the two-handed task, notice fiOwth~two·
limbs seem to be "locked together'' (or"ccn.lpte~?')>
Even though the nondominant handwriting. is
somewhat more uncontrolled thantlw.domina;nt,:;
a characteristic pattern still emerges.,..,_the sc:ime
loops, straight parts, and soon: <··!•>
Sit. in a chair to do the· second e~perif!.leot.
Begin by making clockwise circles on t.he floor.
with your right foot. Then, while continu!("l9.to Fl1ake ~•.·•
circles, draw the number "6" i.n theairwjth,your··.
right hand. What happened to your foot's1J10Ye-.; ·. ·
ment? In all likelihood, your foot. no longer c:on~.
tinued to make circles. Now, try this experiment
again, combining the right-handed 6s"'fithyourJe~
foot making the circles; then againwithyour'fe~
hand making the circles. Nowtry aHthree~gairrby.
making the circles in the counterclockwisedtrec:~ .·.·.
ti on. Was the circle drawing easier to maintain ill! •
any of these six experimental condition~? •··· . ·.•· ... ·... '.
In performing these two. experim~nts,noticE)•:>•
how the two limbs·. tend to want to b~ t~q~ed · •
together, as if you had lost some c:6ntr9tol tnerJ;I.. ·:~
and they· were being controlled Py son:ie ~qth~r. ·.•. .
influence. In the handwriting experif!.l~fl't,.thi~·~~o... •
dency has a facilitating effect~thep~~lli:)'n~~<Ze ··· ._....._ _ _ _....._..._...__
of the nondominant limb lmptoves.reiatl:\!'e•~~. In the stadiums of ancient Rome, the most
the unimanual (control) condition~.lntl'l~~lrpl~~~: t feared trial was the rub-your-stomach-and-
drawing experiment, this sarne tendenci:r~fli'·;· pat-your-head-at-the...same-time event
detrimental effect~theperformanceqf•t~~6fr~J~·>' c.::·•.: ..•. • , .. .,,i
declines relative tethe cllntml conditionn~~s1:19~.~~~e~·~~'*
circle-drawing conditions, though, theten£1e.n~les tb'.d~t;tel~·~~
in.
tasks and the effectors that are involyed per:fqrrnioQ tl'l~~'~X·

and occasionally to a gallop. Although the animal Gait patterns and the transition between gait
could not feel its legs while in locomotion, these patterns in animals with four or more limbs
specific gait patterns, and more importantly, the are fascinating to compare across species (e.g.,
qualitative (or "nonlinear") change from one gait Alexander, 2003, chapter 7). Of particular interest
pattern to another, still occurred on occasion. here is the process by which an animal selects a
280 Ill Motor Control and Learning

particular speed within each gait, and the process the maximum energy efficiency, defined at the
that determines when the animal will change minimum points on the walk and trot curves,
gaits. A leading hypothesis is that gait selection corresponded rather well with the speeds that the
and the decision to change gaits are both based horses selected spontaneously when locomoting
on a principle of minimal energy costs (Alexander, at different speeds. Hoyt and Taylor (1981) sug-
2003). Consider the analysis of horse gaits in figure gested that energy efficiency might represent a
8.11 (Hoyt & Taylor, 1981). The histograms in the basis used by an animal for the selection of specific
bottom half of figure 8.11 illustrate the frequency speeds within a gait, and that energy inefficiency
of gaits that a horse selected spontaneously serves as the catalyst or trigger to change gaits.
when the treadmill speed was changed. These
histograms represent a subset of the total range Human Research
of speeds that the horse could select within each Humans typically use only two gaits, walking
gait. Taking this analysis further, Hoyt and Taylor and running. The transition between these two
measured the energy cost (defined in terms of gaits occurs between 2.0 and 2.2 m/s (about 4.5
oxygen consumption) of horses that were trained and 4.9 mph), but this can change depending on
to walk, trot, and gallop at a range of speeds on a number of factors, such as whether the person
a treadmill (speed now being determined by the is accelerating (a walk-to-run transition) or
experimenter). The data in the top half of figure decelerating (run to walk) (Diedrich & Warren,
8.11 suggest that the energy costs of walking 1995), the rate of acceleration (Van Caekenberghe,
and trotting were rather high at both extremes Segers, De Smet, Aerts, & De Clercq, 2010), and
of the speed ranges within each gait. Moreover, cognitive load (Daniels & Newell, 2003). On the

28

24

E
,.... 20
CD
>
0
E
g 16
"O
CD
E
::I
Ill
z
c
c 12
3
0
CJ
......
g 10 ..
C'
ID

0
C\I 8 -
0
0

..a
C'
Ill
ID
5 <
4 i5'
::I
Ill

4 5 6

Walk I• •I I• Gallop~I
1----Trot -----1
Running speed (mis)

FIGURE 8.11 Oxygen consumption at various speeds in the walk, trot, and gallop gaits of a horse. The histograms just
above the x-axis represent the preferred speeds that the horse selects in each gait.
·Reprinted from Nature Vol. 292, D.F. Hoyt and C.R. Taylor, 1981, "Gait and the energetics of locomotion in horses;• pgs. 239-240, Copyright 1981. By permission of D.F. Hoyt.
 Coordination Ill! 281

basis of the energy-trigger view from the animal sure on the x-axis, the so-called Froude number,
research just presented, one hypothesis would represents speed normalized for leg length. 3) The
suggest that inefficiency in metabolic costs might results are clear: For the relative phase in both
be a leading catalyst for this transition. Support for segments measured, walking became more vari-
this view would be strong if oxygen consumption able (less stable) at speeds higher than the normal
levels were drastically different between perfor- transition point, and running became more vari-
mances at the pre- and postgait transition speeds. able at speeds lower than the normal transition
An experiment to test this prediction, however, point (Diedrich & Warren, 1995; but see also Kao,
showed no such drastic change, even when load Ringenbach, & Martin, 2003). Note, however, that
conditions were added to increase overall energy these data do not provide clear evidence that
expenditure (Raynor, Yi, Abernethy, & Jong, 2002). pattern instability is necessarily the trigger for
An alternative view of the trigger process sug- a gait transition, just that improved stability is
gests that the gait pattern begins to lose stability concomitant with a gait transition.
at speeds beyond the normal transition point Neither the energetic nor the stability hypoth-
(Diedrich & Warren, 1995, 1998a, 1998b). In these esis appears to explain the findings of the role
experiments, Diedrich and Warren examined of vision on gait transitions, however. In a recent
the variability in the ankle-hip and ankle-knee study by Mohler, Thompson, Creem-Regehr, Pick,
segments in terms of within-limb relative phase and Warren (2007), subjects walked on a treadmill
as subjects walked and ran at speeds that were that provided an "endless hallway" virtual envi-
greater and less than those in the normal transi- ronment.Walk-to-run and run-to-walk gait transi-
tion region. Results from one of these studies are tions were stimulated by increasing or decreasing
illustrated in figure 8.12. Here the normal transi- the treadmill speed periodically over the dura-
tion range appears as the shaded region of the tion of the trial (which is typical of this research
graph, and measures of the SD of relative phase protocol). However, the additional manipulation
for the ankle-knee segment are plotted for both used by Mohler and associates (2007) was to
walking and running at various speeds. (The mea- covary the visual feedback provided by the virtual
environment. In a control condition, the virtual
visual feedback was matched to the visual array
8
that the subject would have received in a normal
environment for that biomechanical speed (a 1:1
7 correspondence, or 1.0 gain). In two other condi-
tions, the virtual visual feedback was halved (1:2
6 correspondence, or 0.5 gain) or doubled (2:1 cor-
5l respondence or 2.0 gain) relative to the actual gait
~
Q.
5
speed. Thus, a slower visual feedback condition
~ resulted in the sensation that gait was slower than
~ 4
2! was actually the case (and conversely for the faster
0 3 visual feedback condition). The results of the
c
!/) study are illustrated in figure 8.13. Regardless of
2 the gait transition (walk-run or run-walk), provid-
ing the visual sensation of a slower gait resulted in
a transition that occurred at a faster speed than in
the control condition. Conversely, the faster visual
0.25 0.50 0.75 1.00 1.25 1.50 feedback condition resulted in gait transitions at
Froude number speeds less than normal.
Mohler and colleagues' findings provide dif-
FIGURE 8.12 The variability in ankle-knee relative ficulties for both the energetic and stability gait-
phasing in walking and running at speeds that were greater
transition hypotheses. Since the only manipu-
and less than the normal transition region (Froude number
refers to speed normalized for leg length). lation in the study was virtual visual feedback
Adapted, by permission, from F.J. Diedrich and W.H. Warren, Jr., 1995, "Why (i.e., the treadmill speeds were consistent in all
change gaits? Dynamics of the walk-fun transition;• Journal of Experimental
Psychology: Human Perception and Performance 21: 183-202. Copyright©
visual conditions), there were no biomechani-
1995 by the American Psychological Association. cal or physiological reasons for changes in the
282 111 Motor Control and Learning

"pinches" by touching the finger to the thumb.

2.2 Do this with both hands such that you are making
one "pinch" about every second with each hand.
2.1 When we ask students in our classes to do this
~ simple task, with no further instructions, almost
s
"C
2.0 everyone does it the same way-the pinches are
Cl)
Cl)
coordinated such that the fingers and thumbs on
Q.
!/) 1.9 both hands are making contact at about the same
time. This tendency to produce a recognizable
1.8 timing pattern can be quantified by measuring
the relative phase of the between-hand timing (as
described in chapter 2). In this case, the "pinch" of
Slower Same Faster Slower Same Faster the one hand occurs at about the same time within
Visual speed its cycle as does the pinch of the other hand, rela-
tive to its cycle. (The difference in phase angles
FIGURE 8.13 Effects of actual gait speed and apparent of the two limbs within each of their respective
(visual) speed on the point of walk-run and run-walk gait cycles provides a quantitative measure of coor-
transitions.
With kind permission from Springer Science+Business Media: Experimental Brain
dination, called "relative phase"; see chapter
Research, ''Visual flow influences gait transition speed and preferred walking 2 and figure 2.14 for more details). Moreover,
speed;' 181, 2007, pgs. 221-228, B.J. Mohler et al.
the maximum opening between the finger and
thumb (which occurs about halfway through the
gait-transition speeds. Therefore, these findings cycle) also occurs at about the same time for each
suggest that gait transitions, and perhaps coordi- hand. The timing of these (and other) kinematic
nation more generally, are based on multisensory landmarks within each cycle for the two hands
information sources. result in a mean relative phase of 0°. The standard
deviation of individual scores used to calculate
Temporal Coordination the mean provides a measure of the stability of
the pattern, which is usually very stable for this
A considerable research emphasis that has task (i.e., the SD of these phase angles is typically
emerged in the past two to three decades is about 5°-10° around the mean value). The pattern
devoted to bimanual coordination of the upper is commonly described as moving in-phase, or in
limbs. Some of the initial work was motivated symmetry (chapter 2). For this task, the in-phase
by a desire to understand transitions from one coordination mode is the preferred, or most fre-
pattern to another more fully; a task involving quently adopted pattern.
the continuous cycling of two fingers provided Now, holding your hands and moving at about
a paradigm to study phase transitions between the same speed as before, try performing a differ-
any two limbs (Kelso, 1995). However, the study ent pattern. Make pinching movements with both
of temporal coordination, in general, went far hands such that the pinches alternate on the two
beyond these initial motivations and now rep- hands. You will probably find that this is also quite
resents a complex area of research investigation easy to do. Since the closing of the pinch for one
in itself. In this section we provide an overview hand occurs while the other pinch is about half-
of the research issues, paradigms, and theoreti- way through its cycle (i.e., when it is at maximum
cal orientations that have emerged in the study aperture), the relative phase of this pattern has a
of temporal coordination. Much more in-depth mean of 180°. This pattern is commonly known
treatments of this literature are widely available as moving in anti-phase, or in asymmetry.
(Kelso, 1995; Kelso & Engstmm, 2005; Swinnen, Numerous investigators have studied in-phase
2002; Turvey & Fonseca, 2009). (0° mean relative phase), anti-phase (180°), and
timing patterns with phase relations in between
Bimanual Timing Patterns (e.g., 60°, 90°, 150°); the results illustrated in
Try this simple experiment (cf. Kelso, 1984). Hold figure 8.14 are typical (Yamanishi, Kawato, &
your hands in front of you with the index fingers Suzuki, 1980). In this study, subjects tapped
pointing straight ahead, and make a series of their left and right fingers in time with left and
 Coordination 111 283

50 35

30
40
25
w 30
0 20
s
::I
C/)
0UI c
..Q 15
< 20
10
10
5

oo 180° 360° oo 180° 360°

Metronome phase offset
a b

FIGURE 8.14 (a) Absolute constant error (ICEI) and (b) standard deviation (SD) of relative phase for limbs moving at
phase offsets between 0° and 360°.
With kind permission From Springer Science+Business Media: Biological Cybernetics, "Two coupled oscillators as a model for the coordinated finger tapping by both hands," 37,
1980, pg. 221, J. Yamanishi, M. Kawato, and R. Suzuki.

right visual metronomes that were set to "beat" timing frequency (e.g., 1 beat/ s ). Now, gradually
at various phase offsets. Measures of relative- speed up the pinching frequency until eventu-
phase accuracy and variability are presented in ally you are moving as fast as possible. Then
figure 8.14. 4 The in-phase patterns (0° and 360° repeat the experiment, this time starting with the
on the x-axis) and anti-phase pattern (180°) were anti-phase pattern. If your results are similar to
performed with much greater accuracy (figure those reported by Kelso (1984; Kelso, Scholz, &
8.14a) and with more stability (lower SDs in figure Schaner, 1986), your data will resemble the results
8. l 4b) than for any of the other phase relations. In presented in figure 8.15. The open symbols in this
addition, there was a slight tendency for the in- figure represent the anti-phase pattern; the filled
phase pattern to be performed with more stability symbols represent the in-phase pattern. Mean
than the anti-phase pattern. The performances relative phase is presented in figure 8.15a and
of these two timing patterns illustrate natural relative-phase variability in figure 8.15b. As seen
and stable coordination modes for oscillatory before (Yamanishi et al., 1980; figure 8.14), both
bimanual movements and have been replicated timing patterns were performed close to their
in experiments using finger-oscillation tasks intended goal and with small variability at low
(e.g., Kelso, 1984), finger tapping (e.g., Tuller & oscillation frequencies. 5 This stable and accurate
Kelso, 1989), wrist rotations (e.g., Cohen, 1971; performance continued for the in-phase pattern
Lee, Blandin, & Proteau, 1996), swinging pen- as oscillation frequency increased. However, the
dulums (Schmidt, Shaw, & Turvey, 1993; Turvey, influence of frequency on the anti-phase pattern
Rosenblum, Schmidt, & Kugler, 1986), and other was rather startling: At about 2.25 Hz, the timing
similar movement coordination tasks. This work pattern began to switch from anti-phase to an
in-phase coordination mode. Notice also that in
has been reviewed by Kelso (1995) and Swinnen
figure 8.15b, the anti-phase pattern actually began
(2002; Swinnen & Wenderoth, 2004).
to become more variable at about 1.75 Hz, sug-
Unintended Phase Transitions Here is another gesting that the loss in stability somehow precipi-
little experiment. Perform the in-phase, pinch- tated the switch. Subjects in Kelso's experiment
ing pattern as before, starting at a comfortable had been instructed not to resist these pattern
284 Motor Control and Learning

180 60

160
50
'i 140
~
gi 120 40
~
~ 100 ~::,::;::;;;'i·,~'~;:;:;.~i ,
r::
<U
30 ~
<1> 80
(/)
<U
-a. 60 20
r::
<U
~ 40
10
20

1.25 1.75 2.25 2.75 1.25 1.75 2.25 2.75

Oscillation frequency (Hz)
a b

FIGURE 8.15 Bimanual coordination performance for patterns initiated in anti-phase (open symbols) and in-phase (filled
symbols) for (a) mean relative phase and (b) standard deviation (SD).
Reprinted from Physics Letters A, Vol. 118, JAS. Kelso, J.P. Scholz, and G. Schaner, "Nonequilibrium phase transitions in coordinated biological motion: Critical fluctuations;' pg.
281, Copyright 1986, with kind permission of Elsevier.

switches, so once the original anti-phase pattern Self-Organization Theory Haken, Kelso, and
had become an in-phase pattern at higher speeds, Bunz (1985) presented an influential model of
the new mean relative phase remained close to the results reported in figure 8.15 that has had
0° with low variability. This basic set of findings profound effects on motor control research (for
has been replicated many times, using different more details see Jeka & Kelso, 1989; Kelso, 1995,
effector pairs (see Lee et al., 1996, for a review). 2008; Kelso & Engstrnm, 2005; Wallace, 1996).
Although there are some similarities in the This model of two coupled oscillators (commonly
transitions between patterns of bimanual coordi- referred to now as the HKB model) was grounded
nation and the gait transitions discussed earlier, in the science of dynamics-literally, how physi-
there are just about as many differences. For cal systems change their states over time. Five
example, a stable in-phase pattern can be per- important effects depicted in figure 8.15 laid the
formed at all speeds, whereas a stable anti-phase groundwork for the HKB model. (1) Both the in-
pattern can be performed only at low oscillation phase and anti-phase patterns are performed well
frequencies. As well, when changing gaits as the at low movement frequencies. The system is said
speed increases we do not change from anti-phase to be bistable-there are two stable coordination
to in-phase-that is, we don't start hopping). regimes. (2) At maximal speeds, only the in-phase
However, there are many intriguing similarities pattern remains stable; the anti-phase is destabi-
too. The most important of these is the loss of lized and switches to in-phase. Thus, the system
stability that precedes a phase transition from underwent a phase transition from a bistable to a
anti-phase to in-phase. This result is quite similar monostable coordination regime. (3) The process
to the loss in stability that precedes a gait transi- of change for the initially anti-phase pattern is
tion at nonpreferred speeds (Diedrich & Warren, abrupt and qualitative-the change involved a
1995). And, in a way similar to what is seen with nonlinear transition, called a bifurcation. (4) The
gait transitions (discussed previously; Mohler et phase transition was instigated by an outside
al., 2007; see figure 8.13), the performance of these agency (changes in the pacing frequency in this
timing patterns is highly susceptible to visual case), called a control parameter. 6 (5) The nonlinear
feedback influences (Mechsner, Kerzet Knoblich, phase shift was precipitated by a destabilization
& Prinz, 2001). of the anti-phase pattern during a critical period.
 Coordination !II 285

The general purpose of the HKB model was toward the middle of the chest for upper limb
to conceptualize these movement coordination movements in front of the body), and a directional
findings (and other, related findings) as a self- preference, in which synchrony of the timing of
organizing system, as in the way the physics of two limbs is in the same direction along any plane
chemical actions might be modeled. By self-orga- of motion. For bimanual tasks performed in front
nization, Haken and colleagues assumed that the of the body (e.g., tapping, finger wiggling, and
degrees of freedom involved in movement were our pinching examples), the egocentric preference
governed by patterns (called "collectives") that (in-phase pattern) dominates over the directional
were subject to destabilization and transitions. preference (anti-phase). However, a directional
Compared to more cognitively based models preference exists when one is moving an arm and
or theories of motor control, the key feature of a leg in the up-down direction. But research by
self-organization was a decentralization of the Swinnen and his colleagues has demonstrated
burden of how coordination is achieved. Accord- quite clearly that the relative dominance of these
ing to this view, the system organizes its degrees preferences is subject to spatial interactions
of freedom through preferences that undergo (Bogaerts & Swinnen, 2001; Serrien, Bogaerts, Suy,
transitions as an interaction with the environment & Swinnen, 1999; Swinnen, Jardin, Meulenbroek,
within which it performs. The role of cognition is Dounskaia, & Hofkens-van den Brandt, 1997;
diminished, especially as compared to that in con- Swinnen et al., 1998).
temporary motor control models. In other words, For example, in one study (Swinnen et al.,
self-organization is like an orchestra playing 1998), subjects moved their right and left limbs
without a conductor (Kelso & Engstmm, 2005). in various combinations either along the x-axis
One critical feature that sets the HKB model (left-right, in front of the body) or in the y-axis
(and theoretical approach) apart from many other direction (toward and away from the body). In
theories is the role of variability in motor control. this study, "in-phase" and "anti-phase" were
Recall the discussion of the laws of speed and defined in terms of an egocentric referent. Regard-
accuracy in chapter 7: Increases in variability less of the spatial trajectory of the limbs, in-phase
were associated with declines in the efficiency movements arrived at the egocentrically closest
(or effectiveness, or "skillfulness") of the motor point at the same time and at the point farthest
control system. In the HKB model, variability from egocenter at the same time; for anti-phase
is viewed as a characteristic of the current state coordination, one limb arrived at the closest
of the system-that is, instability gives rise to egocentric point at the same time that the other
change (Kelso, 1992), and change leads to a more limb arrived at the farthest point from egocenter.
stable state. As before, when both limbs were moving along
The HKB model has had an enormous impact the x-axis, in a parallel (180°) spatial orientation,
on coordination research, and in the ensuing the egocentric preference (in-phase) dominated
years hundreds of studies were published in performance. However, when one limb moved
which subjects performed continuous, cyclical along the x-axis and the other along the y-axis, in
tasks. Some of the studies described in the next an orthogonal (90°) spatial orientation, the normal
sections were motivated as direct tests of the HKB stability of these patterns was reversed-anti-
model, and other studies were performed as a phase was more stable than in-phase.
result of the research that it spawned in the ensu- These effects were extended by Lee, Almeida,
ing decades. However, it is important to note that and Chua (2002) and Welsh, Almeida, and Lee
a vast majority of this research has carried on the (2005), who used both the parallel and orthogo-
theoretical tradition to understand coordination nal spatial orientations as had Swinnen and col-
as a self-organizing process. leagues (1998); they also included obtuse angles
Spatial Orientation and Timing Patterns Swin- of spatial orientation (such as one limb moving
nen (2002) concluded that interlimb coordination along the x-axis and the other at a 120°or150° spa-
preferences can be summarized into two general tial orientation to the x-axis). When performance
categories: an egocentric preference, in which syn- was assessed in a high-oscillation frequency con-
chrony of the timing of two limbs is toward and dition, the stability of the patterns showed a clear
away from the center of any plane of motion (e.g., trend, as illustrated in figure 8.16. The shift in
286 Motor Control and Learning

To illustrate the role of intentions, let's go back

40 to our bimanual pinching experiment one more
time. Start by coordinating your pinching move-
Q)
ti)
ments in the anti-phase pattern at a relatively slow
ca
.c pace (say, 1 beat/s). At some point, try to switch
0. 30
~ to an in-phase pattern as quickly as possible. You
~ probably had little difficulty in doing so. Now,
~ trying doing the opposite-try to switch as rapidly
0 as possible from the in-phase pattern to the anti-
§ 20
~ phase pattern. Experiments of this kind typically
·:;: show that it takes longer to switch completely from
Q)
"C
"E the in-phase mode to the anti-phase mode than
.g 10 vice versa (e.g., Byblow, Lewis, Stinear, Austin, &
c
Lynch, 2000; Carson, Byblow, Abernethy, & Sum-
c7l
mers, 1996; Kelso, Scholz, & Schoner, 1988; Scholz
& Kelso, 1990; Serrien & Swinnen, 1999). Similar
differences between the two bimanual coordina-
Orthogonal Obtuse Parallel
Spatial orientation (degrees)
tion modes are seen when a mechanical device
is introduced that perturbs coordination: The in-
phase pattern restabilizes more quickly following
FIGURE 8.16 Effects of spatial orientation on relative-
phase variability in upper limb coordination.
a perturbation than the anti-phase pattern (Scholz,
With kind permission of Springer Science+Business Media: Experimental Brain Kelso, & Schoner, 1987).
Research, Spatial constraints in bimanual coordination: Influences of effector These findings have been interpreted within
orientation," 146, 2002, pgs. 205-212, T.D. Lee, Q.J. Almeida, and R. Chua.
self-organization theory as being consistent with
the unintended phase transitions discussed ear-
dominance between the egocentric (in-phase) and lier. The intention to switch coordination patterns
directional (anti-phase) preferences was mediated creates a destabilization of the current pattern,
by the spatial orientation of the limbs. Findings and that destabilization now facilitates the transi-
such as these (see also Amazeen, Amazeen, & tion to a new pattern. Since the in-phase pattern
Turvey, 1998; Carson, Riek, Smethurst, Parraga, is a stronger, more stable mode of coordination,
& Byblow, 2000; Salesse, Oullier, & Temprado, intentionally destabilizing this pattern is more
2005)-in which the motor system finds new "difficult" and takes more time than destabilizing
stable solutions when changed environmental the anti-phase pattern.
conditions have caused instabilities in the cur- The role of intention also highlights a peculiar-
rent pattern-have prompted some to suggest ity associated with some of the research on self-
modifications to the HKB model (Fuchs & Jirsa, organized coordination. In the early studies by
2000; Newell, Liu, & Mayer-Kress, 2008; Peper, Kelso (1984; Kelso et al., 1986), in which patterns
Ridderikhoff, Daffertshofer, & Beek, 2004). of motion were performed at specified oscilla-
Intention and Attention In the Kelso studies tion frequencies, subjects were instructed to let
described previously (see figure 8.15), the met- the hands "do what came naturally," and to not
ronome pacing frequency served as a control intervene intentionally if a pattern destabilized.
parameter to "perturb" the system (by forcing it From one perspective, this instruction is perfect
to move faster or slower); these changes resulted for the study of self-organization-letting the
in different effects depending on whether the system relax into a pattern that is best suited or
pattern was stable (in-phase) or unstable (anti- "natural" for the given frequency. From another
phase). As we mentioned before, the role of cog- perspective, however, one could argue that the
nition in self-organization theory is minimized. nonlinear phase transition from anti-phase to
But, according to the theory, intentions do have a in-phase occurred because the subjects' task had
specific role-to stabilize or destabilize a move- changed-that is, they were no longer intending
ment pattern in much the same way oscillation to perform the original pattern. If the goal of
frequency and spatial orientation did as described the task was to perform the anti-phase pattern
in the previous sections. as well as possible, then it is peculiar to use an
 Coordination oo 287

instructional procedure that allows the task goal stable than anti-phase patterns, which in turn are
to change midway through a trial. Indeed, results more stable than all other phase relations (e.g.,
of studies in which subjects were instructed to try Yamanishi et al., 1980). Perhaps not surprisingly,
to maintain the goal pattern at all times were quite researchers have found that a similar relation
different: There was an overall linear increase in exists if one merely perceives the phase relation
pattern variability, but not a permanent switch to of two objects in motion. For example, subjects
a stable new pattern (Lee, 2004; Lee et al., 1996; in a study by Zaal, Bingham, and Schmidt (2000)
Smethurst & Carson, 2003). watched two spheres oscillate on a computer
The role of intention in bimanual coordination screen, like two illuminated "balls" at the bottom
has also been examined in the more traditional of pendulums oscillating in a darkened room. On
sense of attention, as in experiments on mental each trial the balls could oscillate with a mean
workload (which was discussed at length in chap- relative phasing between 0° and 180°, with a
ter 4). One group of investigators has conducted variability ranging from 0° (perfectly stable) to
studies in which in-phase and anti-phase patterns 20° (highly unstable). So, for example, a 180°
were performed together with a secondary, probe relative-phase pattern with 0° of variability
RT task (chapter 4) (for reviews see Monno, Tem- would resemble the windshield wipers of many
prado, Zanone, & Laurent, 2002; Temprado, 2004). cars-moving in a perfectly locked asymmetric
As might be expected given the foregoing discus- coordination mode. In contrast, a pattern with 20°
sion, the performance of an in-phase bimanual of variability would show considerable random
pattern was relatively immune to the effects of a fluctuations in which the pattern deviated from
secondary task. In contrast, an anti-phase pattern perfect asymmetry, but overall would have a mean
was rather markedly destabilized if priority in the relative phase of 180°. Following an extensive
"division" of attention was given to the secondary training period with these stimuli, the subject
task. However, performance of an anti-phase task was simply to judge the pattern's mean relative
became more stable if increased cognitive effort phase (using a 0-10 Likert scale, with 0 denoting
was induced through moderate levels of anxiety symmetry and 10 asymmetry) or the pattern's
(Court, Bennett, Williams, & Davids, 2005). Thus, phase variability (0 = no variability and 10 =
the effects of attention (in terms of divided atten- highest variability).
tion) converge well with the role of intentions dis- Judgments of mean relative phase were
cussed earlier-attention can serve to perturb a accurate for all subjects-they could identify
stable system, and to maintain either a stable or a patterns accurately in symmetrical and asym-
destabilized coordination pattern. Inattention has metrical motion as well as the relative phase of
no effect on a highly stable pattern, but gives rise objects with relative phases between 0° and 180°.
to destabilization and pattern switching for less However, as illustrated in figure 8.17, perceptual
stable states (see Lee, 2004; Pellecchia, Shockley, judgments of variability were relatively inaccu-
& Turvey, 2005; and Shockley & Turvey, 2006 for rate, with some exceptions. For objects moving
further discussion on the roles of attention and in-phase (corresponding to 0° on the x-axis
intention in bimanual coordination). in the figure), subjects could accurately judge.
Perception and Bimanual Timing One of the when the pattern was performed with maximal
important advances in self-organization theory stability (filled square symbols )-as evidenced
has to do with the critical role played by infor- by the near-zero score, indicating a judgment of
mation, which is analogous to its role in other no variability. And, subjects perceived added
theoretical accounts of motor control. Specifi- instability quite accurately, too, with correspond-
cally, information is considered to have a strong ingly increased levels of judged variability. In
influence in coordination stability, arising from contrast, the anti-phase pattern (corresponding
intrinsic-feedback sources as well as perception to 180° on the x-axis in the figure) was judged
of the environment. In this section we review evi- to have a moderate level of variability (values
dence suggesting that perceptual information is around two or three units), regardless of how
used to stabilize coordination as well as perturb it. much actual variability was present. Even so,
As we have discussed often in the previous the anti-phase pattern was perceived to be more
sections, simple bimanual coordination patterns stable than all of the other phase relations exam-
differ in stability-in-phase patterns are more ined in the experiment; in general, the anti-phase
288 Motor Control and Learning

One implication of the findings of Zaal and col-

10 leagues is that the ability to perceive relative phase
9
accurately may contribute significantly to one's
ability to perform coordinated movements accu-
8
rately. This implication is supported by results
7 from a study using a related research strategy in
i:
Q) which subjects attempted to coordinate move-
E 6
en ments of one limb with an external object, using
"t:I
.a 5
c:
perceptual judgments as the basis for stabilizing
ca
Q)
4 coordination. Subjects in an experiment by Wim-
a
3 mers, Beek, and van Wieringen (1992) coordi-
2
nated left and right movements of a lever with
a visual metronome that oscillated horizontally
across a monitor. The findings revealed many of
the same self-organizing properties that had been
0 30 60 90 120 150 180
Relative phase (degrees)
observed in studies of bimanual coordination dis-
cussed earlier (e.g., Kelso, 1984; Kelso et al., 1986).
Very similar effects have been observed when
FIGURE 8.17 Judgments of perceived variability for
two objects in motion with different mean relative phases a subject coordinated his or her movements with
(denoted on the x-axis) and levels of variability (corre- another person rather than with an inanimate
sponding to the symbols in the key). Mean judgment was object. For instance, Schmidt, Carello, and Turvey
scored on a Likert scale, where 0 represented no variability (1990) asked subjects to swing one leg in temporal
and 10 represented maximal variability.
Reprinted, by permission, from F.T.J.M. Zaal, G.P. Bingham, and R.C. Schmidt,
coordination with another person, who was also
2000, "Visual perception of mean relative phase and phase variability,'' Journal of swinging one leg (see figure 8.18). The mutual goal
Experimental Psychology: Human Perception & Performance 26: 1209-1220.
between the two subjects resembled the (within
subject) bimanual finger-wiggling task used by
pattern received lower judgments of perceived Kelso (1984). The findings were similar; the anti-
variability (see Wilson & Bingham, 2008, for a phase pattern was more variable and difficult to
review of this research). maintain at high movement frequencies than the

I
I
I
I

C
I

l-,--(
:
I I
I \
'-- ...... .._!

FIGURE 8.18 The between-person coordination task.

Reprinted, by permission, from R.C. Schmidt, C. Carello, and M.T. Turvey, 1990, "Phase transitions and critical fluctuations in the visual coordination of rhythmic movements
between people,'' Journal of Experimental Psychology: Human Perception and Performance 16: 229. Copyright© 1990 by the American Psychological Association.
 Coordination 1\1 289

in-phase pattern (Amazeen, Schmidt, & Turvey, was perceived to be in-phase was always the more
1995; Schmidt, Carello, & Turvey, 1990; Temprado stable pattern. Mechsner's findings, together with
& Laurent, 2004). The importance of these findings the information presented earlier about the role
is that a similar set of effects emerged when these of perception (see also Franz, Zelaznik, Swin-
movements were coordinated, even though each nen, & Walter, 2001; Semjen & Ivry, 2001; Wilson,
individual movement was controlled by a separate Bingham, & Craig, 2003), highlight the key role of
nervous system. (We will have more to say about visual information in stabilizing and destabilizing
such social coordination later in this chapter.) coordination patterns.
The importance of perception in coordina-
tion was probably revealed most spectacularly Upper and Lower Limb Coordination
in experiments by Mechsner and colleagues Common activities such as driving, sewing, and
(Mechsner, 2004; Mechsner et al., 2001; Mechsner playing musical instruments (e.g., the piano and
& Knoblich, 2004). These researchers conducted drums) require that we coordinate the timing of
bimanual timing experiments in which the sub- our hands and arms with foot and leg movements.
jects' visual feedback of their upper limbs was In many of these coordinated actions, the domi-
manipulated. For example, figure 8.19 illustrates nant pattern of movement is a directional preference
one experiment in which subjects moved two (both limbs move in the same direction), not the
hand-wheels either in-phase or in anti-phase. egocentric preference (limbs move toward and
Instead of viewing these movements directly, the away from the center of a plane of motion) that has
subjects viewed "flags" moving in patterns that been observed for many bimanual coordination
were either compatible with the direction of their patterns (Swinnen, 2002). Many of these findings
movements or in opposition to the actual move- were shown in early experiments by Baldissera
ment patterns. That is, sometimes how they saw and colleagues (1982, 1991; Baldissera, Cavallari,
their limbs apparently move was the mirror image & Tesio, 1994; also Carson, Goodman, Kelso, &
of how the limbs were actually moving. Mechsner Elliott, 1995). For example, subjects in Baldissera
and colleagues (2001) obtained a startling result: and colleagues' (1982) study were asked to coor-
The actually-produced anti-phase pattern that dinate ankle movements in the upward direction
(dorsal flexion) or downward direction (plantar
flexion) with specific combinations of wrist move-
ments. When the forearm was fixed in the supine
position (palm of the hand facing up), movements
that were coordinated in the same direction (i.e.,
plantar flexion with wrist extension and dorsal
flexion with wrist flexion) were more stable than
actions coordinated in the opposite direction.
However, with the forearm immobilized in the
prone position (palm facing down), the stronger
coordination mode was observed with the opposite
pairing of muscle groups: Plantar flexion was now
more strongly related to wrist flexion, and dorsal
flexion to wrist extension. The common finding
among these limb pairings was that the stronger
FIGURE 8.19 Experimental setup used by Mechsner
coordination modes occurred for movements in
and colleagues (2001 ). Subjects moved hands in pat- the same direction, and regardless of the pairings of
terns of in-phase and anti-phase but could only see flags flexion-flexion or flexion-extension. The evidence
attached to the handles. Typical coordination effects were for the differential strength of these particular
found when the flags were directionally compatible with the
hand movements. However, when one flag moved in the preferred coordination modes was similar to that
mirror-image direction to the movement of the hand, the for the bimanual modes discussed previously;
typical coordination effects were reversed. the weaker coordination pattern showed higher
Reprinted by permission from Macmillan Publishers Ltd: Nature, F. Mechsner et al.,
"Perceptual basis of bimanual coordination:' 414: 69-73, copyright 2001. http://
relative-phase variability and frequent unintended
www.nature.com/nature~ndex.html transitions to the stronger (same direction) pattern.
290 llJ1 Motor Control and Learning

For the bimanual patterns discussed previ-

ously, one could argue that the in-phase bimanual
pattern was more stable because it involved the
timing of similar muscular groups (simultane-
ous flexion and extension), since movements
in the same direction (i.e., anti-phase) were less
strongly coordinated than were movements in
the opposite direction. Such an argument fails to
explain the findings of Baldissera and colleagues,
however, because the coordination strength of the
ankle movements with wrist flexion-extension
could be reversed by simply changing the fore-
arm orientation (i.e., prone or supine). A similar
effect of spatial orientation can be seen in the
coordination of the wrist and elbow movements
within a single arm (Buchanan & Kelso, 1993;
Kelso, Buchanan, & Wallace, 1991). This evidence
suggests that the pattern of findings for the sta-
bility of the different coordination modes is not
dependent on the specific muscle groups used,
but rather on the spatial orientation of the actions.
The effect of spatial orientation in coordinating
two limbs is illustrated quite well in studies by
FIGURE 8.20 Multilimb coordination apparatus.
Kelso and Jeka (1992) and Serrien and Swinnen Reprinted from Human Movement Science, Vol. 12, J.J. Jeka, J.A.S. Kelso, and T.
(1997a, 1997b). Consider various pairs of limb Kiemel, "Spontaneous transitions and symmetry: Pattern dynamics in human
lour-limb coordination;• pg. 635, Copyright 1993, with kind permission of Elsevier.
movements that could be produced in the sagittal
plane, as illustrated in figure 8.20. Three different
types of interlimb pairings can be produced: those earlier. Thus, these data strongly imply that
involving homologous (same) limb pairs (left and the rules of movement coordination depend
right arms; left and right legs), ipsilateral (same on a number of factors, including the effectors
side) limb pairs (left leg and arm; right leg and involved, their orientation and planes of motion,
arm), or contralateral (diagonal) limb pairs (left interactions with the physical environment, and
leg and right arm; left arm and right leg). These intended goals.
three types of coordination patterns were exam-
ined by Kelso and Jeka (1992) under conditions Complex Timing Patterns
in which the limbs moved either in the same Try the following easy task. Tap the index fingers
direction or in the opposite direction. The results of both your hands simultaneously on a flat sur-
are illustrated in figure 8.21. In general, the limbs face. Now, make two taps of the right hand for
moving in the same direction demonstrated more every single tap of the left hand (i.e., taps of the
stable patterns than the limbs moving in opposite left hand coincide with every second tap of the
directions. But this was true only for upper and right hand-termed a 2:1 rhythm). Then do 3:1
lower limb combinations, which were more stable and 4:1 rhythms; all of these should be easy to
for contralateral pairs than for ipsilateral pairs, do when the two hands are performing the same
especially so when the limbs moved in opposite rhythm, that is, when the rhythms are harmonic,
directions (see also Swinnen, Dounskaia, Ver- or integer related (multiples of each other; e.g.,
schueren, Serrien, & Daelman, 1995). Farnsworth & Poynter, 1931; Lashley, 1951).
Again, notice that for this coordination task Temporal coordination becomes much more
(sagittal plane), movement of the two limbs in difficult when the rhythms are not harmonic.
the same direction was as stable as when they For instance, it is very "difficult" to maintain a
moved in the opposite direction. This effect is rhythm with one hand while doing something
quite different from that found for bimanual else as rapidly as possible, such as tapping with
movements in the frontal plane, as we discussed the other hand (Klapp, 1979) or speaking a syl-
 Coordination WI 291

Recent studies have focused on nonharmonic

rhythms, or polyrhythms, to identify what makes
80
certain timing patterns so "difficult" to produce.
70
Consider a situation in which one hand produces
three beats during a given interval and the other
i 60 hand produces two beats during the same time
I!! interval (a 3:2 polyrhythm). For example, if the
Cl
CD
~ 50 time interval is 1,200 ms, the faster hand pro-
0
en duces a beat every 400 ms and the slower hand
CD 40 produces a beat every 600 ms. The two beats
!:!
.c occur simultaneously only once every 1,200 ms .
C2.
g! 30 Although this 3:2 polyrhythm is much harder
:; to perform than any of the harmonic combina-
'ii
a: 20 tions, it is easier to perform than 5:2, 4:3, 5:3, and
5:4 polyrhythms, which are progressively more
10
difficult (Deutsch, 1983; Summers, Rosenbaum,
Bums, & Ford, 1993). What is the basis for these
Homologous lpsilateral Contralateral control problems? 7
Limb pair Figure 8.22 presents data from an experiment
by Summers and colleagues (1993), who com-
FIGURE 8.21 Standard deviation (SD) of relative phase pared the coefficients of variation (SD divided by
for various interlimb pairings. the mean, expressed as a percentage) for various
Data from Kelso and Jeka 1992.
polyrhythms as a function of hand speed and
musical training. What is evident from this figure
lable (Klapp, 1981). Even more "difficult" is is that nonmusicians were most variable in timing
the situation in which two concurrent activities the beats of the hand that was moving more slowly
have their own rhythms; for example, reciting a in these polyrhythms. Summers and colleagues
nursery rhyme while tapping a rhythm with a (1993; Summers & Pressing, 1994) explained these
different cadence (Peters, 1977). findings by considering the faster hand's rhythm

16

14

l 12
c
0
:;:: 10
Ill
·;:

-
Ill
>
0
8
'E
CD
·u 6
=CD
0
0 4

Musician Musician Nonmusician Non musician

fast hand slow hand fast hand slow hand

FIGURE 8.22 Variability in temporal coordination for polyrhythms of increasing difficulty in musicians and nonmusicians.
Data from Summers et al. 1993.
292 Motor Control and Learning

as the baseline rhythm. The task of coordinat- be effective (Bogacz, 2005; Kovacs, Buchanan, &
ing the polyrhythm involved interspersing the Shea, 2010; Summers, 2002).
slower beats at specific intervals in between the
beats of the faster hand. The bars to the right Spatial Coordination
side of figure 8.22 support the conclusion that Spatial biasing can be observed quite readily in
the variability in performing a polyrhythm is the "pat the head while rubbing the stomach"
positively associated with the number of beats that example ("Costs and Benefits of Coordinating
the slower hand needs to intersperse between Two Limbs" on p. 279). Someone trying to do this
beats of the faster hand (e.g., 5:2 vs. 5:3 vs. 5:4; task tends to bias the spatial trajectories of both
Deutsch, 1983). Moreover, such a strategy is per- limbs: The limbs are drawn toward performing
formed much less effectively by nonmusicians one or the other task, or sometimes a novel combi-
because they tend to intersperse the slow beats nation of the two tasks. An experimental version
at approximately 50% of the interval duration of this coordination task by Franz, Zelaznik, and
between beats of the faster hand (Summers et McCabe (1991) suggests that a novel combination
al., 1993). This unskilled strategy is rudimen- may be the more natural coordination pattern.
tary, although it makes sense because it copes Here the subjects' task was to draw circles and
effectively with this situation by drawing upon lines. In some conditions, a single hand drew only
an anti-phase coordination pattern-which, as one pattern; in other conditions, both hands drew
we have seen, is a relatively natural pattern to the same pattern; and in another set of conditions,
adopt. Other strategies can be used, although one hand drew a line while the other hand drew a
they require considerable practice in order to circle. The results from these conditions are illus-

Single hand Dual hand, Dual hand,

same pattern different patterns

Single hand Dual hand, Dual hand,

same pattern different patterns

FIGURE 8.23 Drawing lines (upper) and circles (lower) under unimanual and bimanual conditions.
Reprinted from Acta Psycho/ogica, Vol. 77, E.A. Franz, H.N. Zelaznik, and G. McCabe, "Spatial topological constraints in a bimanual task:' pgs. 142 and 143, Copyright 1991, with
kind permission of Elsevier.
 Coordination 1111 293

trated in sample trials represented in figure 8.23. workplace (e.g., when several people join forces
Both lines and circles were drawn accurately and to move a heavy object). By "explicit" we mean
consistently in the single-hand and dual-hand that the actions of two or more people are cho-
"same" conditions. However, when the hands reographed with a specific objective of creating
drew different patterns, the variability increased something together that could not be achieved
dramatically; the circles became more linear, and individually, such as the effect on the audience of
the lines became somewhat circular. a ballet duet or the impression made on a judge
One recent study suggests that part of the prob- by a synchronized dive. Predetermined "scripts"
lem encountered during performance of different that describe the unfolding of a play in football
spatial tasks with the two hands lies in conceptu- or a duet in ballet often require considerable
alizing the nature of the two tasks. Franz and col- practice before an acceptable level of coordination
leagues (2001) asked subjects to draw semicircles, is achieved among individuals such as athletes,
one above the other, so that one finger drew a actors, dancers, and musicians. In many respects,
semicircle on top and another finger drew one on social coordination in these situations mirrors the
the bottom. As expected, the bimanual task was concept of a GMP, as plans for action that describe
easy to perform when both hands drew "tr::::,." the order and timing of events remain invariant,
-shaped semicircles and when both hands drew with the parameters (e.g., the specific participant)
"~"-shaped semicircles. However, the authors substituted as needed. How this kind of social
found a dissociation in performance when differ- coordination is achieved in teams represents an
ent semicircle orientations were drawn. The task important research area in sport psychology (see
was easy to do when one hand was required to Eccles & Tenenbaum, 2004).
draw a "tr::::,." on top and a "~" below it; how- Motor control researchers have tended to focus
ever, the task was nearly impossible to do in the on situations in which coordinated actions arise
reverse configuration. One possibility is that in from mutual goals that are implicit-situations in
the former case the spatial representation drawn which coordination emerges spontaneously when
with the two hands is a full circle, but in the latter each person is assigned an individual goal. This is
it is a nonidentifiable symbol. The dissociation in more analogous to a self-organizing system, as we
performance in these two cases occurred despite have discussed previously, because coordination
the fact that the spatial coordinates to produce the arises despite the fact that no intention or goal
two tasks are similar. Perhaps the contribution of related to coordination was specified. The nature
of the coordination also differs from what we have
a conceptual model provided a strategy that was
discussed in previous sections; instead of two
sufficient to reduce the interference inherent in
(or more) limbs becoming locked in a particular
the task. Or perhaps this effect is related to the
coordination mode (called absolute coordination
strong influences of perception on performance
by von Holst, 1937/1973), unintentional coordi-
that we discussed earlier. Much more study of
nation between people is characterized as a type
spatial coordination effects is needed in order to
of relative coordination-the tendency to perform
establish firm principles (see Chan & Chan, 1995;
with a relative phase near a preferred coordina-
Franz, 1997; Franz, Eliassen, Ivry, & Gazzaniga,
tion mode(s), but without being phase locked for
1996; Spijkers & Heuer, 1995; Swinnen, Jardin, &
extended periods of time (von Holst, 1937 /1973).
Meulenbroek, 1996).
For example, two people sitting in rocking
chairs, with no explicit goal to coordinate their
Social Coordination
rocking, tended to do so near an in-phase pat-
Although interlimb coordination is organized by tern about 30% to 70% of the time (depending
the pathways within the CNS (Swinnen, 2002; on the physical similarities of the two chairs), but
Swinnen, Heuer, Massion, & Casaer, 1994), this only if the subjects were looking directly at each
cannot be the case when coordinated actions other (Richardson, Marsh, Isenhower, Goodman,
occur between two (or more) people. Coordi- & Schmidt, 2007; see also Schmidt & O'Brien,
nated activities between people often have a 1997). A greatly reduced tendency toward an in-
well-defined, explicit mutual goal (Schmidt, Chris- phase pattern was found when only peripheral
tianson, Carello, & Baron, 1994), as in music, in vision was available, and relative coordination
sport (e.g., synchronized swimming), and in the was eliminated altogether when vision was
294 Motor Control and Learning

occluded. Thus, in this task, relative coordina- on temporal versus spatial factors during move-
tion was dependent on the availability of direct ment planning.
visual information (for a review see Schmidt & Discrete bimanual actions have also been an
Richardson, 2008). important focus of study. Studies using a two-
But visual information is not the only medium hand version of the Fitts task reveal consider-
that facilitates social coordination: A conversa- able influences of one limb's movement on the
tion between two individuals results in mutual, spatial and temporal actions of the other limb.
socially coordinated interactions involving the These influences are even more pervasive if the
speaking patterns of the two individuals (see actions of one limb have increased kinematic
Fowler, Richardson, Marsh, & Shockley, 2008, complexities. Recent evidence suggests that
for a review). Interestingly, shared postural coor- the development of a new GMP that controls
dination between two individuals was highest the two limbs simultaneously may be a way in
when they were engaged in conversation and which the motor system solves these coordina-
was influenced by how they spoke to each other tion difficulties.
(Shockley, Baker, Richardson, & Fowler, 2007), but The coordination of continuous, cyclical
was not dependent on whether or not the indi- actions represents a contrast to that of discrete
viduals could see each other (Shockley, Santana, actions; in the latter, the GMP seems to play an
& Fowler, 2003). important role. Some continuous interlimb coor-
Coordination can also be studied as a mutual dination patterns are more stable than others, and
collaboration between more than just a pair of indi- these stable states are subject to change. Impor-
viduals, although such a study is usually designed tant insights regarding pattern organization are
to investigate explicit cooperation among mem- revealed by phase transitions, in which the desta-
bers of a team, as occurs in sport or the workplace bilization of a pattern leads to a change in the
(Eccles & Tenenbaum, 2004). For example, the basic form of the pattern. These transitions have
spontaneous applause that occurs among mem- been shown in animals and in humans, leading
bers of an audience following a concert reveals some to suggest that these patterns have a strong
periods of synchronicity (Neda, Ravasz, Brechet, self-organizing basis, as conceptualized in the HKB
Vicsek, & Barabasi, 2000). We anticipate that more model (Haken, Kelso, & Bunz, 1985). Evidence
complex forms of unintentional coordination will from a number of bimanual and interlimb coor-
be studied in years to come. dination paradigms has provided support for
this view. Other patterns of coordination, such
as polyrhythm timing, spatial coordination, and
Summary patterns of social coordination, have been the
focus of recent investigations and pose exciting
Even quite simple movements require the orga- challenges for future research.
nization of various (potentially independent)
moving parts of the motor system. Aiming move-
ments, though simple at one level, involve the Student Assignments
coordination of the eyes, head, and hand when
perception and action are guided visually. One 1. Prepare to answer the following questions
method of analyzing the organization of discrete during class discussion:
tasks is to search for separate units of action-
a. Describe the coordination components
parts of action sequences that are independent
of reaching for and opening a jar of
temporally from other parts of the action. In
peanut butter.
contrast, reaching/ grasping represents a type of
discrete movement in which the limb-transport b. Describe two different methods by
component is quite obviously distinct from that which an artificial limb is used to reach
of object manipulation. Evidence suggests that, for and grasp objects.
although these components seem to be separable, c. Describe any daily activities that char-
their actions are highly interdependent. Two acterize in-phase and anti-phase move-
theoretical views of reaching and grasping have ments. Describe an activity that either
been suggested; these differ in terms of the focus combines in-phase and anti-phase
 Coordination Iii 295

movements or that characteristically to the extent to which the hands are coordinated together
has an alternative phasing. in time. The covariance-ratio data are consistent with the
view that the two hands are controlled by a single GMP,
2. Find a research article that uses the Haken- but details of this analysis are beyond the scope of this
Kelso-Bunz model as a theoretical basis. Be presentation.
prepared to explain these concepts using 3 The Froude number was discovered in the 19th century
examples from daily activities. by a mathematician and naval architect (William Froude)
as a way to characterize the efficiency of boats of differ-
ent hull lengths. An excellent historical review of the
Web Resources application of the Froude number to gait biomechanics
of animals, and other uses, is provided by Vaughn and
Motor control in zero gravity: O'Malley (2005).
4 The measure of accuracy presented here is different
http://mvl.mit.edu/ASTRODYN/html/astrodyn.
from that presented by Yamanishi and colleagues (1980).
html They reported constant error, averaged over subjects.
More on the Haken-Kelso-Bunz model: However, as noted in chapter 2, the average constant
errors for subjects who are biased differently can under-
www.scholarpedia.org/article/Haken-Kelso-Bunz_ estimate the average inaccuracy for a group-and these
model individual differences, in opposite directions of bias, were
evident in the data from Yamanishi and colleagues. We
have replotted their data in terms of absolute constant
Notes error (chapter 2). The results show much the same pattern
as the variability data.
1 Different taxonomies have been proposed that extend 5 Note that mean relative phase is plotted in figure
Napier's (1956) original classification scheme of precision 8.l5a, rather than absolute CE as in figure 8.14a. A rough
grips versus power grips. However, the choice of which equalization would require a subtraction of 180° from
grip to use is determined not by the object but rather by each anti-phase value in figure 8.15a, then a conversion
the intentions of the individual regarding how the object to a positive value.
will be used (Jeannerod, 1996; Marteniuk, MacKenzie, 6 Use of this term can be confusing here, as we have

Jeannerod, Athenes, & Dugas, 1987). For instance, one defined the word" control," as in motor control, much dif-
normally uses a precision grip to write with a pen. How- ferently than as it is used in the term" control parameter."
ever, if one were intending to use the pen to puncture a As well, earlier we discussed parameters of a GMP (e.g.,
cardboard box, use of a power grip would be more likely. handwriting in large or small script). A control parameter
2 This work extends the pioneering research of Wing is defined as a nonspecific variable that, when changed,
and Kristofferson (1973a, 1973b), which was discussed results in a nonlinear change in the behavior of the system
in chapter 7. Analyses of the bimanual coordination data as a whole.
here were done with nonoverlapping intervals among the 7 Interestingly, the task used by Heuer and colleagues

landmarks, which eliminates the bias in the correlations (1995), as described earlier, was a rapid, 3:2 "polyrhythm,"
because of overlapping of some of the landmarks. Heuer which was performed rather easily (after practice). That
and colleagues (1995; Schmidt et al., 1998) have developed discrete and continuous versions of a similar task might
a theory of bimanual coordination based on GMPs that have remarkably different effects on coordination perfor-
uses a statistic termed a covariance ratio, which is sensitive mance is deserving of more research.
 TER . . 9

hy is one person a better gymnast than Experimental Versus

another, even after the same amount of
practice? What are the abilities or aptitudes that Differential Approaches
contribute to success as a skilled woodworker?
How many basic, inherited motor capabilities There are a number of fundamental differences
do humans possess, what are they, and how between the ways in which motor behavior is
can they be measured? These are just some of studied and understood in the experimental
the questions considered by researchers who and differential approaches. In the following
investigate individual differences-the study of discussion, we focus on just two of the major
factors that make individuals different from one differences-the scientific goals and the scientific
another. The approach to motor behavior rep- methods used by the two approaches.
resented by this chapter is a marked departure
from the approaches in the previous chapters. Different Scientific Goals
With the earlier approach, the concern was The most obvious difference between the experi-
for the effect of certain independent variables mental and differential approaches lies in the
on certain other dependent variables, using goals of these two traditions. The experimental
the mean (usually) of a group of people as the approach is concerned primarily with under-
measure of primary interest. With individual standing the effects of certain independent vari-
differences, however, the concern is for how ables (e.g., the target size of an aimed movement)
the individuals within a group differ from each on some dependent variable (e.g., the accuracy of
other. Because these two scientific traditions the movement). In this example, the interest is in
are so different in method and goal, they are how aimed movements are regulated in humans
usually treated as separate points of view. We generally. A fundamental belief is that humans
examine more closely some of the differences are not really very different from one another
between these two approaches in the next sec- (especially when human-to-human differences
tions. are contrasted with human-to-giraffe differences,

!11 297
298 Motor Control and Learning

for example); so when the subjects in a group are among people rather than (b) general phenomena
treated alike, one can estimate the behavior of a that are seen in the "average" person. As might
"typical" human by considering the effect of the be expected, such differences in goals naturally
target size on average accuracy. Through experi- create differences between the points of view
ments the researcher can hope to arrive at state- of the differentialists and the experimentalists.
ments such as "decreasing the size of the target Indeed, such differences have become so great
will require humans to slow their movement in that the two groups of psychologists are almost
order to maintain accuracy," as Fitts (1954) did, totally separated, with separate methods of doing
to cite just one example. research, statistical designs, goals, textbooks,
In these kinds of experiments, there is little con- and scientific journals. These differences were
cern regarding any one individual in the group. described, and decried, by Cronbach (1957) in
Almost no interest is shown in the possibility that his article titled "The Two Disciplines of Scientific
one person might use a different strategy than Psychology" (see also Underwood, 1975).
another or have more skill than another. These
factors are usually averaged out and are rarely Different Scientific Methodologies
seen in the group data. In fact, if some individual The experimentalists and the differentialists
differs "too much" from the mean behavior answer their respective questions quite differ-
level of the group (as determined by a statistical ently. While we have already devoted consider-
procedure), common experimental practice is able space to the experimental approach, a brief
to remove this outlier from consideration. Thus, review of its characteristics will help to make clear
variations among people are considered "noise," how it contrasts with the differential approach.
or a nuisance, and many methods are available to
eliminate or control such between-subject varia- Experimental Methods
tions in experiments. Essentially, the "true" experimental method
With the differential, or individual-differences, involves the manipulation (or artificial variation)
approach, on the other hand, the primary focus is of some independent variable while as many
on the differences between or among individuals. other variables as possible are held constant. This
Thus, many of the things that the experimental- can be done by administering one level of the
ist considers "noise" in experiments are the very independent variable to one group of people and
things that the differentialist considers interesting another level to another group, and noting the
and worthy of study! Generally, the differential differences in some dependent variable (called
approach deals with two basic issues. a between-subjects or between-groups design).
First, concern is directed toward the nature of Sometimes only one group of people is used
the underlying abilities (or capabilities), the ways (called a within-subjects design). One level of
in which these abilities differ in "strength" in the independent variable is administered to the
different people because of genetic variations or group at one time and the other is administered
experience, and the ways in which different tasks at another time, with differences between the
are interrelated. For example, are tasks involving two times (in terms of the dependent variable)
strength related to other tasks involving accuracy being the chief comparison of interest. We have
. in motion? Does high strength performance on a discussed many examples of both of these designs
strength task imply that the person will have low in previous chapters.
accuracy performance on some other task requir- In such experiments, the critical comparison is
ing accuracy? A second problem, closely related usually between the means of groups of people
to the first, entails prediction, or the estimation of (or of a single group tested under two different
performance in one situation based on measure- conditions). Typically, no regard is given to the
ments taken in some other situation. For example, variations among people within the group, except
how do intelligence test scores relate to success for the usual reporting of statistics that describe
in graduate school? Or, how does height relate to the extent to which people differed. Such disper-
success in gymnastics? sion statistics are rarely the primary concern in
The differential approach deviates from experiments; rather, they are included to ensure
the experimental approach in that it concerns that the variations among people were not so
attempts to explain and predict (a) differences large as to obscure or change the conclusions
 Origins of Expertise
We often wonder at how some elite performer in some activity or occupation got to>be thatwaY.We<
are tempted to say that such an individual was fortunate to have Men born with the "right''ablllMs
(i.e., those needed for performance in his specialty), coupled, of course,. with suffici~nt P:relcilce
and training. Of these two factors-abilities versus practice-it is our impression thatmosfpeClple
seem to place far more credence in genetically determined abilities as the major deterrnir)aqts qt..•
elite-level performance. . .
Such a view is severely challenged in a study by Canadian psychologist Roger Barn,$1ey (Barn,.
sley, Thompson, & Legault, 1992) and others (Dudink, 1994, in Dutch soccer; Helsel'l, $tarkes1 &
a
van Winckel, 2000, in Belgian soccer; see Musch & Grondin, 2001, for review). Here.• itis argued ·
that genius (or elite-level performance, or expertise) is not so heavily dependef\ton inheritedabifi"
ties, but rather is the product of various important factors that are effective almost.by chance alOl'le:
(This is all nicely summarized, with more examples, in Gladwell's [2008] book Out/iel'S;.Which WfJ·
recommend strongly; see also Gladwell's The Tipping Point for more.) This view. is that the< elite
performer-rather than being gifted with an extraordinary collection of the "right'' a~ittties_.,_simply
happened to be lucky or fortunate enough to exist in an environment where encouragement, cpach:. ··
ing, or teaching (or more than one of these), the opportunity for practice and experiehce, anel9:ther ·
factors all converged to make this person great. · ··
There are many examples. In one of these (Barnsley et al., 1992), Canadian Major JuniorAJeague .
ice hockey players (who are about high school age) and professional ice hockey players were studied, •
{It almost goes without saying that Canadian ice hockey is played at an extremely: higt) level, and ·
that the Canadians have a highly structured youth hockey system designed to identify and protriqte
promising young players.) Barnsley and colleagues examined the rosters of several Canadian Major
Junior A teams, as well as those of several professional hockey teams. These rosters includeimany ·
statistics such as each player's home town, his right- or left-handedness, his year and mqnth of ;
birth, his height and weight, and so on. Barnsley and colleagues noticed that almost every:play:erat ·
the Major Junior A level and at the professional level was born in the early months of the celehdar
year (January, February, March, April), and that almost no players were born in the. later .months
(September, October, November, December). This was a very large, and clear, effect. Otherth~A
by resorting to some bizarre astrological explanation, how could one explain why the monthofoirtli
should have any effect at all on hockey performance in these situations? .
One interpretation of this finding was as follows. When these elite players were kids, the ma.jprity:
played age-group hockey. In order to be on the 8-year-old team, for example, the player needed tcf
have his eighth birthday during the calendar year (January to December) in question. This meant
that a player having his birthday in January could be almost one full year older than a player borl"f
in December of that year. This has been termed the "relative age effect:' in thatthose otrt~:e. team
born in January are "relatively older'' than those born in December. Of course, eJ:>peciatly aithe
younger ages, older kids are bigger, faster, better skaters, and so on than younger kidsi andx~her .>;
year out of eight years can result in a huge difference in skill levels. Therefore, as the hypothesis ·. ·•
goes, the older players on the age-group teams (who are bigger, faster, etc.) attracted. the attention ·.·.. ··.
of the coaches; the coaches then gave the early-month players more coaching, encouragement,
practice and playing time, and so on as compared to the relatively younger kids. This. bias was then
repeated in each subsequent year. So, according to this argument, after several years of pref~r~ntial
treatment, the relatively older kids tended to be playing at the highest levels, while the relatlv~ly.
younger kids were not. This is a kind of "rich get richer'' effect.
According to this view, elite performers are not so "special" in terms of their abilities; rather, they
just happened to be "lucky" enough to have been born in the "right" month .. Of.course,. this d9es..
not say that the birth month is the only factor operating to produce elite performers, as ot~er abili-
ties discussed in this chapter are probably also important. This. evidence and interpretation slf11ply
reminds us that inherited abilities are not as critically important as we might have once thought.

WI 299
300 && Motor Control and Learning

drawn about the differences between means. a procedure is a study of the relationship between
Finally, the conclusions that come from these age and throwing accuracy. Such variables (age,
experiments are usually stated in cause-and-effect in this example) are usually called individual-
terms-the variations in the independent variable difference variables. Thus, studying which indi-
caused the changes in the dependent variable. As vidual-difference variables are related to certain
such, experimental methods provide relatively kinds of performances is a primary concern of
powerful ways of coming to an understanding the differential approach. Indeed, textbooks have
of one's scientific area. been written solely about individual-difference
variables, such as life span motor development
Differential Methods (e.g., Haywood & Getchell, 2009; Piek, 2006),
The differential methods contrast starkly with aging (e.g., Spirduso, Francis, & MacRae, 2005),
those just described, relying substantially on cor- individuals with various movement disorders
relational (or associational) techniques whereby (e.g., Weeks, Chua, & Elliott, 2000), and expert-
the relationships between or among variables novice differences in sport skill (Farrow, Baker,
are studied. In its simplest form, the differential & MacMahon, 2008; Starkes & Ericsson, 2003).
approach uses one group of people and at least Some of these individual-difference variables
two tests measured on each individual. (Remem- will be discussed in more detail toward the end
ber that the simplest "true" experiment often of this chapter (see also "Origins of Expertise").
uses at least two groups of people and one test, With the differential approach, conclusions
or dependent variable.) The primary concern is about the results tend to be phrased in lan-
the extent to which one test (e.g., height) relates guage quite unlike that with the experimental
to another test (e.g., accuracy) in the same people, approach. In experiments, one might conclude
with the nature of the relationship being deter- that the independent variable caused changes in
mined by the size and sign of a statistic called the the dependent variable (because other variables
correlation coefficient (a discussion of correlation were held constant or "controlled"). In differen-
statistics appears in chapter 2). With these corre- tial studies, however, causation can seldom be
lational methods, the chief concern is the relation- inferred logically. The major reason is that many
ship between the two tests, or among several tests other things may differ, or intervene in the causal
if more than two are used. Sometimes the relation- nature of the relationship. For example, consider
ship is computed between a group of tests (called the relationship between children's height and
a test "battery") and some other single measure. their throwing accuracy. Since weight is usually
An example is the relationship between a fitness associated with height, one cannot be certain
battery (consisting of five subtests) and some that a relationship between height and accuracy
other measure, such as probability of becoming is really not due to the relation between weight
a successful firefighter. and accuracy. Also, taller people are usually older
A second major method of individual- (if one is considering children), and one could
differences research uses essentially the same easily confuse the height-accuracy dependency
logic, but the appearance of the procedures may with an age-accuracy dependency. The primary
make it seem that the methods are experimental. limitation in these studies is that the level of the
Consider a study to determine the relationship variable of concern is not manipulated (artificially
between age and throwing accuracy. Typically the determined by the experimenter). Rather, the
researcher chooses one group of people at one age variable is allowed to vary naturally, and the
and another group at another age and compares scientist measures its value and attempts to
the group means on some "dependent" variable, understand how it relates to some other variable
such as throwing accuracy. This appears to be an that is also varying naturally. Such procedures are
experiment, because there are two groups and one often called natural experiments.
dependent variable and the focus is on the group People can differ from each other in at least
means. But it is not really experimental, because two fundamental ways. First, two people might
the level of the independent variable (age) is be fundamentally and consistently different from
not manipulated by the experimenter; that is, the each other in some stable characteristic, such as
ages of the people in the group were already height. Such differences will be enduring and
established when the subjects were chosen. Such constant across both time and testing conditions.
 Individual Differences and Capabilities 111 301

But the two individuals might also be different individual differences be reliable, and various
in other ways. For example, if on just a single methods to assess reliability have been devised
trial, one person makes a successful pool shot (see "Reliability and Individual Differences").
and another does not, we might not be willing
to say with certainty that the two people are dif-
ferent in pool-shooting capability, as on the next Abilities
shot the performance success may be opposite.
Finding differences between two individuals on Probably the most important topic in the area of
some measure of performance does not necessar- individual differences is abilities. In this section,
ily indicate that these differences are reliable. The we begin by providing a definition of the concept
stable, enduring differences among people are the of ability, and then we tum to some of the research
subject of this chapter on individual differences. indicating the structure of abilities.
In fact, the definition of individual differences
is the stable, enduring, and underlying differences
Abilities Defined
among people (Henry, 1959; Schmidt, 1975a). It The term ability, which is often used interchange-
is critical that measures used in the study of ably with the terms capability and aptitude, usually

Reliability and Individual Differ~nC·~~

The reliability coefficient provides a way to evaluate the extent.to which. o.b~;rv~d·~l;~~Jci.~.;·
o' '', '' '' "" ,'',

people on some tests are due to individual differences (i.e., stable, e~duringdiff~renqeS')°Qrc
or transitory effects. Reliability is really another use of the correlation, butin.thi~ in~ta.nq~1ij
is with the correlation of a test ''with itself:· For example, assume .th!:i.t fiVe subj~st~.~~cl'): · ·
trials on a reaction-time (RT) task. These six scores for each person are di\l,'iclecl t,qfoit~l.
("halves") for each person according to one of a number of different rnethocls. One sqrnrti
called the "odd-even" method; the sum of the odd-numbered trials is computed a~•on~a{f
for each person, and the sum of the even-numbered trials is taken a$. a secqnd, S,~parlite scot~ qt.;F.
each person. The extent to which the odd a.nd even sums for each indiyidUal tet'lcUo ·. .· ·.. ·• · ' ·
each other is one measure of the random variations in the indiVi:Ouartrjalcla~· ~r1At~~cl.
no random variation at all, the sum of the odds and evens woi:Jldbe ~X(lctiythe sarp
subject. Next, the across-subjects correlation is computed between the odcf a.,rt4~~e
correlation is typically called the reliability coefficient. Reliability cah, theoreti2aljy/ ·.· ·
between +1.0 and -1.0 (review Chapter2); but, in practical situati()A!:>i reli~pfli~~~;
and usually ranges from 0 to +1.0.
One way to interpret reliability·is to multiply it by 100 to·expre$$lt~.p~l'1
variation among people is made up of (a) differences a.morig peopteJ~. tnet~ ..
(termed individual differences) and (b} random (or other) fluct~atiofl.i;th~tJM~.
appear to be different. If so, reliability is the percentage of the observe,d~t!a
differences. With low reliability of, say, 0.20, only 20% of the pbser'.ved,vc:i'd~tf
differences, with about 80% being due to random (and other) variations,.,. 0
A primary concern of individual-differences research is the cei(re!ei.tion bet
Statistically, the size of the correlation between two tests is limited by tneielia
of the tests being correlated, 1 so one must be certain that the reliability'ofs~qh'"P
correlated is reasonably high. Reliability also represents a measure of t~e"stabifity"
different applications. Certainly as the number of trials administered to'eei.~,l;l~L.!Pi~.cr
performances become more stable, and reliability increases. Not surptisinglyttb~r:l
that make up a test is a strong determinant of reliability; indeed, if the nurnberoftrii¥tl.
enough, the reliability can actually be brought to 1.0 (e.g;, Gullicksen, 1950),Jl.~ ~ll!~Y'
do (usually), there is seldom a good excuse for a test wi.tti low relia;piHty:
302 Motor Control and Learning

refers to a hypothetical construct that underlies effort we devote to the game, because we do not
(or supports) performance in a number of tasks possess the requisite abilities. Two individuals
or activities. An ability is usually thought to be could have the same skill level at a given time,
a relatively stable characteristic or trait. These but one of them could have far greater potential
traits are typically regarded as having been either because he has greater abilities for the skill in
genetically determined or developed during question. The other person is likely to be frus-
growth and maturation, and they are not easily trated by attempts to improve beyond the limita-
modifiable by practice or experience. Abilities tions defined by his underlying capabilities.
represent the "equipment" that a person has at
Varieties of Abilities
her disposal, determining whether or not a given
motor task can be performed either poorly or well. Abilities can take many forms. One form
Abilities are usually inferred from patterns of common to many sports is body configuration.
performance on groups of tasks, largely using For example, the ability (or trait, if you prefer) of
correlations as the primary method of measure- height is important to basketball; similarly, small
ment. For example, suppose we find that, for a people rarely succeed in American football, and
group of individuals, those people who perform large people rarely succeed in gymnastics. Such
well on task A also perform well on task B, and characteristics are surely genetically defined and
that those who perform poorly on task A also are almost impossible to modify by training or
perform poorly on task B. This pattern is what one practice. Another variety of ability is related to
would expect if tasks A and B were related, or cor- certain emotional or personality characteristics.
related statistically. If two tasks are related, then For example, certain personalities are more ame-
they might be related because some underlying nable to team sports than to individual sports,
property or process is included in both. The differ- and some people are more excitable or anxious
than others. Finally, the abilities that will most
ential scientist is interested in the possibility that
concern us here could be called "motor abilities."
the underlying properties common to these two
They are the underlying characteristics that tend
tasks are abilities-or enduring, stable traits that
to contribute to success in moving the limbs in
contribute to the performance of the two tasks.
particular ways such as reaction time, movement
Abilities Versus Skill speed, and manual dexterity, among many others.
Another way to understand the concept of abili- These abilities are often not as easy to measure or
ties is to distinguish it from the notion of skill. An isolate for study as are abilities relating to body
ability is a relatively stable, underlying trait that is configuration, but they are no less important for
largely unmodifiable by practice. Skills, of course, understanding why certain people become more
can be modified by practice or experience (the skilled than others.
last part of this book covers these changes with
practice-called motor learning). Thus, abilities The Structure of Motor Abilities
underlie or support certain skills. The capability How many motor abilities are there? What are
to react quickly may underlie a number of specific they, and how can they be measured? These
skills such as sprint starts in swimming or a quick questions have been asked for many years, and
reaction in driving a car. Also, think of a given the answers have changed systematically as more
skill as composed of a number of different abili- effective techniques for studying abilities have
ties. Thus, the skill of driving a car may be made been developed. First, consider some of the earlier
up of various abilities, such as those involved in thinking about motor (and cognitive) abilities that
vision, an ability to switch attention from one has led to present-day beliefs.
event to another, and an ability to anticipate.
General Motor Ability
Abilities as Limiting Factors An early notion about motor abilities-one that is
Abilities can also be conceptualized as represent- still widespread among people not familiar with
ing limitations on performance, or as defining a the research-was the idea that all motor perfor-
person's potential for success. Neither of your mances are based on a single, all-encompassing
authors will ever become a professional basket- ability (Adams, 1987). This idea goes by different
ball player regardless of the amount of time and names, such as "athletic ability," "coordination,"
 Individual Differences and Capabilities II!! 303

"motor ability," or a more formal label of general

/1
frequently, concerning the sizes of the correlations
motor ability." All these terms imply essentially the among tests of various skills. Take a group of indi-
same thing: that we are structured with a single viduals and test them on task A; after consider-
capability to move, with this capability having able practice the people tend to order themselves
relevance to any motor task in which we choose on this task from "best" to "worst." According to
to engage. the general motor ability hypothesis, the "best"
These ideas, which were explored largely in the performers are most proficient because of a strong
1930s, were no doubt supported by the common general motor ability; conversely, the "worst"
observations (on playgrounds, in athletics, and performers have a weak general motor ability.
so on) that certain individuals seemed able to Thus, the performance on this task can be taken
do anything they tried (so-called all-around as a measure of the "strengths" of these subjects'
athletes). A well-known example at the time general motor ability. Now, suppose these same
was Mildred "Babe" Didrikson Zaharias, who people are tested on some other task B. The argu-
won Olympic medals in 1932 in the javelin, high ment is that, because the individuals who are
jump, and hurdles. She also excelled in baseball, "best" on task A have a strong general motor abil-
basketball, and tennis; and later she went on to ity, then these same individuals should be "best"
win 55 amateur and professional events in golf. at task B. Similarly, the individuals "worst" at task
In contrast, it seemed that other individuals A should be "worst" at task Bas well.
could not do well at any motor task they tried. This kind of prediction can be seen by the
Given these casual observations, it made sense hypothetical data in the scatterplot shown in
to postulate an underlying factor relating all the figure 9.1, where each individual is represented
various tasks in sports to one another: a general by a dot whose position on the scatterplot is
athletic (or motor) ability. determined by scores on each of the two tasks A
This idea of general motor ability was prob- and B. This pattern suggests a strong relationship
ably led by the analogous research on cognitive between the two tasks, and a high correlation
abilities, which was prevalent in the 1930s. It is (near +1.0 in this case) between tasks A and B.
beyond the scope of this chapter to present much This is a major prediction of the general motor
detail, but one important concept that emerged ability hypothesis. If the correlation turned out to
from this work was that of intelligence. It was be low (i.e., near zero), this would not support the
during the early 20th century that measures of
IQ (intelligence quotient) were developed, and
educators and parents became strong believers
in the predictive power of IQ tests as measures High
of a child's capacity for "success" in society-a
hypothesized general mental ability. m
~
In the 1950s and 1960s, however, the concepts ti)
.f!
of general intellectual ability and general motor c
0
ability both came under serious attack. With Cl)

respect to motor skills, the threat came from "

c
Ill
essentially two major sources: (a) the work on ...0E
individual differences conducted by Hemy (see 't:
Cl)
c.
"Franklin Hemy" on p. 12) and his students at
Berkeley and (b) the research program related
to individual differences in pilotry and similar Low
tasks that was conducted by Fleishman and his
associates through the U.S. Air Force (see "Edwin Low High
A. Fleishman" on p. 309). Performance on task A

Predictions From the General Motor

Ability Notion FIGURE 9.1 Hypothetical data presented in a scatter-
plot showing a relationship between tasks A and B; each
The general motor ability hypothesis has one subject is represented as a dot, positioned according to the
important prediction that has been examined scores on task A and task B.
304 Motor Control and Learning

general motor ability concept. Next let's examine tion with an armed services testing program. This
correlations among tasks that have been found in produced a 50 X 50 correlation matrix, whereby
the literature. every test is correlated with every other test and
the resulting correlations are placed in a large
Correlations Among Skills
table. The majority of tests correlated about 0.40
A large number of separate investigations in the or lower with each other; only rarely was there a
published literature deal with the correlations correlation of 0.50 or higher, which again fails to
among well-practiced skills, but we discuss support the idea of general motor ability.
only three of these studies in order to give the Many more studies like these have been done.
general idea (for a review see Marteniuk, 1974). Generally, low correlations are found among
One example from Henry's laboratory is a study different skills. This pattern does not support
by Bachman (1961). A group of 320 people prac- the notion of a general motor ability or even of
ticed two motor tasks that supposedly involved general motor subabilities such as balance (Bach-
the ability to balance. One of these, the Bachman man, 1961) or quickness (Lotter, 1960). Marteniuk
ladder task (see figure 2.8), required subjects to (1974) published a critical review of the research
climb a free-standing ladder. Bachman' s second conducted during the 1950s and 1960s that
task was the stabilometer (see figure 2.5c). This remains important and relevant today.
apparatus was an unstable balancing board on
which the subject stood; it pivoted so that the Henry's Specificity Hypothesis
right foot moved down as the left foot moved up. In the late 1950s, Henry (1958/1968, 1961) pro-
The subject's task was to keep the unstable board posed the idea, in direct contradiction to the gen-
still, with feet level. Bachman (1961) found that eral motor ability hypothesis, that motor abilities
for various subgroups of subjects (defined by age are specific to a particular task. Essentially there
and gender), the correlations between success on were three aspects to this hypothesis. First, Henry
the ladder and success on the stabilometer ranged thought that the number of motor abilities was
from +0.25 to -0.15, with most correlations being very large-perhaps in the thousands. Second,
very close to zero. These results ran contrary to a he believed that these abilities are independent,
general motor ability hypothesis, or even to the so that the strength of any one particular ability
concept of a general balance ability. is unrelated to the strength of any other ability.
Lotter (1960) had subjects perform striking Third, each task or skill that we perform depends
actions of the hand and kicking movements with upon a large number of these abilities. When the
the foot, and measured reaction and movement task is changed, the particular collection of abili-
times for each. The hand action involved a for- ties that supports the performance must change
ward-downward movement to hit a suspended to meet the new task demands.
tennis ball as quickly as possible (for the left Probably the most important prediction of the
and right hands separately); the leg movement specificity hypothesis is that two tasks, even if
involved kicking a small plate with a move- they appear to be quite similar (such as throw-
ment similar to a place kick in American football ing a baseball and throwing a javelin), will tend
(again, for the right and left legs). The arm-arm to correlate nearly zero with each other. The
correlation was 0.58, and the leg-leg correlation groups of abilities that underlie these two tasks
was 0.64. These correlations were considerably are, according to this view, two distinct collec-
higher than those found by Bachman (1961), but tions with few or perhaps no abilities in common.
note that they involve the same task performed Because these abilities are assumed to be inde-
with limbs on opposite sides of the body. The pendent, the correlation among skills should be
correlations between arm and leg, on the same zero, or at least very low. As we discussed in the
side of the body (e.g., right arm vs. right leg) or previous section, the evidence supports such a
on opposite sides (e.g., left arm vs. right leg), viewpoint.
were considerably lower: 0.24, 0.36, 0.23, and 0.18. In addition, the Henry hypothesis predicts
These findings do not support the general motor that transfer among skills should be quite low
ability hypothesis either. (Schmidt & Young, 1987). Transfer is defined
Finally, Parker and Fleishman (1960) had 203 as the attainment (or loss) of proficiency in one
subjects perform a battery of 50 tests in conjunc- task as a result of practice or experience at some
 Individual Differences and Capabilities fflll 305

other task (we discuss the concept of transfer in classified as motorically "clumsy" were more
greater detail in chapter 14). If the two tasks have variable in both motor and perceptual timing than
no abilities in common, then no element practiced age-matched controls. Research involving groups
in one of them will contribute (or transfer) to the of individuals who have neurological damage
other. Generally, the transfer literature supports suggests that the cerebellum may play a specific
Henry's hypothesis, showing essentially that role in timekeeping ability (Ivry & Corcos, 1993;
motor transfer is generally low and positive. Ivry & Keele, 1989; Ivry, Keele, & Diener, 1988).
These findings have led Keele and Ivry (1987) to
Exceptions to Henry's Specificity
suggest a modular view of individual differences
Hypothesis (see also Jones, 1993). In this view, the brain is
Although Henry's hypothesis about specificity of organized to perform certain functions rather
individual differences has been widely accepted than certain tasks. Modules represent neural
for many years, recent work by Keele, Ivry, systems that support a particular function. A
and others indicates that timing may represent timing module represents a type of timekeeper
an exception to Henry's view. This new work that functions to support the performance of both
suggests that a general "timekeeping" ability perceptual and motor tasks under a wide range
underlies performance of a number of tasks. The of sensory and effector mechanisms.
research focuses on how the temporal aspects of We note some controversy in this literature,
movements are organized in the central nervous however, as not all studies of motor timing have
system (CNS), and examines correlations among supported a general timekeeping ability. Only
various tasks requiring central control of timing. moderate correlations (0.36 to 0.48) were found
In one study (Keele & Hawkins, 1982), correlations when the timing of limb and jaw movements
between maximum rates of tapping by various were compared (Franz, Zelaznik, & Smith, 1992).
body parts (finger, thumb, wrist, arm, and foot) Studies in which timed movements were made by
ranged from 0.60 to 0.80, considerably higher tapping, line drawing, and circle drawing showed
than the correlations seen in the earlier studies. mostly small correlations; and correlations were
In another study, Keele, Ivry, and Pokorny (1987) significant only when the same tasks were per-
asked subjects to maintain a regular beat by tap- formed at similar timing rates (Robertson et al.,
ping with the finger or arm at 400 ms intervals; 1999; Zelaznik, Spencer, Doffin, 2000).
the measure of performance was the regularity
Factor-Analytic Studies
of tapping, measured by the SD of the produced
time intervals. The correlation between the SDs A second major research thrust uses the factor-
produced with finger and arm tapping was 0.90, analytic method, so called because of a statistical
suggesting a high degree of commonality between tool termed factor analysis. Various investigators
the two tasks. Interestingly, Keele and colleagues have used this general method, but certainly the
also included a task in which subjects were most active was Fleishman (1964, 1965, 1967; see
required to produce certain forces with the finger Fleishman & Bartlett, 1969, for reviews). After a
and arm. The correlation between performance in brief discussion of the factor-analytic method,
the force production task and in the timing task this section presents some of the major findings
was about 0.20. The dissociation of performance from this body of research.
abilities within the same subjects-showing speci- Factor Analysis For factor analysis, typically, a
ficity for the force and timing tasks, but generality large number of people (e.g., 100 to 200) perform
across different timing tasks-is evidence against each of a number of tests (e.g., 50). Factor analysis
a strict view of Henry's hypothesis. groups the tests into clusters, or factors, so that the
Support for a common timing ability has been number of factors is considerably less than the
found in a number of studies. For example, Keele, number of original tests (e.g., 10 or so, depend-
Pokorny, Corcos, and Ivry (1985) found that the ing on a number of other considerations). The
correlations between perception of timing and tests that make up a particular cluster or factor
the production of timing were relatively high have the property of showing relatively high cor-
(with rs of about 0.60), suggesting a link between relations with each other, whereas tests that are
perception and production of action. Williams, members of different factors tend to show low
Woollacott, and Ivry (1992) found that children correlations with each other.
306 1111 Motor Control and Learning

This perhaps will be clearer if we consider group of tests into a smaller group of factors,
the diagram in figure 9.2. Say there were 20 each of which is thought to represent one or more
tests, and the factor analysis grouped them into separate abilities.
five clusters, or factors, each represented by a Identifying the Abilities The next step is to
square. The circled numbers within the squares determine what abilities the factors represent. We
refer to the test numbers (which are, of course, can illustrate this process with one of Fleishman's
purely arbitrary), and the numbers on the (1957) studies, in which 200 people were each
arrows joining two tests represent the correlation given 18 tests. The result of a factor analysis is a
between the two indicated tests. For simplicity, factor matrix, shown in table 9.1. Across the top
only some of the 20 tests and five factors are are the nine factors that emerged in the analysis,
shown. analogous to the clusters or groupings of tests in
Notice that for Factor I, the tests (1, 8, and 9) figure 9.2. On the left side are the tests that were
show relatively high correlations with each other administered, ordered arbitrarily. The numbers
(0.52, 0.49, and 0.65), indicating that these tests in the body of the table are called the factor load-
tend to be measures of some common underly- ings, which indicate the extent to which the test in
ing ability. For Factor II, the tests (6, 13, and 18) question is a "measure of" the ability or factor in
tend to correlate with each other as well (0.60, that column; the factor loading is often thought of
0.48, and 0.50) and therefore tend to measure as the correlation between the test and the ability
some common ability. However, a test in Factor I that it represents.
(e.g., test 9) does not correlate well (r = 0.06) with For example, consider test 2 (Reaction Time).
any test in Factor II (e.g., test 13). Also, test 6 (in This test has a high loading (0.60) on Factor I but
Factor II) and test 2 (in Factor III) do not correlate a relatively small loading on Factor II (-0.15). On
well with each other (r = 0.10). The interpretation the other hand, test 4 (Pattern Comprehension)
is that the tests within a factor tend to measure has a high loading on Factor II (0.66) but a very
one or more basic abilities. Each of the factors small loading on Factor I (0.12). Here, we can say
tends to represent a different set of abilities. In that Reaction Time is a measure of Factor I (but
this general way, factor analysis divides a large not of Factor II), whereas Pattern Comprehension
is a measure of Factor II (but not of Factor I). Had
we displayed these factors as clusters as in figure
Factor I Factor II 9.2, Reaction Time and Pattern Comprehension
would have been represented as members of dif-
ferent clusters.
The values of the factor loadings help the sci-
entist to understand the structure of the abilities
that underlie the various tests. Consider Factor
IV. Scan the factor loadings for Factor IV, noting
which ones are high. All the loadings are less.
than 0.30 except for two-0.52 for the Mechanical
.10 Principles test and 0.65 for General Mechanics.
The scientist would be tempted to believe that
Factor IV has something to do with mechanical
knowledge and to name this factor accordingly.
Fleishman named Factor IV "Mechanical Expe-
rience." Now look at Factor IL The tests with
high loadings are Pattern Comprehension (0.66),
Mechanical Principles (0.53), and perhaps Speed
of Identification (0.44). Fleishman named this
factor "Visualization." In some cases (Factor IV),
FIGURE 9.2 Three clusters (represented by boxes)
of tests (shown as circles) that might result from a factor
the name of the factor is rather obvious from the
analysis. (Correlations between tests within a cluster are nature of the tests that "load on" it. In other cases
higher than correlations between tests in different clusters.) it is not (as with Factor II).
 Individual Differences and Capabilities 307

2. Reaction time
.... •"··~····-"-''''''

3. Rate of movement
4. Pattern comprehension
5. Mechanical principles
6. General mechanics

8. Visual pursuit

-03
01
28
16
15. Kinesthetic coordination -01 -16
16. Unidimensional matching 14 16 34 14 08 19
17. Two-hand matching 16 21
18. Discrimination reaction time 28 24
Note. Decimal omitted. Factors are identified as follows: I = Speed of Arm Movement; II =Visualization; Ill = Perceptual Speed; IV= Mechanical Experi-
ence; V =Spatial Orientation; VI =Response Orientation; VII =Fine Control Sensitivity; VIII= Complex Coordination "Within Task" Factor; IX= Residual
Factor; h' is called the communality, and is the sum of the squared factor loadings for that test (e.g., .18' + .22' + ... + .20' = .48).
From Fleishman 1957.

Notice that a given test can load on two factors of the factors appeared to represent a separate
at the same time. An example is test 5 (Mechani- ability or group of abilities.
cal Principles), which loads on both Factor II Motor Abilities Identified
(0.53) and Factor IV (0.52). One interpretation is
by Factor Analysis
that this test is made up of at least two abilities
(Visualization, Mechanical Experience) and that it The following list of abilities has been deter-
measures both of these abilities at the same time. mined in a number of separate studies, using the
This is in keeping with the idea that any given methods outlined in the previous sections. The
list is not exhaustive; it merely provides an idea
performance or skill (e.g., test 5) can be thought
of the kinds of abilities that have been inferred
of as composed of many abilities.
from these methods. These abilities all come from
Finally, the pattern of factor loadings is differ-
Fleishman's work (e.g., Fleishman, 1964, 1965,
ent for the various factors. That is, the tests that
1967; Parker & Fleishman, 1960) and bear the
load highly on one factor are typically not the
names given by him. After each we have provided
same tests that load highly on the other factors.
an example of a "real-world" task in which this
Compare Factors I and II, for example. Here, the
ability might be used, based largely on conjecture.
tests with the highest loadings on one have very
low loadings on the other. This observation is • Control Precision. This ability underlies
really the same as that in figure 9.2, where each the production of a movement for which the
308 111 Motor Control and Learning

outcomes must be rapid and precise, but which • Arm-Hand Steadiness. This ability under-
is made with relatively large body segments. lies tasks in which the person must be quiet and
Example: swinging an ax. steady. Example: aiming in riflery or archery.
• Multilimb Coordination. This ability • Wrist-Finger Speed. This ability underlies
underlies tasks for which a number of limb tasks for which alternating movements (e.g.,
segments must be coordinated while moving tapping) must be made as quickly as possible; it
simultaneously, such as the two hands, the two seems to represent the rapid coordination of the
feet, or the hands and feet. Examples: juggling, muscles required for up-and-down movements
playing a piano. of the fingers and wrist. Example: piano trills.
• Response Orientation. This ability under- • Aiming. This ability underlies tasks for
lies tasks for which rapid directional discrimina- which the subject must aim or point at a target,
tions among alternative movement patterns must attempting to hit it with very quick movements.
be made, and it is apparently related to the ability Example: dart throwing.
to select a correct movement in choice-RT situa- • Physical Proficiency Abilities. In addition
tions. Examples: actions performed by a defensive to the abilities listed for the movement control
lineman in American football or a hockey goalie. area, other abilities have to do with physical or
• Reaction Time. This ability underlies tasks structural aspects of the body. Some of these as
for which there is one stimulus and one response outlined by Fleishman (1964) are Extent (Static)
and which require the subject to react as quickly Flexibility, Dynamic Flexibility, Static Strength,
as possible after a stimulus in simple-RT situa- Dynamic Strength, Trunk Strength, Explosive
tions. Example: sprint start. Strength, Gross Body Coordination, Gross Body
• Speed of Arm Movement. This ability Equilibrium, and Stamina (Cardiovascular
underlies tasks for which the limb must be moved Endurance). These nine abilities can be thought
from one place to another very quickly and the of as underlying dimensions of physical fitness
measure of performance is movement time (MT). or physical proficiency, and they appear to be
Example: a jab in boxing. separate from the skills-oriented abilities listed.
• Rate Control (Timing). This ability under- Role of the Factors in Skills
lies tasks for which the movement speed of the How do these factors contribute to particular
limbs must be adjusted to the movements of the skills? It is interesting to consider that a number
environment so that the person's limbs are timed of factors seem to represent what we might call
correctly. Example: tracking tasks, as in steering "quickness." For example, Response Orienta-
a race car. tion involves RT with more than one stimulus-
• Manual Dexterity. This ability underlies response alternative (choice RT); Reaction Time
tasks for which relatively large objects are refers to RT tasks involving only one stimulus-
manipulated, primarily with the hands and arms. response alternative (simple RT); and Movement
Example: hammering a nail. Time refers to making an arm movement quickly.
Thus, "quickness" seems involved in each of
• Finger Dexterity. This ability underlies
these three kinds of tasks, yet they are repre-
tasks for which small objects are manipulated,
sented by different underlying abilities. Also, note
primarily with the fingers. Examples: repairing
that Manual Dexterity and Finger Dexterity are
a wristwatch, sewing.
separate abilities. Both pertain to the hands, but
• Postural Discrimination. This ability for different-sized objects. If so, then what does
underlies tasks for which subjects must respond it mean to say that a person "has good hands"?
to changes in postural cues, in the absence of To answer this question adequately, we would
vision, in making precise bodily adjustments. have to know about at least two different abilities.
Example: walking in the dark. Clearly, the structure of human motor abilities is
• Response Integration. This ability underlies far more complicated than common sense would
tasks for which the person must utilize and apply lead us to believe. Statements such as "John is
sensory cues from several sources into a single, good with his hands" must be combined with
integrated response. Example: throwing a pass information about how the hands are used in
in a football game. order to be meaningful.
 Edwin A. Fleishman
Certainly one of the most influential figures in the field of research on individual differences-par-
ticularly as it relates to motor behavior-has been Ed Fleishman (figure 9.3). In the years after World
War II, when pilot performance and pilot selection procedures were being examined, Fleishman
took a position at the Air Force Human Resources Research Center in San Antonio, Texas, from
1951to1956. There, he developed a program that linked correlational and experimental methods in
the study of perceptual-motor abilities. During this period and afterward at Yale University and his
later affiliations, Fleishman and his colleagues published numerous studies in which factor-analytic
methods were used to examine the fundamental structure of human motor abilities. These studies
identified and defined the abilities required by individuals to perform a variety of perceptual-motor
tasks, ranging from those involving fine manipulations of objects to those involving gross and coor-
dinated movements or complex timing operations (see Fleishman, 1954, 1972). Fleishman later
broadened this focus to measures of physical fitness; he found that the abilities required for physical
fitness are largely separate from those involving perceptual-motor skill. In this area he identified and
defined the abilities involved in performing physically demanding tasks and specified the tests most
diagnostic and reliable in measuring each ability (Fleishman, 1964).
With an understanding of the fundamental perceptual-motor abili-
ties developed in his factor-analytic studies, Fleishman then sought
to determine how these abilities could be used in the prediction of
success in piloting and in many other occupations. Fleishman was
also interested in how individual differences could be used in the
study of learning (Fleishman & Rich, 1963) and retention (Fleishman
& Parker, 1962) of complex perceptual-motor skills. He was one of
the first to show that the particular combination of abilities involved
in learning such skills changes at different stages of skill acquisition.
In later years, Fleishman and Quaintance (1984) and Fleishman and
Reilly (1992) developed methods for analyzing the ability require-
ments of jobs; these methods were based on his earlier taxonomic
work and were extended to cognitive abilities. Fleishman showed that
tests predicting job performance could be administered with the use
of these methods.
Fleishman's work leaves a legacy for future efforts on solving
problems of prediction. In 2004 he was awarded the American Psy-
FIGURE 9.3 Edwin A. Fleishman
chological Foundation's Gold Medal Award for Life Achievement in the (b. 1927).
Application of Psychology (see American Psychological Association, Courtesy of Edwin Fleishman.
2004). A reading of this retrospective of his career provides a more
complete picture of his immense contributions to psychology in general
and to the field of motor behavior in particular.

Selected Bibliography
American Psychological Association. (2004). Gold Medal Award for Life Achievement in the Application
of Psychology. American Psychologist, 59, 352-354.
Fleishman, E.A. (1954). Dimensional analysis of psychomotor abilities. Journal of Experimental Psy-
chology, 48, 263-272.
Fleishman, E.A. (1972). Structure and measurement of psychomotor abilities. In R.N. Singer (Ed.), The
psychomotor domain: Movement behavior (pp. 78-106). Philadelphia: Lea & Febiger.
Fleishman, E.A., & Quaintance, M.K. (1984). Taxonomies of human performance: The description of
human tasks. Orlando, FL: Academic Press.
Fleishman, E.A., & Reilly, M.E. (1992). Handbook of human abilities: Definitions, measurements, and
task requirements. Potomac, MD: Management Research Institute.

309
310 li!I Motor Control and Learning

Note that the particular collection, or pattern, to some because they do not appear to agree
of these abilities appears to change markedly with with a number of common observations. On
only minor changes in the task or situation. For playgrounds, for example, some children seem to
example, use of a choice-RT ability or a differ- be able to do well at nearly any motor task they
ent simple-RT ability depends on the number of try whereas others are nearly always ineffective,
stimulus-response alternatives presented to the which makes abilities appear to be more general
subject. The task remains generally the same with than they are. How can this apparent contradic-
respect to the movements made, but the way in tion be rationalized?
which the movement is signaled by the stimuli on A number of forces are at work to produce suc-
the display is changed. Again, different abilities cess in activities like sports. Some parents encour-
are used when the task is changed. age their children to participate in many sports;
Finally, it has been tempting to align the human for these children, this contributes to motor learn-
abilities defined by factor-analytic methods with ing in a number of sport tasks and can provide
the experimental research described in the first the appearance of a strong general motor factor.
sections of this book. Fleishman and Bartlett Also, playground activities favor those children
(1969) discuss the hypothesis that the abilities who are larger and more physically mature than
are measures of separate ways in which humans average, leading mature children to practice many
process information. In some situations, this con- sports and relatively immature children perhaps
nection is easy to imagine, such as with simple- to avoid sports. A kind of "rich get richer" phe-
versus choice-RT abilities being differentiated on nomenon develops, whereby a little experience
the basis of the need to resolve uncertainty. Simi- and encouragement can result in more experience,
larly, abilities like Rate Control (or anticipation, more encouragement, and so on. Such processes
timing) involve the processes related to analysis can be completely absent for other children.
of incoming sensory information.
Criticisms of the Abilities Approach Taxonomies
The abilities approach has not been without
its critics. One problem is methodological (see A taxonomy is a classification scheme used to
Kleine, 1982, 1985). There are many varieties of assign things to various categories. We discussed
factor analysis and of other techniques (called a few of these classification methods in chapter 2,
"rotations") that are applied to factor-analytic in connection with open versus closed skills and
outcomes to aid in interpretation. Choices about continuous versus discrete skills, for example.
the way in which the same data are analyzed More elaborate systems for classification have
can change the nature of the factors that emerge. been developed on the basis of the underlying
Fleishman' s work in the armed forces has been structure of the abilities involved in motor tasks.
criticized-unfairly, in our opinion-as limited in The basic notion is that one of the ways to clas-
scope, dealing primarily with young servicemen. sify tasks is in terms of the pattern of abilities that
Rarely were women used as subjects; children underlie them, rather than more superficial and
were never included, nor were older people. obvious characteristics such as whether the per-
Thus, the studies suffer some limitation in gen- formance is discrete or continuous. For example,
eralizability. Finally, the abilities that emerged performance on the pommel horse in gymnastics
from these studies have been based on skills for might be made up of Strength, Rate Control, and
Multilimb Coordination, with each contributing
which the person is typically seated and using
a certain proportion to the whole. Again, the
the hands (occasionally the feet as well), and for
leader in the early development of this area was
which the performance is noncompetitive. These
Fleishman (e.g., Fleishman & Stephenson, 1970).
kinds of performances are but one kind of action
Two ·of these general taxonomic methods are
involved in the total spectrum of motor activity,
described next (others are presented in Gawron,
and the abilities that emerged from these studies
Drury, Czaja, & Wilkins, 1989).
probably are somewhat limited as a result.
Ability Structures in Everyday Activities Factor-Analytic Classifications
The conclusions from the previous sections-that After a great deal of experience with the vari-
abilities are very specific-are often troublesome ous motor tasks in previous factor analyses,
 Individual Differences and Capabilities 111 311

the scientist sees that certain tests seem always In theory, this procedure can be applied to any
to result in the emergence of certain factors or new task, such as the game of golf. We could use
abilities. A good example is the various tests of a large number (e.g., 50) of well-practiced and
rapid movement that, when included in factor skilled golfers and administer reference tests to
analyses, typically produce a factor that is labeled them, perhaps the eight tests we have mentioned.
Movement Speed. Such stable findings lead to the (We might wish to use somewhat different tests
establishment of a particular MT test as the ''best" if there is reason to believe that some factor defi-
measure of the ability of Movement Speed, with nitely is, or definitely is not, involved in golf.)
the "best" test being defined in terms of ease of Then, we administer a golf performance test and
administration, the sizes of the factor loadings, examine the reference tests and the golf test with
and so on. These tests are often elevated to the factor analysis. The loadings that emerge should
status of reference tests-the generally agreed- give an indication of the nature of the abilities
upon measures of a particular motor ability. involved in golf.
Now, if this reference test is included in However, all of this is not as simple as we
another, subsequent factor analysis that involves have perhaps made it sound. A major problem is
tests for an entirely different activity (e.g., having "good," well-understood reference tests in
pilotry), we can understand the structure of the first place. Fleishman's research has provided
this new task somewhat if we note the extent to a start, but there are limitations to the generality
which Movement Speed ability loads on the new of this work, as we have pointed out. And many
task. If high loadings with the Movement Speed abilities that might be involved in golf may not
ability are obtained, then we can say that the be detected, as they may not be represented by
new activity has an important Movement Speed any of the reference tests. But with progressive
component. By including other reference tests at additions to knowledge about the underlying
the same time, each one representing a different abilities, the reference tests will become more
predetermined ability, we can pinpoint the abili- numerous and more effective, making the task
ties represented in this new activity. of discovering new tests easier in the future. This
A good example is provided in the factor method has a great deal of potential for practical
matrix shown in table 9.1. Tests 1 through 8 are application for new jobs in industry, for which the
tests of various kinds of skills, such as Instrument nature of the abilities needs to be known.
Comprehension, Reaction Time, and so on. These
eight tests are the reference tests-items that had Task Analysis
been studied extensively so that their underlying
abilities were understood. Now notice that tests 9 A considerably easier but less effective way to
through 12 are all based on the same apparatus, determine the nature of underlying abilities is
called the complex-coordination task (see figure through a series of procedures that has been
2.Sd), in which the person manipulates an aircraft- termed task analysis. The essential idea is to ana-
type "joystick" and foot pedals to respond to a lyze (i.e., break down) the task to consider what
pattern of lights presented on a display. kinds of abilities might be involved in it. Look at
What abilities underlie the complex-coordina- the flowchart in figure 9.4, from Fleishman and
tion task? For simplicity's sake, consider only the Stephenson (1970; Fleishman, 1975). Consider
initial performance measures, trials 1through5 a skill such as the throwing task performed by
(test 9). The highest factor loadings for this task a shortstop in baseball. Begin at the top of the
were for Factor VII (0.42) and Factor VI (0.38), chart, asking whether or not speed is important
with Factors II and V having slightly smaller load- to performance of this task. It is, so move to the
ings (0.35). We can therefore say that this task (at right and ask whether accuracy is important also.
this stage of practice at any rate) is based on Fine Follow the arrow dictated by that decision. If
Control Sensitivity (VII) and Response Orienta- baseball players or coaches are asked to use such
tion (VI), and to some extent Visualization (II) and procedures to evaluate throwing, researchers can
Spatial Orientation (V). By comparison, none of arrive at a tentative understanding of the kinds
the other factors seem to be involved very much. of abilities required for this action. We would
In this way, such a procedure "defines" this task probably come to the decision that this task
in terms of its underlying ability structure, just involves Control Precision, because speed, accu-
as a recipe defines what a licorice milk shake is. racy, and fine control are all needed for success.
312 Motor Control and Learning

Start

No

Exit

FIGURE 9.4 A binary flow diagram used to make decisions about abilities underlying particular tasks.
Adapted, by permission, from E.A. Fleishman and R.W. Stephenson, 1970, "Development of a taxonomy of human performance: A review of the third year's progress:' Technical
Report No. 726-TPR3 (Washington, D.C.: American Institutes for Research).

These methods are only in the beginning stages ences is the problem of prediction. The funda-
of development and problems obviously exist, mental problem is to be able to say (or predict)
such as ambiguities that make it difficult to give with some degree of accuracy what a person's
satisfactory answers to the questions in the boxes. score or level of skill will be on one task as a result
of information about that person's measure on
some other task or tasks.
Prediction These problems exist everywhere in the prac-
tical world. For example, insurance companies
Companion to the research on abilities, the second attempt to predict the probability that you will
major aspect of the work on individual differ- have a car accident on the basis of information
 Individual Differences and Capabilities 1111 313

such as age, gender, driving record, and where over rate of 80% per year. The employees simply
you live. Also, universities attempt to predict could not, or would not, continue in the job for
success in graduate school from tests that suppos- very long, even though the pay was quite good.
edly measure certain intellectual abilities (e.g., the Management suspected that they were hiring
Graduate Record Examination). More important people with the wrong abilities.
to motor skills is the need to predict success in
Task Analysis
jobs or occupations.
Prediction in this sense implies the concept The first step in developing a test battery was to
of futurity-forecasting some event or behavior perform a task analysis to generate an estimate
before it happens. While this is the primary kind of the skills involved on the job. The authors then
of application of individual-differences work, asked what kinds of abilities might be required
it is not the only way the idea of prediction is in these particular skills. Earlier studies provided
used. A second way involves the study of which a number of possibilities such as Manual Dex-
tests best "predict" (or are correlated with) some terity, Finger Dexterity, and Steadiness. These
other score that we already have in hand. Such made up a preliminary group of reference tests
techniques are intimately involved in the devel- that were probably involved. A second group of
opment of test batteries, and some of these pro- tests consisted of face valid tests, meaning that
cedures are described in the next section. the performances involved in the tests closely
mimicked some of the skills on the job, such as a
Test Battery Development: test in which subjects matched two-dimensional
An Example pictures of teeth with three-dimensional models.
A third group of tests concerned fundamental
Suppose that you are a personnel director for a mechanisms in motor control, as discussed in the
large company and that you want to predict suc- earlier chapters here. For example, to measure the
cess in some particular job in the company. For consistency of muscular impulses, subjects were
many jobs, this is easy. If you know the relevant asked to make rapid ballistic movements of the
skills, you can simply measure whether an appli- fingers, with a goal MT of 100 ms; measures of
cant has these skills (e.g., plumbing skills). But timing consistency were emphasized. The Fitts
a problem arises with a job for which the appli- tapping task was also used, as was aiming a
cants have had no previous experience. Common stylus to a target.
examples involve accepting applicants into a The result of this process was 24 predictor
flight training program in the Air Force or hiring tests-tests from which success on the job was
a person to perform a particular assembly line job to be predicted. These 24 predictor tests, about
that exists only in your company. The problem is equally divided among the three categories
to determine whether or not the applicant has the described in the previous paragraph, were then
necessary abilities to do this job after a training administered to 44 of the current employees, all
period, as no applicant would be expected to of whom had had at least one successful year on
have the particular skill without prior experience. the job.
Posing the Problem Criterion Variable
Consider the occupation of dental technician, Probably the most critical aspect of the project
studied several years ago by Schmidt and Pew was the development of a criterion score. This score
(1974). In this occupation, the primary task is was the best estimate of the employee's level of
to make fittings for the mouth, such as crowns, skill on the job. The supervisors were asked to
bridges, and artificial teeth, according to a den- rank the employees from "most skilled" to "least
tist's prescription. The job involves use of the fin- skilled," and to do this in a group session in which
gers and hands in the manipulation of tools and each employee's skill was discussed and a con-
various materials such as gold, plastics, and wax. sensus about the final ordering was achieved. It
Employees usually perform their tasks seated at was this rank that was to be predicted with the
a workbench, and the job is relatively sedentary. battery of tests, and this rank represented the
The major problem giving rise to this study was measure that one would like to predict. In other
that the firm requesting the analysis had a turn- occupations, the criterion score might have been
314 Motor Control and Learning

one's yearly production level, quality of play in Essentially, the method allows the determination
a sport as judged from game films, peer ratings of which predictor tests are the strongest predic-
of performance on the job, or any other measure tors of the criterion score, minimizing duplica-
providing a "best" estimate of success. tion, and allows the generation of a regression
equation that assigns a weight to each predictor.
Validation In the present case, the final regression equa-
Next, the task was to determine which of the 24 tion had the form
predictor tests were the most valid (or useful)
RANK = 0.599(BA) + 0.42(TL) + 0.56(SS)
in predicting the criterion. First, the correlation
+ 0.30(RL) (9.1)
between each of the predictor tests and the cri-
terion was computed, and the authors rejected where the abbreviations (BA, TL, SS, and RL) are
any test with a correlation less than 0.15. This the predictor test names: Ballistic Aiming, Tap-
preliminary screening left 14 tests that correlated ping (Left Hand), Spatial Scaling, and Rotated
0.15 or greater with the criterion. Letters, respectively. This regression equation had
Next, the problem was to determine which a multiple R of 0.50, meaning that the combina-
tests, and in what combination, were most effec- tion of tests shown correlated 0.50 with the cri-
tive in predicting the criterion score. Decisions terion score (rank). This provided a preliminary
like this are based on a statistical procedure called battery of tests that could be used in the evalua-
multiple regression (see "Multiple Correlation"). tion of new applicants. The potential employee

Multiple Correlation
Multiple-correlation methods are very similar to "regular" correlation· measures, except that the mul-
tiple correlation (abbreviated R) is the correlation between some criterion score (a measure of what
one is attempting to predict) ahda weighted combination of what are called predictor scores (which
are test scores from Which one would like to predict the criterion). Imagine combining the predictor
scores (for example, [1.0 x test 1] plus [2.5 x test 2] plus [8.0 x test 9] plus [5.5 x test 24] to form
a single sum for each subject. If we correlated this weighted sum (now a single score) with the cri-
terion score (another single score) using traditional correlation as described earlier, we would have
a kind of multiple R. The only difference is that inherent in the multiple-correlation procedures is the
automatic determination of theweightsin the sum, so that the correlation between the weighted sum
and the criterion score is maximized. Thatis, the multiple-correlation technique automatically adjusts
theweights so as to achieve a maximum correlation between the predictors and the criterion, and
thus lt "decides"which ones of the predictors are most and least effective in predicting the criterion
and in what combination. This procedure produces what is called a regression equation, which is
the specifi¢ation of the weights and the variables that best predict the criterion score.
ljsually, the prediction is stronger With more tests. In the case of the example with the dental
technicians,tf1e prediction would be best if we were to use all 14 tests, but the time needed to
measure 14 test performances .on every new applicant would not be worth the effort. Thus, there
is a trade"off between the number of tests and the size of the multiple R. Adding tests raises the R,
but with each added test the multiple R is raised by a smaller and smaller amount, eventually to the
poihttha.t adding another test does not yield enough more useful information to justify the effort. In
the example cited here, the final battery involved four tests for which the multiple R was 0.50. The
battery of four tests accounted for about 0.502 x 100 = 25% of the variance in the criterion score
with the various predictors. Thus, about 75% of the abilities that were important for this job were
hbtincluded in the battery of tests. This level of success (or lack of it) in prediction is unfortunately
typical for many different situations-including the U.S. Air Force and prediction of pilot success, on
which considerable effort, time, and money have been spent. This highlights the idea that effective
prediction is difficult to achieve in the motor behavior area.
 Individual Differences and Capabilities 315

would perform the four tests; the weights would of success in pilot school. This level of predict-
be applied as shown in the equation, and a pre- ability is not particularly strong; it represents
dicted rank would be computed. Such a rank only somewhat better predictability than was
would be at least one basis for hiring or not hiring seen in the example with dental technicians (50%
that person. in that case).
This general method can be applied to a variety
of situations with only slight modification. The
techniques are relatively time-consuming and Individual-Difference
expensive (because of testing time), however. Variables
The area of industrial personnel selection has
used these techniques for years, and the armed
In this final section we describe some research
services have employed them in pilot selection
about certain kinds of individual-difference vari-
programs. A few professional football teams use
ables and the ways in which they relate to skilled
batteries of tests that have been developed in
performance. An individual-difference variable is
ways like those described here, but this is more
usually some definable trait that can be measured
the exception than the rule.
in people, such as age, height, weight, gender,
Some Efforts at Prediction of Skill or ancestry. It has been of some interest to deter-
mine what relationships these variables have to
With use of the basic methods just described, the performance of certain kinds of skills. Such
attempts have been made to develop batteries questions as "Do men outperform women on tests
of tests that would predict success in various of pursuit tracking?" or "How much does move-
situations in the military, in industry, and in ment speed decline as age increases beyond 40?"
professional sport. Clearly the most systematic are tied to this general way of studying individual
and large-scale effort in this regard has been differences. Other individual-difference variables
the attempt to develop prediction batteries for have to do with the ways in which people of dif-
U.S. Air Force and Navy pilots. A great deal of ferent skill levels perform certain activities, such
research has been directed toward the problem, as sport skills.
mostly during the post-World War II years, the Space does not permit a discussion of many of
1940s and 1950s. Fleishman was one of the most these individual-difference variables, so we con-
active researchers in this area, and some of his centrate on those that have been studied the most
important findings are presented in the next sec- and that have the most interest and relevance for
tion. For more information on these programs of understanding human skilled performances. The
research, see Fleishman (1956) or Adams (1953, discussion focuses on three important variables:
1956, 1987). age, neurological impairments, and expertise.
In this research, the criterion score is usually a Others have obvious importance but are not
measure of success in flight school. This in itself is dealt with here, such as gender; race or country
a complex score, consisting of instructors' ratings, of national origin; sociological variables like
performances on knowledge tests, evaluations of number of siblings and birth order; and body
"leadership" qualities, and subjective evaluations configuration variables like height, eye color,
of personality. Typically, the research is done weight, percentage of fat, and fitness. For more
using a large group of potential pilot trainees, on these kinds of variables, see Noble (1978) or
measuring a variety of motor and perceptual Singer (1975).
tests (the predictors) as well as the criterion score
(success in flight school), and then using multiple- Effects of Age on Skilled
correlation methods to determine which predic-
Performance
tors are the most useful for predicting success.
About 10 tests were identified in this program The study of the effects of age has traditionally
of work. Some of these seem to resemble the task been divided according to two separate age cat-
of flying an airplane, whereas others are not so egories. There are studies of what is called motor
obviously linked. From the regression equations development, that is, of children as they mature
developed, the combination of these tests corre- and gain experience, usually from birth through
lated approximately 0.70 with the criterion score the teenage years. A second category of studies
316 Motor Control and Learning

involves aging, or the processes related to the Another view considers the development of
effects of age on performance of individuals in motor skills within a dynamic systems approach
the upper age categories, often considering adults (Newell, 1986; Thelen & Ulrich, 1991; see chap-
65 years of age and older. The kinds of changes ter 8). By this view, motor development (infant
in skills seen during these two age spans are development in particular) involves the emer-
markedly different, even though the variable of gence of motor skills as a function of the complex
interest (age) is the same in both cases. interaction of the individual with the task and
the environment in which the task is performed
Motor Development
(Newell, 1986). The emergence of this view as a
One major finding that emerges from the motor new paradigm in the study of motor development
development research is that large and systematic is probably most closely associated with the work
gains occur in nearly every conceivable aspect of of Esther Thelen.
motor performance as age progresses from birth to
about 18 years (Haywood & Getchell, 2009; Keogh Motor Performance in Older Adults
& Sugden, 1985; Piek, 2006). These improvements One of the fastest-growing areas in the behavioral
can be divided into categories related to growth and biological sciences is gerontology, or the study
and strength (Malina, Bouchard, & Bar-Or, 2004), of aging. The primary concern here is with the
biomechanical changes (Zernicke & Schneider, upper age levels and the associated changes in
1993), posture (Woollacott & Sveistrup, 1994), the motor system that affect movement abilities
capacity to anticipate and predict (von Hofsten, in healthy older adults. Aging studies, like motor
1983), ability to process information from com- development studies, have been conducted on
plex displays, speed of decision and movement, numerous variables related to motor behavior,
and accuracy in throwing (Halverson, Roberton, such as postural control; coordination; rapid,
& Langendorfer, 1982), as well as many other aimed hand movements; tracking; timing; and
specific areas of research (see also Gabbard, 2008; learning. For a review of these variables and
Payne & Isaacs, 2007). We could probably make many others related to aging processes, readers
the case that, in some way, most of the laboratory can refer to Spirduso, Francis, and MacRae (2005).
and real-life tasks studied involve the various One of the most frequently studied questions is
abilities discussed in earlier sections of this chap- the effect of aging on RT. For example, Fozard,
ter. And these abilities appear to develop, through Vercruyssen, Reynolds, Hancock, and Quilter
maturation or experience or both, with increasing (1994) examined the changes in simple and choice
age up to about 18 years. So, it is not surprising at RT in males and females who ranged from 20 to
all that, on the average, older children typically 90 years of age. As can be seen from the data in
outperform younger children. figure 9.6 (p. 318), there is a marginal slowing in
One of the common hypotheses in the motor simple RT with age, but a more dramatic slow-
development literature today is the idea that as ing in choice RT. The decline in performance of
humans become older the capacity to process tasks requiring speeded decisions, especially for
information increases (e.g., Thomas, 1980). tasks of increasing complexity, seems to be one of
Children appear to be relatively deficient in the the most general findings in aging research (Sal-
rate and amount of information that they can thouse, 1985; Spirduso, Francis, & MacRae, 2005).
process, which, as we discussed in earlier chap- Another consistent finding in the aging litera-
ters, was an important limiting factor in motor ture is that people move more slowly as they age.
control tests such as Fitts task. Children appear These effects are seen not only in tasks for which
to have smaller capacities to hold information in speed is measured directly, but also in tasks for
short-term memory, shorter attention spans, and which speed is evaluated indirectly. For example,
perhaps less effective mechanisms for processing in tasks requiring accuracy and speed, the overall
information necessary for movement (e.g., feed- score often suffers with increasing age because
back or environmental cues-see Thomas, 2000; speed decreases. Or there may be a marked shift in
Thomas, Thomas, & Gallagher, 1993). Therefore the speed-accuracy trade-off, with speed decreas-
it is a logical theoretical prediction that motor ing and accuracy either increasing or being held
performance improves as processing capacity constant. According to early thinking about these
increases with age. speed deficits with age, the slowing reflects the
 Individual Differences and Capabilities 317

Esther Thelen
The study of infant development, and particularly so the research on
complex movement and behavior development, was influenced greatly
by the remarkable work of Esther Thelen. With graduate training in both
animal and human development, Thelen was unmoved by the dominant
views of the day suggesting that the emergence of stereotypical move-
ments in infants unfolded according to a predetermined, neural matura-
tion process. For example, the newborn stepping reflex is characterized
by "walking-like" rhythmic movements that are seen when the bottom
of an infant's feet are stimulated, but these usually disappear from the
infant's behavioral repertoire after 6 to 8 weeks of age. A dominant view
was that the disappearance of the stepping reflex reflected the natural
neural maturation of the infant's motor development. However, Thelen's
research showed that the reflex "reappeared" when an infant was sus-
FIGURE 9.5 Esther Thelen
(1941-2004). pended in water. Conversely, the reflex "disappeared" when an infant who
Photo courtesy of Indiana University. normally showed the reflex was fitted with small weights on their feet.
The disappearance and reemergence of these reflexes, therefore, were
not a matter simply of neural maturation. Rather, they reflected a much more complex interaction of
the infant's sensory-motor system with the gravitational influences of the environment, suggesting
that movements are largely determined as emergent behaviors.
Thelen's research and thinking extended beyond the boundaries of motor development to include
perception and cognition. She viewed action as a process by which infants learn to solve problems.
Her views therefore embodied cognition as part of a much more complex system of interactions of
infants with their environment. She passed away in 2004, but her legacy lives on as a true paradigm
shift in the study of infant development.

Selected Bibliography
Smith, L.B., & Thelen, E. (2003). Development as a dynamic system. Trends in Cognitive Sciences,
7, 343-348.
Thelen, E. (1995). Motor development. A new synthesis. American Psychologist, 50, 79-95.
Thelen, E., Schaner, G., Scheier, C., & Smith, L.B. (2001 ). The dynamics of embodiment: A field theory
of infant perseverative reaching. Behavioral and Brain Sciences, 24, 1-34.

Esther Thelen Tributes

Galloway, G.C. (2005). In memoriam: Esther Thelen May 20, 1941-December 29, 2004. Developmental
Psychobiology, 47, 103-107.
Smith, L.B. (2005). Lessons from Esther Thelen. Journal of Motor Behavior, 37, 83-84.
Spencer, J.P., Clearfield, M., Corbetta, D., Ulrich, B., Buchanan, P., & Schaner, G. (2006). Moving toward
a grand theory of development: In memory of Esther Thelen. Child Development, 77, 1521-1538.
See also the papers in the special issue of Infancy, 2008, 13(3), 197-283.

slowing in neurological activities in the CNS-not 1984). Being unwilling to make errors on a task
only those involved in nerve conduction times but makes a person-old or young-appear to be
also those involved in decision making and other slow. Strategic factors are currently the subject
aspects of information processing. Later studies of much continued research in aging and human
suggested, however, that the slowing may also be performance (e.g., Morgan et al., 1994).
related to the fact that older people appear to be However, although the most obvious findings
more "cautious" than younger people (Welford, from aging research are the factors that decline
318 Motor Control and Learning

550

500

450

400
(jj'
.s 350
l-
a:
300

250

200

20 30 40 50 60 70 80 90
Mean age (range of± 5 years)

FIGURE 9.6 Effects of age and gender on simple and choice reaction time (RT).
Reprinted, by permission, from J.L. Fozard et al., 1994, '1\ge differences and changes in reaction time: The Baltimore longitudinal study of aging;' Journal of Gerontology:
Psychological Sciences 49: 182.

with age, these somewhat pessimistic findings ing can go a long way to offset the declines in
do not represent the entire picture-there are the more physical accompaniments of aging
many aspects of performance that are main- (Starkes, Cullen, & MacMahon, 2004).
tained well into advanced years. For example,
Wishart and colleagues (2000) found that healthy Individuals With Neurological
adults over 80 years of age performed as accu- Impairments
rately and consistently as 20-year-old subjects
To this point in the book we have focused our
on an in-phase bimanual coordination task
discussion, with only a few exceptions, on the
(see chapter 8), suggesting that the declines in performance of persons who have an intact CNS.
motor performance may be attenuated in tasks However, important information has also been
that are highly automated. But cognition also discovered in numerous investigations examin-
serves a critical role. Older typists in an impor- ing movement in people who have had damage
tant study by Salthouse (1984) were slower to to the CNS. In chapter 6 we discussed Lashley' s
make alternate hand taps and in choice RTs, patient, who could accurately position his knee
but equal to younger typists in typing speed. despite the absence of sensory feedback resulting
The older adults compensated for hypothesized from a gunshot wound (see also the neuropathy
declines in motor and sensory processes by patient shown in figure 6.3). The primary purpose
accumulating more advance information about for studying individuals with particular neu-
the information to be typed. Similar findings are rological impairments is to gather information
emerging in older adults who have specific sport that might improve quality of life. However, a
expertise too-maintenance in performance (or side benefit of this research is the comparison
maintained performance relative to expected or with information about performance by persons
typical declines) has been found for tasks that with a nonimpaired CNS. Such knowledge can
have important perceptual components, such as lead not only to a better understanding of the
tennis (Caserta, Young, & Janelle, 2007). Thus, effects of the specific impairment, but also to a
aging research also yields the very optimistic theoretical advancement of specific roles of the
findings that perceptual and cognitive train- CNS thought to be involved in the impairment.
 Individual Differences and Capabilities m1 319

We discuss some specific aspects of research in (e.g., Kennerly, Diedrichsen, Hazeltine, Semjen,
this area to illustrate the diversity of this work. & Ivry, 2002). Figure 9.7 illustrates a demonstra-
In most studies of individuals with very low tion of the loss of interhemispheric connections
cognitive functioning, the nature of the neuro- in bimanual coupling. In this study (Franz, Eli-
logical impairment is unknown. Individuals with assen, Ivry, & Gazzaniga, 1996), subjects drew
Down syndrome, however, have a known genetic three-sided rectangles that were oriented to be
disorder that is manifested in certain physical and either spatially compatible (top half of figure 9.7)
cognitive impairments. Elliott, Weeks, and Elliott or incompatible (bottom half of figure 9.7). The
(1987) proposed a model in which hemispheric effects are dramatic: The incompatible rectangles
functioning in people with Down syndrome are nearly impossible for subjects with an intact
is partially reversed, compared to that in age- corpus callosum to draw well (bottom left of
matched controls. According to the model, this figure 9.7), although the individual with a cal-
reversal in functioning has implications regarding losotomy was able to draw these symbols with
specific aspects of motor performance in these little difficulty (bottom right of figure 9.7). These
individuals, particularly with regard to speech findings suggest an important role of the corpus
perception and production (e.g., Heath & Elliott, callosurn in tasks involving the control of both
1999). upper limbs (see chapter 8).
Persons with Parkinson's disease and Hun- One of the most highly investigated patient
tington's disease have provided important groups comprises individuals who have suffered a
information about the role of the basal ganglia stroke. In addition to the unilateral motor control
in movement (Prodoehl, Corcos, & Vaillancourt, deficits that often result from a stroke, these indi-
2009). Individuals with Parkinson's disease viduals demonstrate remarkable plasticity of the
typically have slowness of movement, problems CNS in the recovery of lost function. For example,
in movement initiation, and tremor. The basal one of the most recent developments in the treat-
ganglia pathology in Huntington's disease results ment of stroke has been termed constraint-induced
in excessive movements. A research strategy that therapy. This therapy is based on the principle that
has been popular in recent years, for example, has
applied what is known about bimanual coordi-
nation in young adults (chapter 8) and healthy
older adults to contrast with information gath-
ered about bimanual coordination performance
in those with Parkinson's disease (Almeida,
Wishart, & Lee, 2002; Byblow, Summers, Lewis,
& Thomas, 2002; Lin, Sullivan, Wu, Kantak, &
Winstein, 2007; Serrien, Steyvers, Debaere, Stel-
mach, & Swinnen, 2000) and Huntington's dis-
ease Gohnson et al., 2000; Serrien, Burgunder, &
Wiesendanger, 2002). Studies of this type provide
a greater depth of knowledge regarding the role
of the CNS (and the basal ganglia in particular) in
the control of movement than could be obtained
if experiments were restricted solely to the study
of nonimpaired individuals.
Another example of the use of knowledge
about bimanual coordination in nonimpaired
subjects is investigations of individuals who have Control Patient
undergone callosotomies (people with a surgically
severed corpus callosum-the neural bundle
FIGURE 9.7 Spatial coordination in subject with an
that connects the two brain hemispheres). These intact corpus callosum (left) and a callosotomy patient.
studies have revealed that the tight coupling E.A. Franz, J.C. Eliassen, R.B. lvry, and M.S. Gazzaniga, Psychological Science,
normally apparent in bimanual timed movements "Dissociation of spatial and temporal coupling in the bimanual movements of
callostomy patients; 7: 306-310, © 1996, Reprinted by permission of SAGE
does not occur in individuals with a callosotomy Publications.
320 Motor Control and Learning

self-generated, goal-directed movements involv- Problems in Movement Control"). Such investi-

ing the affected limb induce permanent changes gations have included the study of persons with
in recovery due to a resultant reorganization of the Alzheimer's disease (e.g., Dick, Hsieh, Bricker, &
CNS (see "Constraint-Induced Physical Therapy" Dick-Muehlke, 2003; Patterson & Wessel, 2002),
on p. 350 for more details). Recent clinical trials with damage to the cerebellum (e.g., Spencer,
have revealed quite impressive recovery of upper Zelaznik, Diedrichsen, & Ivry, 2003), and with
limb function in survivors of mild to moderate acquired brain injury (e.g., Holden, Dettwiler,
stroke, with retention of these changes up to two Dyar, Niemann, & Bizzi, 2001). These investiga-
years poststroke (Wolf et al., 2006, 2008). tions have provided valuable insights regarding
Comparative research investigations involving the functioning of the motor control system in
other patient groups have also been frequent in people with damage to the CNS, as well as a com-
recent years (e.g., see the extended discussion parative analysis to advance our understanding
of patient RS in "Neurological Disorders and of the CNS without damage.

Neurological Disorders and Problems

in Movement Control
Recently we hc;we become· aware of a neurology patient (RS) who demonstrates some curious
movement coorclination disorders. These .disorders seem especially interesting when considered
in relation to some of the issues discussed in this chapter and chapter 8. RS, who is right-handed
and in his late 60s, originally sought medical treatment because of clumsiness in his left hand and
arm as well as balance problems. His right hand and arm seem .to be nearly unaffected; conscious
detection oftactile stimuli in the left hand seems not to be disrupted.
Testing (e.g., X-rays, MRls, behavioral testing) revealed a rather diffuse pattern of neurological
degeneration that apparently affects several brain structures, including the frontal cortex, basal gan-
glia, and cerebellum (see lvry & Hazeltine, 1999; Mauk et al., 2000; Mauk & Buonomano, 2004). A
conclusive diagnosis for RS is unclear, but a degenerative process called multiple systems atrophy
has been suspected.
Following are some of the movement difficulties that RS experiences. Note the relationships of
these deficiencjes to some of the issues about motor control and coordination discussed throughout
this book andparticularly in the work of Franz and colleagues.

• RS has a marked decrement in left-hand fine motor control that is manifested in tasks such as
typing, fastening buttons, picking up small objects with the fingers, using eating utensils, and
tying shoelaces. Because consciously perceived tactile stimulation is essentially unaffected,
these difficulties may then bE:l due to some lack of nonconscious feedback control (full vision of
the objects does not compensate for these difficulties). See the discussion of tactile feedback
in c;hapter 5.
• Using a medical version of the Fitts tapping task (see Desrosiers et al., 1995), where the neu-
rologist has the patient tap alternately between the examiner's finger and the patient's nose
as quickly as possible, the number of taps RS could produce in 15 seconds with his left hand
was only half the number that he produced with the nearly unaffected right hand.
• The tendency for the two. hands to be coupled strongly to each other appears to be absent.
(See the section on bimanual control in chapter 8.) In fact, even when RS tries to move his
hands in the same pattern together, the left hand appears disconnected from the right and the
patterns do notresemble each other.The coordination of the hands together as one, which is
one of the most striking phenomena of "normal" motor behavior, seems almost totally absent
in HS.
 Individual Differences and Capabilities Ill 321

• Making a large (2 tootoi~m~t~f.

plays the requisite *oQf~ihati~6
attempting thistaskwlththS ·. · ·
This is also seen in ari ~noe.r
release is in gross, ef!'or, ·.·.. •. •• . •
• We appear to control limbs Wtt~
cally Cloes some nearlytrJ ··· ··
with the righfhand ~no~il')"l
hano. When HS tries su~~ ~
drawer ano will not l'elease·fn.e·i6·

Selected Bibliography
Desrosiers, J., Hebert,R., Bra'lo, G.,c&d~11l··~;
elderly people. Age andAgeiAg,:?~~·Jpa"l1~~
lvry, R.13., & Hazeltine, E (l999):su~c;:'QI". · · · ·
of a callosoton;ty patient, l-fu111M iwt:> ·.
Mauk, M.D., & Buonornano~ t>c.•'lk (~6()#);.tK;
Neuroscience, 2i 307-34.Q. ·· ·
Mauk,~-~·· Medina, J;E,Nores,\N+,;~gt\y~
or timing?. Current Biology, 10, R52~-ij9~5~•

Expert-Novice Differences they had played previously. To examine this

apparent superior "memory" under controlled
Issues regarding motor learning are the focus experimental conditions, expert and novice chess
in the next part of the book. However, a grow-
players were briefly shown chessboards of games
ing research area that has become a topic of
that had been partially completed; they were then
individual-differences research is the comparison
asked to recall the information by reconstructing
of performances in people who possess varying
the pieces' locations on the game board after the
levels of skill (Ericsson, 1996; Starkes & Allard,
pieces had been removed from view. As expected,
1993; Starkes & Ericsson, 2003). The fundamental
the experts were much better than the novices at
question concerns the ways in which experts per-
this recall task. However, when the "game" that
form differently than novices. This expert-novice
was shown was in actuality a series of randomly
research is characterized by various methodologi-
arranged chess pieces, the experts were no better
cal approaches, each contributing unique infor-
mation about the ways in which distinctions can at reconstructing the board than were the novices.
be made among people of varying levels of skill The arrangement of the pieces in an actual game
in certain activities (Abernethy, Burgess-Limerick, had meaning and relevance to the expert, but not
& Parks, 1994; Chamberlin & Coelho, 1993). We so for random patterns. The recall advantage was
briefly summarize some of these approaches. specific to the domain of the subject's expertise.
These important findings have been repli-
Recall of Briefly Presented Information cated in experts in other intellectual games (e.g.,
Early work in cognitive psychology (Chase & Scrabble; Tuffiash, Roring, & Ericsson, 2007) and
Simon, 1973; DeGroot, 1946/1978) revealed that in sports. For example, using a paradigm similar
master chess players had excellent recall of games to the chess work, Abernethy, Neal, and Koning
322 Motor Control and Learning

(1994) compared expert and novice snooker nonexperts under these occlusion methods to
players' recall of ball positions during an actual assess a key temporal marker or an important
match. Although the experts were better at the spatial location in the display that differentiates
task than the novices for recall of a partially the performance of the subjects. Presumably, the
completed game, no differences were found when nature of the specific skills that define expertise
the balls were scattered randomly on the table. for that sport underlies observed differences in
Similar patterns of findings, in which the recall performance (Williams & Ward, 2003).
advantage for experts is specific to the nature of A good example of this research was provided
the expertise, have been reported for many other by Muller and colleagues (2006; 2010), who vid-
sports (see reviews in Chamberlin & Coelho, 1993; eotaped cricket bowlers from the perspective of
Starkes, Helsen, & Jack, 2001; Williams, Davids, the batter. Edited videos were shown to groups
& Williams, 1999). of expert, intermediate, and novice players, who
made judgments about the location at which the
Eye Movement Recordings
ball would land under varying conditions. Tem-
Eye movement recording devices allow the experi- poral occlusion was manipulated by stopping the
menter to analyze the parts of a visual display that video at time markers during the bowler's deliv-
a subject fixates using visual search during per- ery-either well before or shortly before ball deliv-
formance (see chapter 2). Researchers have used ery, at the point of release, or with no occlusion.
this technique to identify how experts and novices Muller and colleagues found that only the experts
differ in terms of obtaining visual information could make use of the earliest time-occlusion
from the environment, particularly when watch- information-intermediate players showed some
ing the actions of an opponent. A recent review moderate success (relative to the novices) when
of the literature suggests that, in general, experts the video was occluded at the point of ball release,
tend to have fewer eye fixations on a target, but but novices only improved their accuracy well
each has a longer duration than nonexperts (Mann, after the ball had been released. Spatial occlusion
Williams, Ward, &Janelle, 2007). One particularly involved editing out parts of the display during
important time when these differences emerge the ball delivery up to the temporal marker of
is immediately prior to movement onset-what ball release, at which point the entire video was
Vickers (2007) has called the quiet-eye effect. stopped. The features that were spatially occluded
Experts in such sports as billiards, basketball, and included relevant parts on some trials (e.g., hand
golf are consistent in their tendency to have fewer and ball; bowling arm) and irrelevant parts on
eye fixations of longer duration than nonexperts. other trials (e.g., nonbowling arm; lower body)
It is unclear, however, what purpose this quiet-eye (see figure 9.8). The results revealed that a complex
period serves, although a probable role is given to relationship existed for the pickup of advanced
facilitating movement preparation. information; in impoverished viewing conditions,
only the experts were able to predict with accuracy
Occlusion Methods
when relevant viewing data were available. The
Two methods of studying expertise involve differences in performance between experts and
adapting the visual displays that are seen by nonexperts in occlusion experiments of this kind
the subject. Although this research paradigm suggest that the acquisition of perceptual skills
began many years ago with the use of film and in sport is not a simple, quantitative change that
videotape (e.g., Jones & Miles, 1978), more recent occurs with practice. Rather, the acquisition of
opportunities afforded by digital displays have expertise requires the learning of specific percep-
provided researchers with precise methods of tual cues to which information is acquired with
preventing the subject from viewing certain parts increasingly earlier temporal landmarks.
of a display. Spatial occlusion methods involve
masking certain relevant and irrelevant parts of Point-Light Displays
a display, such as the various parts of the arm A different approach to the occlusion of informa-
and racket of a badminton opponent (Abernethy tion uses a paradigm in which whole-body images
& Russell, 1987). Temporal occlusion methods are are replaced with visual markers (white dots
used to stop a display at critical points during an against a black background). When these point-
action (e.g., Salmela & Fiorito, 1979). A frequent light displays are set in motion, they reveal infor-
experimental approach compares experts and mation about movement kinematics to the viewer
 Individual Differences and Capabilities 111 323

a b c

d e

FIGURE 9.8 Five different types of spatial occlusion, which prevented subjects from seeing different body segments,
were shown under five temporal occlusion conditions in a cricket batting task.
Reprinted from S. MOiier et al., 2010, "Expertise and the spatio-temporal characteristics of anticipatory information pick-up from complex movement patterns,'' Perception 39:
745-760. By permission of authors.

who is sufficiently skilled to perceive that informa- Domain Knowledge Reports

tion accurately. The use of point-light information An alternative type of methodology for studying
has therefore become another important paradigm expert-novice differences is the analysis of self-
in the identification of factors that differentiate the report data. In this method the verbal reports of
skills of expert and nonexpert athletes. athletes engaged in certain sport-specific situations
In a sense, most of us are "experts" in some are subjected to a protocol analysis (e.g., French &
types of perceptual identification from some Thomas, 1987). The objective of this analysis is to
types of point-light displays. People can often identify domain-specific differences in knowledge
make the distinction between males and females, about certain tactics, procedures, rules, nuances,
and between friends and strangers, simply from and so on that might distinguish the expert from
the information contained in these moving white the novice. Obviously, this qualitative type of
dots. The identification of perceptual information information is not obtained through the more con-
in point-light displays of specific athletic actions, ventional methods that we have described. Studies
for example, would probably be observed only in of this type are becoming more frequent and pro-
the expert, by this thinking, and research using vide a number of important details about factual
these tools is becoming more common (Shim, knowledge and strategic differences between
Carlton, Chow, & Chae, 2005; Williams, Ward, & experts and novices (e.g., MacMahon & McPher-
Smeeton, 2004). son, 2009; McPherson, 1993, 1994; McPherson &
324 Motor Control and Learning

MacMahon, 2008; Starkes, Cullen, & MacMahon, A large number of individual-difference vari-
2004). One needs to be mindful, however, that even ables have been studied. These variables provide
experts are subject to illusions and biases-what an easy basis for classifying people, so that the
they "think" may be happening is often quite performance on some task can be studied "as a
subjective and not always supported by objective function of" these variables. Three major research
data (e.g., Lee et al., 2008a ). variables have attracted considerable research
interest: (a) age effects, both from the point of
view of growth and maturation and from the
Summary point of view of older age levels (gerontology);
(b) the effects of neurological impairments; and
The study of individual differences concerns the (c) expert-novice differences.
variations among individuals on various tasks or
behaviors. This kind of research differs strongly
from the experimental approach in which differ- Student Assignments
ences among individuals are ignored in order to
concentrate on the average performances of larger 1. Prepare to answer the following questions
groups of people affected by certain independent during class discussion:
variables. Two major subdivisions of individual- a. Use the concept of specificity of indi-
differences research can be identified. vidual differences to explain the per-
The major focus is on abilities, defined as formance characteristics of two video
stable, enduring characteristics or traits, probably game experts whom you know.
genetically defined, that underlie certain move-
ment skills or tasks. Since the 1930s, there have b. Add a novel example to each of the
been many points of view about the structure of abilities listed in the textbook in the
motor abilities. Some convergence has occurred section "Motor Abilities Identified by
on the idea that humans possess a relatively Factor Analysis."
large number of separate motor abilities (perhaps c. Timing is one ability that may be gener-
50 to 100), that these abilities are independent, alizable. Discuss how this concept might
and that any particular skill or task may have be examined for musical performance
many of these abilities underlying it. Henry's in a large group of 10-year-old children.
specificity theory and Fleishman' s work on factor- 2. Find a research article that contrasts the
analytically determined abilities tend to point to abilities of experts and novices in the per-
this kind of conclusion. Both of these approaches formance of a motor skill.
indicate that the idea of a general motor ability
defining proficiency on all motor tasks is surely
not correct. However, some abilities might turn Web Resources
out to be more general than the other abilities
mentioned here, though more research is needed A discussion of factor analysis:
in order to provide certainty about this.
www.fact-index.com/f/fa/factor_analysis.html
A second major division of individual-dif-
ferences research is related to prediction. Here, Link to Research Quarterly for Exercise and Sport, a
the relationships among scores on various tests journal that published many of the early investi-
(the predictors) are used to predict or estimate gations of individual differences:
the scores on some other test (the criterion),
www.aahperd.org/rc/publications/rqes/index.
such as the prediction of an applicant's prob-
cfm
able success on the job from various measures
of abilities. The development of predictive tests
is time-consuming, and the effectiveness of such Notes
batteries in predicting some criterion behavior is
disappointingly low even when a great deal of 1 The limiting factor is that the correlation between
research funding and effort have been applied. test X and test Y is theoretically less than the square root
The problem is that many abilities seem to exist, of the reliability of either test X and/ or test Y; that is, rxy
and these abilities are not well understood. s; ~and/or rxy s; ~·See Gullicksen (1950).
CHAPTER 10 Motor Learning Concepts and Research Methods
CHAPTER 11 Conditions of Practice
CHAPTER 12 Augmented Feedback
CHAPTER 13 The Leaming Process
CHAPTER 14 Retention and Transfer

t this point in the text we change the emphasis with respect to the treatment
of motor skills. The concern so far has been with skilled performance-often
at high levels of proficiency-and the numerous internal processes that make these
performances possible. We now shift to a different, but related problem: the learning
of skills as a result of practice or experience. The problem is different from the issue
of skilled performance, because we will be focusing on the changes in skill, rather than
the nature of skill at some particular level. As such, different methods and logic are
needed to understand these performance changes; and in chapter 10, we document
some of the more important methods used to understand the principles that have
been discovered about motor learning. Chapters 11 and 12 describe how conditions
of practice and augmented feedback influence learning. In chapter 13, we present the
various ways in which the learning process has been conceptualized by researchers.
This part closes in chapter 14 with a discussion of factors that influence retention (how
learning is retained over periods of no practice) and transfer (how learned behavior
can be applied in novel situations).

325
L earning is a critical part of our existence.
Think where humans would be if we could
not profit by the experiences and practices in
Motor Learning Defined
Learning in general, and motor learning in par-
which we all engage. You would not be able to
ticular, have been defined in a variety of ways.
read the words on this page, we would not be
Four distinct characteristics are included in the
able to type the words that appear here, and no
definition: (1) Learning is a process of acquiring
one would be able to speak. In short, we would
the capability for producing skilled actions. That
be simple creatures indeed if we were forced to is, learning is the set of underlying events, occur-
behave in the world equipped only with the skills rences, or changes that happen when practice
we inherited. The fact that we can acquire new enables people to become more skilled at some
knowledge and skills has led to a robust interest task. (2) Learning occurs as a direct result of
in the ways in which people learn, in the critical practice or experience. (3) Learning cannot be
variables that determine how people will profit observed directly, as the processes leading to
from experience or practice, and in the design of changes in behavior are internal and are usually
instructional programs. not available for direct examination; rather, one
We will not attempt to review the entire topic must infer that learning processes occurred on
of learning. There are examples of learning in the basis of the changes in behavior that can be
all organisms (even the simplest of single-celled observed. (4) Learning is assumed to produce
organisms), and the learning that humans enjoy relatively permanent changes in the capability for
is the most complex of all. Thus, many forms of skilled behavior; for this reason, changes in
human learning are not discussed here, such as behavior caused by easily reversible alterations in
the learning of verbal materials, the learning of mood, motivation, or internal states (e.g., fatigue)
concepts, and the learning of interpersonal skills. are not thought of as due to learning.
We concentrate on the acquisition of motor skills A synthesis of these four characteristics pro-
as defined in chapter 2. Essentially, the concern duces the following definition: Motor learning is a
will be with the effects of practice and experience set of processes associated with practice or experience
on performance, in an attempt to understand leading to relatively permanent changes in the capabil-
the relevant variables that determine gains in ity for skilled movement. We discuss these aspects
proficiency. in more detail next.

Ill 327
328 Motor Control and Learning

Motor Learning Is a Set minor reference to the usual use of the word),
of Processes but it has been named in other ways by other
theorists. Regardless of its label, the notion of
A process is a set of events or occurrences that, some internal state that is the product of learn-
taken together, lead to some particular product, ing represents a critically important distinction.
state, or change. For example, in reading we are Defining learning as producing a capability for
interested in processes that transform visual movement directs our focus to the internal state
information to provide meaning; in motor con- and the processes that have led to it, rather than
trol we may focus on processes of retrieving a simply to the behavioral changes. More impor-
motor program from memory; and in physiology tantly, the concept of a capability for movement
we can discuss processes that result in muscle implies that if the capability is "strong," then the
hypertrophy. Similarly, practice and learning skilled behavior may occur if the external condi-
can be seen as a set of analogous processes that, tions, motivation, and other surrounding factors
taken together, lead to the acquisition of the are present; if the conditions are not favorable,
capability for moving skillfully-with "skill" then the skilled behavior might not occur-for
being defined as in chapter l. These processes are example, if fatigue is present or motivation is
assumed; in other words, we assume that some set low. That is, behavior may vary for a number of
of processes must have taken place in order for reasons, only some of which are a result of change
learning to have occurred with practice. However, in the internal capability for movement produced
what these processes are (exactly) is not specified, by practice. This concept provides a basis for the
and in fact the nature of these processes is what distinction between learning versus performance,
learning theorists try to understand. Thus, an which is a major theme of this chapter.
important focus is on what happens-in terms of Numerous changes in humans can contribute
the underlying processes-when people practice to their capability for movement in skilled situ-
and acquire new skills. ations, but many of these have little to do with
learning as defined here. For example, we know
Learning Produces an Acquired that increased maturation or growth can lead to
Capability for Skilled Movement improvements in skill, as older children generally
(Habit) outperform younger ones. Similarly, changes in
strength or endurance from physiological training
The processes involved in learning-like all could contribute to certain kinds of skills, such as
processes-generates or results in a product or those used in weightlifting or soccer. However,
internal state. In the case of motor learning, this we would not want to include such improve-
state is an increased capability for moving skill- ments in a definition of learning because practice
fully in the particular situation. Notice that we or experience is not the basis for the changes in
have not defined learning as a change in behavior capability. We will be searching for situations in
per se, as many have done (e.g., Morgan & King, which changes in capability are primarily related
1971). In this sense, the goal of practice for the to changes acquired through experience.
learner is to increase the "strength" or the "qual-
ity" of this internal state, so that the capability for Motor Learning Is Not Directly
skill will be maximized in future attempts. The
researcher's goal is to understand the nature of the
Observable
internal processes that have led to the increases It should be clear that motor learning is not
in the state; thus, theorists propose hypothetical directly observable. The processes that underlie
processes to account for learning in experimental changes in capability-and the nature of the
settings. Also, the researcher wants to understand capability itself-are highly complex phenomena
the nature of the state itself, perhaps in terms of in the central nervous system, such as changes
the codes involved or the kinds of control it exerts in the ways sensory information is organized
on behavior; such knowledge will tend to provide or changes in the patterning of muscular action.
an answer to the question of what was learned. As such, they are rarely directly observable, and
William James (1890) used the term "habit" for one must infer their existence from changes in
this internal capability for movement (with only motor behavior. This feature of motor learning
 Motor Learning Concepts and Research Methods 329

makes it particularly difficult to study. Experi- behavior (skill), but these may or may not change
ments must be designed carefully so that the the internal structure of the person in a relatively
observed changes in behavior allow the logical permanent way. If the effect of some independent
conclusion that there were associated changes in variable can appear and disappear as the value
some internal state. of the variable is changed, then this change in
behavior cannot be associated with anything
Motor Learning Is Relatively relatively permanent, hence is not thought to be
Permanent due to learning.
How permanent is "relatively permanent"?
Another important feature of motor learning is This is a vague concept, and scientists studying
that it is relatively permanent. Something lasting learning are rarely clear about it. But the inten-
occurs when one engages in practice and learns tion of this discussion should be clear; learning
some activity-something that does not simply should have some lasting effect.
pass away in the next few minutes or hours.
More dramatically, we could say that when you
practice and learn, you will never be quite the Measuring Motor learning
same person as you were before. Learning has the
effect of changing the learner (if only slightly) in Given that motor learning is a set of processes that
a relatively permanent way. underlie the changes in a capability for movement,
With respect to skill learning, this distinction how can such a capability be measured in order to
is important because it rules out the changes in understand what variables affect it during prac-
skills that can come from a variety of temporary
tice? It will be helpful to consider a typical motor
performance factors. For example, skills might
learning experiment to explain some of the major
improve if the person is in the "right" mood, if
points in the measurement process.
motivation is temporarily high, or if certain drugs
are administered. Yet each of these changes in Performance Curves
behavior will probably vanish when the tempo-
rary effect of the mood, for example, "wears off." In a simple experiment on learning, a large group
Thus we should not attribute these changes in of individuals are asked to practice on some
behavior to motor learning, because they are not motor task, and the experimenter charts their
sufficiently permanent. performances as a function of "trials," resulting
An analogy may help to clarify this point. If in a performance curve. For example, figure 10.1 is
you cool water sufficiently, you will find that it a graph from Fleishman and Rich (1963) showing
becomes solid (ice); you can reverse the effect performance on the two-hand coordination task.
completely to produce water again simply by Subjects had to follow a moving target through
warming the ice. This is not so with boiling an movements of two crank handles, one controlling
egg. Boiling an egg for 10 min produces changes the forward-backward direction of a pointer and
that are not reversible when the egg is cooled. one controlling right-left movements (see figure
Some relatively permanent change has been made 2.Sd). The average time on target (in minutes)
in the egg that was not the case with the water. for a group of 20 subjects is plotted as a function
This analogy applies well to the concept of of successive blocks of four 1 min trials. A clear
motor learning and performance. The nature of trend can be seen for the scores to increase with
the water or of the egg can be observed directly, practice, with the increases being somewhat more
and they both behave in predictable ways when rapid at first and then leveling off later.
the independent variable (temperature) is When the measure of performance is an error
applied. But beneath the surface is some unob- score, for example, the scores will decrease, as
servable change in the nature of the substance; shown by the performance curve in figure 10.2.
in one case (the water), this change is completely These data, from Quesada and Schmidt (1970),
reversible and not relatively permanent, while in are from subjects' performances on a timing task.
the other (the egg), the change is not reversible. Here, the subject's task was to operate a switch
With human learning, many analogous variables when a moving pointer became aligned with a
can be applied in order to change the observed stationary one, and error was the time interval
330 11!1 Motor Control and Learning

c 3
§.
..
'&i
Cl
sc
0 2
GI
.§
..
GI
Cl
Ill
GI
>
<C

2 3 4 5 6 7 8 9 10
Blocks of four 1-min trials

FIGURE 10.1 A performance curve showing increases in the score with practice.
Reprinted, by permission, from E.A. Fleishman and S. Rich, 1963, "Role of kinesthetic and spatial-visual abilities in perceptual motor learning;· Journal of Experimental
Psychology 66: 9. Copyright© 1963 by the American Psychological Association.

between the switch movement and the actual practice, they both represent gains in performance
moment of coincidence. Average absolute error and almost certainly can be interpreted as caused
decreased rapidly at first, and more gradual by motor learning. Such effects are usually among
decreases occurred later in practice. the most powerful in the study of motor behavior.
Even though the two performance curves (fig- But what is usually of more interest than whether
ures 10.1 and 10.2) change in opposite ways with or not learning occurred is whether learning was

600

500

UI 400
....2
.§.

GI 300
.!!
::I

~
~ 200

100

2 3 4 5 6 7 8
Blocks of 5 trials

FIGURE 10.2 A performance curve showing decreases in the score with practice.
Adapted from D.C. Quesada and R.A. Schmidt, 1970, 'i'\ text of the Adams-Creamer decay hypothesis for the timing of motor responses," Journal of Motor Behavior 2: 273-283,
adapted with permission of the publisher (Taylor & Francis Ltd, http://www.informaworld.com).
 Motor Learning Concepts and Research Methods !'Ii 331

greater in one condition than in some other condi- procedure has a number of advantages, such as
tion. Thus, the question of interest relates to the "smoothing" or "canceling out" many random
role of variables that influence learning of the motor (perhaps meaningless) variations in scores due
task. In order to make meaningful inferences to inattention, errors in measurement, and other
about whether or not condition A produced more factors not directly related to the internal habit
learning than condition B, special procedures are changes. But at the same time, this averaging
needed to analyze these performance scores in procedure tends to hide any differences that may
motor learning experiments. have existed among people on a particular trial,
or it may hide important trends in improvement
Motor Learning Experiments with practice.
Graphs such as those shown in figures 10.1 and These effects can be particularly important in
10.2 are often thought to represent the acquired the study of learning. Consider two hypothetical
capability for movement in the subjects during subjects whose performances over a number of
practice from trial to trial, and to some extent they trials are shown in figure 10.3. Subject 2 seemed
probably do. For this reason, such curves are often to improve in performance early in the sequence,
loosely termed learning curves, as it is tempting to with little change occurring later. Subject 1 has
regard the changes in performance as reflecting a difficult time improving in the task until later
the product of the internal capability for move- in practice, when finally performance improves
ment generated by learning. The notion that these markedly.
curves "mirror" the internal state (the amount of Now consider what happens if these two sub-
habit) is oversimplified, however, and scientists jects' scores are averaged in ways that are typical
are very cautious about interpreting the changes for studies of larger groups. In figure 10.3, the
in curves like those in figures 10.1and10.2 as a center line is the average of the performances for
reflection of the amount of motor learning. Some the two subjects for each of the trials. The pattern
reasons for this caution are outlined next. of improvement with practice is considerably dif-
ferent now. It might be tempting to say that the
Performance Measures average capability for responding accumulated
The first reason performance curves perhaps gradually and consistently. This, of course, would
should not be assumed to reflect learning is be misleading, as neither of the subjects whose
that skilled performance-not the capability for
moving skillfully-is plotted as a function of
trials. Since the capability (habit) cannot be
measured directly, any change in habit that has
occurred must be inferred from the changes in
performance. Thus, it seems more logical to refer
to the curves exemplified in figures 10.1and10.2
as performance curves rather than learning curves.
CD
As such, they reflect both the momentary changes u
c
ca
in performance and the relatively permanent
...
E
consequences of practice. .g
CD
II.
Between-Subject Variability
A second problem in making inferences about
learning from performance curves is that they are
insensitive to the differences among individuals
that arise as a function of practice. Consider how
a performance curve is produced. A large number
of people (the larger the better, usually) is used. Trials
The first-trial scores for all of the subjects are
averaged to obtain the data point for trial 1; trial FIGURE 10.3 Hypothetical performance curves for two
2 scores for all subjects are averaged to obtain the individuals, together with the curve representing the aver-
data point for trial 2, and so on. This averaging age of their performances.
332 II! Motor Control and Learning

data produced the average performance curve

showed this trend. Thus, it could certainly be 0.50
that learning does not occur gradually at all, as is
usually evidenced by the individual performance
curves. Rather, learning might sometimes occur
as more of an abrupt "revolution" than a gradual UI

"evolution" in the ways in which the subjects ~

u
perform the task. ::I
UI
0
Within-Subject Variability ~ 0.25

As mentioned previously, one of the important :c

-8
aspects of the averaging procedure is the reduc- e
D..
tion of errors in measurement and of factors that
seem to obscure the "true" capabilities of the
people on a particular trial. As seen in chapter 7,
a typical finding is that people inherently vary
from trial to trial, even if they are attempting to 50 100
do the "same" thing each time. But is trial-to-trial Trials
variation for a particular person due to some
meaningless random fluctuation in the motor FIGURE 10.4 An average performance curve for which
system, or is it due to some meaningful change probability of success is the dependent measure. (On a
in the way the person attempted the task on a particular trial, any given subject can receive only O or 1.)
particular trial?
The problem can be better illustrated with an have shown this pattern of performance! Indeed,
example. Consider the task of free-throw shoot- there is no way that a given person could ever
ing in basketball when the subjects are relatively achieve a score of 0.20 on a particular trial, as a
inexperienced. Certainly a great deal of variability single subject can only achieve 1.0 (hit) or 0 (miss)
exists in this task's performance from trial to trial, on a single trial. Thus, the average performance
and much of this variability does not seem to curve in figure 10.4 obscures all the variations that
represent fundamental changes in the ways that occurred within people across trials and encour-
people attempt the task. If a large group of people ages us to draw conclusions about the learning
is examined on this task, with performance scored process that may be incorrect.
as "correct" or "incorrect," each person will have
a pattern of scores that shows a large number of Ceiling and Floor Effects
apparently randomly ordered hits (baskets made) Ceiling and floor effects are a third kind of prob-
or misses, with somewhat more hits as practice lem that can lead to erroneous conclusions about
continues. And different people will have the hits learning processes from group performance
and misses scattered across the trials differently. curves; these are present in many of the tasks
A performance curve can be plotted with these used in motor learning experiments. In most
data, with the measure for a particular trial being tasks, absolute scores exist that no person will
the probability of success on that trial (figure 10.4). ever exceed. For example, there can be no fewer
With this measure, the data point for trial 1 will than zero errors on a trial, no less than 0 s for
be the total number of hits divided by the total some movement-time task, and no more than 30
number of attempts (i.e., the number of sub- s for a time-on-target score in a 30 s performance
jects)-that is, the proportion of subjects that shot trial. Thus, as people approach these ceilings (the
successfully on trial 1; the method is similar for limitation in score at the top of the scale) or floors
trial 2. When plotted, the average performance (the limitation at the bottom of the scale), the
curve usually rises gradually, perhaps moving changes in the performance levels of the people
from 0.10 to 0.40 in 100 trials. From such a curve doing the task become increasingly insensitive
it is tempting to conclude that the capability for to the changes in learning that may be occur-
accurate shooting grew slowly as a result of ring in the people as they practice. As a person
practice. But note that not a single subject could approaches some ceiling or floor, it becomes
 Motor Learning Concepts and Research Methods 1111 333

increasingly "difficult" to improve performance;

in gymnastics, for example, it is far "easier" to 100
improve one's score from 6.0 to 6.5 than it is to
improve from 9.0 to 9.5, when "perfect" is 10.0. 90
Similarly, in golf it is much easier to reduce one's
total by five strokes, say from 140 to 135, than it is 80
to improve by five strokes from 75 to 70. In addi- ~
'Iii 70
tion to these absolute scoring ceilings and floors,
psychological or physiological floors and ceilings .
Cl
.I!!
c 60
can be present. For example, the 4 min mile was at 0
GI
one time a barrier that we thought would remain E
unbroken. Now the barrier is considerably lower, i= 40
but it could very well be that no human will ever 30
break a 3 min mile. Will anyone ever long jump
more than 30 ft? As performers approach these 20
physiological limits, it becomes increasingly more
difficult to improve. 2 3 4 5 6 7 8 9 10
Trials
Scoring Sensitivity and the Shape
of Performance Curves FIGURE 10.5 Time on target for a tracking task as a
The primary problem is that the "rate" of prog- function of trials for three different scoring criteria. (The 5%
ress (the slope of the performance curve) toward criterion indicates that the target used for scoring was 5%
of the screen width, and so on.)
some ceiling (or floor) is usually quite arbitrary Reprinted from H.P. Bahrick, P.M. Fitts, and G.E. Briggs, 1957, "Learning curves:
and dependent on the ways in which the task is facts or artifacts," Psychological Bulletin 54: 260.

measured. The rate does not seem directly linked

to the rate of change in the capability for move-
ment that underlies this change in behavior. A The authors then performed this procedure a
powerful example of this principle comes from second time, scoring the subjects using a differ-
a study by Bahrick, Fitts, and Briggs (1957, their ent tolerance for error (again, there was no target
"simple task"). The authors studied 25 male width as far as the subjects knew). Here, the target
subjects on a continuous tracking task for ten 90 width was 15%.of the width of the screen, and the
s practice trials. The pattern of the track that the subjects were evaluated in terms of the number
subjects had to follow, and the movements of the of seconds during a trial the pointer was in this
lever that the subjects made when following it, target zone. As the criterion of success was much
were recorded for later analysis. more lenient (the "target" was wider, therefore
The authors then analyzed the single set of hit more often), the time-on-target scores were
performances in three different ways. First, they naturally larger, forming a second performance
assumed that the width of the target the subject curve marked "15%" in figure 10.5. Finally, the
had to follow was small, 5% of the total width of procedure was done again, this time with a very
the screen. (There was, in fact, no target width wide target that was 30% of the width of the
as far as the subjects were concerned, as all they screen. In figure 10.5 the performance curve for
saw as a target during the performance trials was these data is labeled "30%."
a thin line that moved on the screen.) By going The important point about the data in figure
over the tracking records and examining the 10.5 is that all three curves are based on the
number of seconds in a trial during which the same performances, but the differences between
subject was in this imaginary 5% target band, the them are produced by the ways in which the
authors obtained a separate measure of the "time experimenter has chosen to evaluate those per-
on target" for every subject and trial. Then, the formances. We might conclude (if we did not
data from trial 1 were averaged for all the subjects know that all the data came from the same perfor-
to form the trial 1 data point in figure 10.5 for the mances on the same subjects) on the basis of the
curve marked "5%." The data for other trials for 5% curve that habit gains are a positively accelerated
this target size were handled in a similar way. function of trials (because the shape of the curve
334 1$ Motor Control and Learning

is concave upward), and on the basis of the 30% closer to hitting the target than are the children.
curve that habit gains are a negatively accelerated With a little practice, their initial advantage in
function of trials (because the shape of the curve is motor control begins to show up in terms of
concave downward). This is, of course, nonsense; increased time on target, whereas the children
only one pattern of habit gain emerged (whatever show no such effects even though they may be
it was), but evidence about this gain was obtained moving slightly closer to the target. We may
in three different ways that gave three different (erroneously) conclude that the adults learned
answers about how the capability for performing more than the children because their performance
progressed with practice. gains were larger.
The differences are apparently caused by the Now consider what happens if we do the
fact that making the criterion "easier" (moving 11
same" experiment, but with an easy" version of
/1

from 5% to 30% target widths) moves the person the pursuit rotor task in which the target is large
through the range from floor to ceiling at different and the speed of rotation is slow (figure 10.6b).
rates, depending on the sensitivity of the scores Now the adults begin very near ceiling, and the
and the level of skill. Thus, despite the fact that children are somewhere in the middle range. The
the same learning occurred in all three curves adults have little capability to demonstrate con-
from trial 1 to trial 10, very different amounts of tinued improvement, whereas the children start
performance improvements are displayed depending in a sensitive area of the scoring range where a
on which target zone one chooses to use. So, what little practice produces maximum score gains. In
is the pattern of habit change that occurred with this case, the gains in score are much larger for
practice? We have no idea, on the basis of these the children than for the adults. We might (again,
data, and we can conclude (erroneously) just about erroneously) conclude that the children learned
anything we choose merely by selecting the "right" more than the adults.
target width to study (see also Wilberg, 1990). This nonsense is caused by the marked differ-
ences in the sensitivity of the scoring system to
Implications for Experiments changes in the subjects' behaviors and movement
on Learning patterns-that is, in the sensitivity to changes in
These considerations present strong limitations the level of habit of the subjects. The central region
on what can be understood from experiments on of the scoring system (around 15 s time on target)
learning, often making it impossible to provide is very sensitive to changes in the subjects' habit,
clear interpretations about what happened to whereas the regions near the ceiling and floor
habit in the study. Consider this hypothetical tend to be relatively insensitive to such changes in
example. We want to study whether children habit. In fact, if we wanted to show that children
learn more than adults as a result of a given and adults learned the same amount, we could
amount of practice at some new task. (We could easily choose a scoring criterion intermediate in
as easily ask about males versus females, older "difficulty." So, given the choice of the sensitiv-
versus younger adults, and so on.) We choose a ity of the scoring system, we can produce just
task such as the pursuit rotor (figure 2.5a), which about any conclusion we desire about the relative
is foreign to both the children and the adults, and amounts of learning in children and adults.
allow both groups to practice. But games like this are not science, and it
A finding that emerges consistently from the makes no sense to play them. Even so, scientists
work on children's motor behavior is that adults do not know how to resolve this particular prob-
nearly always perform better than children lem, and thus we have no idea whether children
(although video game performance may be an learn more than adults or vice versa. The lesson
exception). If we were to use a pursuit rotor with is that such effects are always present in learn-
a very small target and a fast speed of rotation (a ing studies, and scientists have to be aware of
relatively "difficult" task), allowing the children the potential artifacts that they may produce
and adults to practice for fifty 30 s trials, we might in coming to their conclusions about learning
expect curves such as those that appear in figure (see also Estes, 1956; Sidman, 1952). Fortunately,
10.6a. Here, both hypothetical groups begin with experimental designs that minimize this kind of
nearly no time on target, but the adults improve problem are available. Some of these designs are
more than the children because they are relatively described in the next section.
 Motor Learning Concepts and Research Methods mi 335

Ceiling

30
i

~
Gi
...
Cl

.f! 15
c
0
Cl)
E
i=

a
Ceiling

30
i

~
Gi
...
Cl
....""
c 15
0
Cl)
E
i=

b Trials

FIGURE 10.6 Hypothetical performance curves for adults' and children's performances on a "difficult" or an "easy" task.
(Depending on the scoring criterion, one can [erroneously] conclude [a] that adults learn more than children or [b] that chil-
dren learn more than adults.)

Designing Experiments method with a more traditional old" method,

/1

where the independent variable is the method

on Learning used in practice. (Hundreds of other examples
could have been chosen.) We begin by assign-
One of the major goals in the study of motor ing participants from a large group of people at
learning is to understand which independent random to two groups. Then we administer one
variables are involved in maximizing learning, level of the independent variable (the "new"
which variables impair learning, and which have method) to one group and another level of the
no effect whatsoever. Clearly, such knowledge independent variable (the "old" method) to the
is important both for the development of useful other group, and record the performance levels
theories of learning and for practical application achieved during considerable practice of a novel
in teaching and other learning situations. motor skill. It might seem logical that the answer
Given the definition of learning and the limita- to the question of which of the two methods
tions on the kinds of behavioral changes that sci- produced more learning would be based on the
entists are willing to classify as learning changes, performance levels achieved during practice, and
how do we go about deciding whether a certain especially at the end of practice. But, for the rea-
variable influences learning or not? What follows sons discussed in the previous sections, we need a
is a discussion of a rather typical, but hypotheti- way to separate the relatively permanent effects of
cal, example of an experiment on motor learning. the independent variable from the transient effects·
Imagine that one is interested in some "new" on performance. To do this, researchers have typi-
method for practice and wishes to contrast this cally used so-called transfer (or retention) designs.
336 111 Motor Control and Learning

Transfer or Retention Designs Transfer and retention designs are quite simi-
lar, and often the terms are used interchangeably
Transfer or retention designs in their simplest by researchers to refer to the same test. In general,
form involve two related operations. First, the however, tests involving the same task as prac-
learners are provided a retention interval (or
ticed in the acquisition phase are called retention
rest period away from practicing the task) of
tests, as they evaluate the extent to which a given
sufficient duration that the transient effects of the
skill has been retained over the retention interval.
experimental variable (here, the practice method)
Transfer tests, on the other hand, typically involve
will dissipate. It is difficult to say how long such
new variations of the tasks practiced in acquisi-
an interval must be, but it is clear that we are
interested in the lasting, relatively permanent tion (e.g., performing the task at a new speed,
effects of practice and not the momentary benefits or with different lighting conditions), or they
afforded by the experimental variable. As a result, might involve essentially new tasks that have
retention intervals of 24 h or more are commonly not been practiced before. The essential feature
used before the subjects perform the transfer test. of both retention and transfer designs, however,
The second feature is that the learners are all is that the tasks and conditions are the same for
tested under a common level of the independent the two groups.
variable-that is, under identical conditions. This Returning to the example of the "new" and
is done so that any transient effects of the inde- "old" practice methods, the results might look
pendent variable (especially ones that are differ- something like those shown in shown in figure
ent for the different experimental variables) will 10.7, where the curves in acquisition are plotted
not return in the test to again hide the relatively for each of four different hypothetical situations.
permanent effects. Thus, in the retention test, we In each of the four cases, the "new" method (the
assume that the transient effects of the indepen- open circles in figure 10.7) exerts an effect that
dent variable have dissipated, but that the rela- tends to increase the performance measure rela-
tively permanent effects of the variable have, by tive to the "old" method (the filled circles). We
definition, remained, so that performance on the consider next some of the possible effects that
retention or transfer test will reveal differences in could occur on retention or transfer tests, and the
amount learned. interpretations we could make.

..
Cl)

0
(.,)
I!!
0
(.,)
CJ) CJ)

Acquisition Retention/ Acquisition Retention/

transfer transfer
a Trials b Trials

Acquisition Retention/ Acquisition Retention/

transfer transfer
c Trials d Trials

FIGURE 10.7 Hypothetical effects of an independent variable on performance (in acquisition) and learning (in retention or
transfer): (a) Size of the performance effect in acquisition is maintained in retention or transfer; (b) size of the performance
effect in acquisition is reduced somewhat in retention or transfer; (c) size of the performance effect in acquisition is elimi-
nated in retention or transfer; and (d) the performance effect in acquisition is reversed in retention or transfer.
 Motor Learning Concepts and Research Methods m1 337

In the first case, .shown in figure 10.7a, the method was better for performance in acquisi-
levels of performance after the retention interval tion. Apparently, the "new" method involves
are essentially similar to those shown at the end of many temporary benefits to performance, but
acquisition. Here we are forced to the conclusion these benefits do not survive until the retention
that the "new" method is better than the "old" for test. The effect is stronger than that, however.
learning. Further, all of the effect of the different With the "new" method-even though it is
methods in acquisition was apparently due to the performing very well in acquisition-subjects
variable's relatively permanent effects, and there are learning the task less effectively than with
was no transient effect to dissipate. the "old" method, which we see only on the
In the second case, in figure 10.7b, the "new retention or transfer tests. It may seem that such
method" group still outperforms the "old reversals in performance level from acquisition
method" group on the retention test, but the dif- to retention would never occur in real situations.
ferences are not as large as they were in acquisi- However, there are several cases like this, high-
tion. Still, though, we conclude that the "new" lighted in chapters 11 and 12, that force rather
method produced more learning than the "old" counterintuitive conclusions about the effects of
because the performances still favor the "new" practice.
group in retention. We also see that not all of the
difference present in acquisition was relatively Double-Transfer or -Retention Designs
permanent, as some of the advantage seen for There is a potential problem with using only one
the "new" group dissipated with rest, bringing level of the independent variable for the transfer
the groups closer together than they were in or retention test as was done in the previous
acquisition. In the situation shown, we conclude examples. Consider figure 10.7d again, and fur-
that there is a learning advantage for the "new" ther assume that all subjects switch to the "old"
method. method in retention or transfer. This naturally
In the third case shown, in figure 10.7c, favors the "old" practice method in acquisition,
performance of the two groups in retention is because those learning under the "old" condi-
essentially equivalent. Our decision rule leads tions are tested under the same conditions while
to the conclusion that there is no learning benefit those learning under the "new" conditions must
of the "new" method over the "old." During switch conditions in the transfer test. The same
acquisition, the "old" method tends to produce problem, but in reverse, would have occurred if
somewhat depressed scores, whereas the "new" we tested all subjects under the "new" conditions
method tends to elevate performance. We see in retention/ transfer.
this by viewing the retention performances One way to reduce this difficulty is to use
in relation to those in acquisition; the "new so-called double-transfer designs, in which the
method" subjects have lost some of their gains two acquisition groups are each split into two
in acquisition (which gains, therefore, must have subgroups for the retention test. For each of the
been temporary), and the "old method" group acquisition groups, one of the subgroups has
shows gains (i.e., from acquisition to retention) its retention test under "new" conditions and
presumably resulting from the dissipation of one under "old" conditions. This forms essen-
temporarily depressing effects in acquisition. tially four groups in the experiment-"new"
Overall, though, because (and only because) the and "old" in acquisition and "new" and "old"
two groups performed similarly in the reten- in retention. As we will see, these methods are
tion/ transfer test, we argue that there was no frequently used to avoid problems in interpre-
learning benefit for the "new" versus "old" tation, and we give many examples of them in
methods at all. chapters 11 and 12.
Finally, the case shown in figure 10.7d is an
interesting one. Here, the groups on the reten- Which Group Learned More?
tion test have reversed their orders relative to Notice that in all of these cases, the fundamental
their performance in acquisition. Of course, question "Which group learned more?" was
we argue that the "old" method is better for answered by an analysis of the performance
learning than the "new," based on the better levels in the retention or transfer test. We did not
performance in retention. Yet, the "new" consider the level of performance in acquisition
338 Motor Control and Learning

at all when considering the question of learning, Alternative Methods

because this performance confounds relatively
permanent and transient effects of the indepen- for Measuring Learning
dent variable. Nor did we use the difference
between the performance at the end of acquisition Many situations exist in which the measurement
and the performance in retention as a basis for the of performance, and thus the measurement of
answer. Simply, the group that performs better learning, does not give a good estimate of the
on the retention or transfer test was concluded relative amount" that someone has learned in
/1

to have learned more. practice. The problem is often that the perfor-
mance scores have approached a ceiling or floor
Learning and Performance during the course of practice, so that all the sub-
Variables jects appear to be the same on the task because
all the scores hover close to the ceiling or floor.
With the use of these experimental designs, it has
In such situations, attempts to show that a given
been possible to classify experimental variables
independent variable has effects on learning are
into essentially two categories. One of the catego-
ries is that of the performance variable. According thwarted because continued practice on the task
to its definition, this kind of variable has effects can result in no changes, as the scores are already
on performance while it is present, but when the maximized or minimized.
level is altered in transfer, the effect is altered as These problems can arise in at least two dif-
well. A performance variable is thus one that ferent settings. One of these involves simple
influences performance but not in a "relatively tasks, for which all subjects perform nearly
permanent" way. Using the analogy presented maximally in only a few practice trials. Here,
earlier, cooling water to make ice is only a "per- continued practice can result in no effects on the
formance variable" and not a "learning variable" performance score. A second situation relates to
because the effect of the variable vanishes when a particular type of complex task-those that
it is taken away. people have had a great deal of experience with
A learning variable, on the other hand, affects in the past, such as driving a car. Because sub-
performance after the variable has been removed. jects are so well practiced, little improvement
That is, the variable influences performance in a in skills will be evidenced as they continue to
"relatively permanent" way, affecting the learn- practice. The problem again is that subjects are so
ing of the task. Examples can be seen in the first close to a performance ceiling that no additional
hypothetical outcome in figure 10.7a, where the improvements can be shown. Other examples
effect of the independent variable remains even are the performance of high-level sport skills
when the level of the variable is changed. To and the performance of various highly skilled
extend our analogy, boiling an egg for 10 min is jobs in industry.
a "learning variable." In chapter 2, we discussed secondary-task
Finally, variables may be both learning and methods for the measurement of skills. The
performance variables. Like the hypothetical problem is similar to the present one, as the mea-
examples in figure 10.7, many of the variables sures of the subject's behaviors on a particular
that will be examined influence performance task may not give a good indication of the level
when they are present, with some part of the of skill this person possesses. The example we
effect dissipating when the variable is taken used was driving under the influence of fatigue.
away. Yet some other part of the effect remains The accumulations of fatigue from long, uninter-
when the level of the variable is changed, sug- rupted stretches of driving were not observed
gesting that the variable also has affected learn- in vehicle-control movements at all. However,
ing in the task. Examples like this can be seen in decrements were observed as a function of the
the hypothetical situation in figure 10.7b, where duration of the previous driving when subjects
not all of the effect of the independent variable were asked to perform a simultaneous secondary
has dissipated upon transfer. We will see other task, suggesting that there was a decrement in
examples of effects like these in chapters 11 "spare capacity" with increasing levels of fatigue
and 12. (e.g., Brown, 1962, 1967).
 Motor Learning Concepts and Research Methods iii 339

Secondary Tasks and Alternative ued practice at the ceiling did something to the
Learning Measures subjects, but we have no way to make this con-
clusion from the performance curves shown in
The measurement of performance on a task often figure 10.8, as both groups have the same final
does not tell us much about the person's level of score-essentially at the ceiling. We address these
learning. Additional practice (e.g., driving a car) questions using this basic experimental design
will probably result in some additional learning in the next three sections, so keep the situation
of the skill even at advanced levels of proficiency, indicated in figure 10.8 in mind. In each case,
but the experimenter may not be able to detect various methods with secondary tasks can be
these effects because the subjects are so close to used to answer these questions.
the performance ceiling or floor, as the case may
be. By using secondary-task methods, one can Automaficity and Learning
often see these changes more clearly. One hypothesis that has received considerable
Assume that two groups of subjects practice a empirical support is that skills become more
task and that they have reached a performance automatic (see chapter 4) with practice, in the
ceiling. Figure 10.8 shows some hypothetical sense that systematically less interference with
curves that could result. Now, suppose that one certain simultaneous secondary tasks will be
of the groups (A) is told to discontinue practice, shown. What would be the effect of imposing
whereas the other group (B) continues to practice. a simultaneous secondary task in the example
Group B's performance is shown as a continua- shown in figure 10.8? We could (as Brown [1962]
tion of the earlier curve along the ceiling, as there did) have the subjects do a mental task requiring
can be no further improvements in the score after the detection of a duplicated letter in a stream
the ceiling has been reached. This procedure, in of auditorily presented letters. The measure of
which a person practices further after having importance would be the extent to which the sub-
reached some criterion of success, is often called jects could improve on this secondary task as they
overlearning. practice the main task. Some hypothetical results
Did any learning of the task go on during the are presented in figure 10.9. We would probably
overlearning trials? Which of the two groups see continued improvement in the accuracy of
had learned more after all the practice had been the secondary task, even during the overlearning
completed? We might suspect that the contin- trials for which the score for the main task was

Practice trials

FIGURE 10.8 Hypothetical performance curves from original learning to a ceiling (groups A and B) and "overlearning"
trials at the ceiling (group B only).
340 iii Motor Control and Learning

CD
u
c:
Ill

0
_
...E~
u
't:
CD._
!!!
c.o
~u
UI ._
Ill CD
-• .a
E
~::I
c:
-gIll ,_.
0

Original learning Overlearning

Practice trials

FIGURE 10.9 Hypothetical data from a secondary task measured during practice trials of the task shown in figure 10.8.
(The secondary-task score continues to increase even though the main-task score is at a ceiling.)

essentially fixed at the ceiling. The improvement Speed of Decision and Learning
in the secondary task would suggest that learn- Another method is effective in situations for
ing was going on during the overleaming trials, which the main task involves decision making,
with practice reducing the attentional load and such as in learning to make the correct move-
allowing more accurate task performance. This ment when one of several different stimuli is
technique has not been used often, but it is useful presented. Early in the overlearning trials, sub-
in situations like this. jects are just able, with much time and effort,
Effort and Learning to generate the correct response. But, later in
Closely related to the notion of automaticity is overlearning, they can choose the correct answer
the notion of effort (Kahneman, 1973). As people more easily and far more quickly. What if, in
learn a motor skill, they appear to be able to do addition to measuring whether or not the subject
the task with less and less physical and mental could make the correct response, we measured
effort, possibly because they learn to perform the latency with which the subject did so (Adams,
with more efficient movements or because they 1976a)? We would probably see that the latency
process information more efficiently. If so, then of the response (which is not yet at a floor) would
simultaneous physiological measures could be decrease markedly even though the accuracy of
used to show that, during the overlearning trials the response did not change at all (since accuracy
in which the subjects are practicing the task at of performance was at the ceiling). This proce-
the ceiling, the effort in the task continues to be dure will not work with all motor tasks, but it
reduced with additional practice. Measures like seems ideally suited to those situations in which
oxygen consumption (assessed by techniques there are time pressures to make accurate deci-
associated with physiology of exercise) or pupil sions, such as deciding where to throw a fielded
dilation (also a measure of effort) are frequently baseball, or choosing a defensive maneuver in a
associated with levels of effort. If we find less dangerous driving situation. If such outcomes
effortful performances during these overlearning occurred, we would conclude that the decreased
trials, such data could be interpreted as showing response latencies indicated a continuation of
that the continued practice trials produced some learning even though the subjects were at the ceil-
additional learning, and that it was manifested ing, the learning being manifested as increases
as a decrease in the effort expended. in speed with constant accuracy.
 Motor Learning Concepts and Research Methods mi 341

Memory and Learning In addition to data on the initial performance

Inherent in the notion of learning is the concept of in the retention test, useful information can be
memory (chapter 14). In fact, most experimental found in the patterns of improvement during
psychologists define learning in terms of memory, retention-test performance. Typically, group B
saying that something has been learned when a would be expected to improve at a faster rate than
person has a memory of it. In this sense, memory group A, reattaining the ceiling somewhat more
~nd ~abit are very similar constructs. Memory
quickly as a result (and also because they had
is evident when one has learned a skill or can less loss in the first place). But another measure,
perform it again at some time after the original- first discussed long ago by Ebbinghaus (1913)
and more recently by Nelson (1985) and Seidler
practice session.
(2007), is the savings score (see also chapter 14).
With respect to the problem of overlearning, if
Here, the amount of additional practice "saved"
the group with overlearning trials was practicing
in reaching some criterion score (in this case, the
at the ceiling, the two groups should differ on a
ceiling) on the retention test is measured. From
retention test given some weeks later, perhaps
figure 10.10, if group A required, for example,
producing a pattern of results something like
25 trials to reach the ceiling again, and group B
that shown in figure 10.10. Here, the group with
reached the ceiling in about 10 additional trials,
overlearning trials should outperform the group
overlearning would have produced an average
without these trials on the first retention-test trial.
"savings" of 15 trials. Savings scores are also of
Both groups would have lost, in this hypothetical
considerable practical interest in industry and
example, some of what they had learned in the
the military because they indicate the amount of
original session; but the group without the over-
additional (time-consuming and costly) practice
learning would be farther below the ceiling than that will be needed to return people to various
the group with the overlearning trials, leading to criteria of "readiness" for a particular job.
the conclusion that a stronger memory (habit) for
the task existed after the overlearning condition. Generalizability as a Learning
Evidence for a stronger habit after a retention
interval could indicate that learning continued
Criterion
during the overlearning trials, even though the We normally think of learning as having the goal
ceiling prevented a change in performance scores of improving our behavior on a particular move-
in these trials. ment (e.g., to serve more effectively in tennis),

Ceiling

Cl)
u
c
I'll
...0E
'!::
:.

Overlearning Retention test

Practice trials

FIGURE.10.10 Retention as a ~easure of original learning. (Original learning and overlearning trials are shown at left with
hypothetical sc~res fro?l a retention test shown at right; the finding that B outperforms A on the retention test indicates that B
learned something during the "overlearning" trials even though no change in task performance could be seen.)
342 fill Motor Control and Learning

but broader benefits of practice should also be complex, having many forms in many different
recognized. One of these is generalizability-the situations. Above all, we stress here the deficiency
extent to which practice on one task contributes in the oversimplified idea that learning is merely
to the performance of other, related skills, perhaps a change in behavior on the task in question. It
in different contexts. As highlighted in chapter is clearly a much more complex set of processes
6 in the section on generalized motor programs than that.
(GMPs), our capability to perform a task like
throwing is not based on one particular throwing
movement. Rather, we appear to be able to use Issues About the "Amount"
a generalized throwing program for a variety of of Learning
throwing tasks, but with the selection of appropri-
ate parameters for the kind of object to be thrown On the basis of experiments on learning, research-
as well as parameters to achieve the distance and ers are often tempted to make statements phrased
trajectory of the throw. We could regard throw- in terms of the amount of learning that has
ing practice as having the goal of contributing occurred as a result of practice. For example, we
to one's overall throwing capability as well as to might wish to say that a group of subjects practic-
the skills actually practiced. In this sense, we can ing with one condition learned 20% less than a
measure the effectiveness of a practice session not group with another condition. Or you may wish
only by how well the particular skills practiced to say that Luc learned twice as much as Jack
are acquired, but also by how well performance on this task. Do such statements really have any
improves on similar skills (that are not practiced meaning?
directly). This would involve measuring perfor- The problem is that habit is a construct that
mance on other similar skills in a transfer test- cannot be observed directly. Usually little basis
analogous to the measurement of the retention exists for making quantitative statements about
characteristics of some task as in figure 10.10. it, because it can only be estimated from perfor-
The acquisition condition producing the most mance scores. Recall the experiment by Bahrick
effective performance in this transfer test would and colleagues (1957) discussed earlier. By a
be judged as having the highest generalizability. simple change in the way the task was scored
We can therefore think of the measurement of (change in the strictness of the criterion of suc-
generalizability as another in a group of alterna- cess), the authors obtained almost any scores
tive measures of learning. The details of these they wanted, and they could easily change the
evaluations are related to the measurement of shapes of the performance curves. For example,
transfer, which we consider in detail in chapter 14. looking back at the data from their experiment in
figure 10.5, we computed that the subjects with
Summary of Alternative Learning the 5% scoring criterion improved 86% as a result
Assessments of practice, that the same subjects improved 29%
with the 15% criterion, and that they improved
All these methods are consistent with the funda- 8% with the 30% criterion. How can we assess the
mental notion that learning is the set of internal, "amount" of habit gain that occurred here? Will
unobservable processes that occur with practice we say that it was 8%, 29%, or 86%? That deci-
or experience, resulting in a changed underlying sion is purely arbitrary, and we would be equally
capability for moving. Therefore, it is not sur- wrong in making any of these claims.
prising that even when obvious changes do not The best one can do is to make statements
occur in the main task, these learning processes about the relative amounts that two groups have
can be demonstrated by a variety of other means, learned, or about the relative amounts that two
such as decreases in the interference created for people have learned, essentially with statements
other simultaneous activities, reduction in effort, like "group A learned more than group B." The
increases in the speed with which a main task issues are somewhat different for the case in
demanding accuracy is completed, changes in which groups are compared to each other versus
retention capabilities, or alterations in generaliz- the case in which individuals are compared
ability to other similar skills. These techniques against each other, so we discuss these two situ-
highlight the idea that learning is internal and ations in turn.
 Motor Learning Concepts and Research Methods 111 343

Group Differences formance and hence different levels of initial

capability. This initial difference could be attribut-
Figure 10.lla is a typical example of the trans- able to some systematic difference in the nature
fer paradigm discussed earlier, where practice
of the subjects (e.g., males vs. females or children
tends to increase the scores on the task. Here,
vs. adults), or it could be attributable to simple
the experimenter does not really have any idea
random sampling effects, whereby the groups
about how much learning occurred, but can make
simply differ by chance. In either case, both
statements about two things: the fact that both
groups learned, and the fact that group A learned groups learned, but we are in a difficult position
more than did group B. Because the two groups with respect to saying which group learned more.
began practice on trial 1 with the same level of You can see that both groups gained about the
performance, and because, during the course of same amount in terms of the score on the task, but
practice, they progressed to different levels of you have no way of knowing whether the amount
performance, the conclusion that group B learned of change in the capability (i.e., habit) was larger
less than group A in this case seems unavoidable. or smaller for group A or group B.
Remember that the test of the relative amount The problem is that the differences in the initial
learned should always be on the transfer trials, as level of performance have confounded interpre-
it is only there in the sequence that the temporary tation about which of the groups learned more.
effects of the independent variable are equated Thus, in designing these kinds of experiments it
for the two different levels of initial capability. is essential, if this question is to be asked about
Now consider figure 10.llb, where the two the data, to be certain that the two groups of
groups began practice at different levels of per- subjects are equated at the beginning of practice

Original learning Transfer

!!!
0
(,)
en

b Trials

FIGURE 10.11 Outcomes from an experiment on learning in which the original level of proficiency for the two groups was
(a) controlled or (b) not controlled.
344 Motor Control and Learning

so that differences in the performance on the too large this adjustment will be, and we have
transfer test can be attributed to changes in the no way to find out, such adjustments probably
amount learned. 1 This can be accomplished by should not be used.
administering a pretest on the learning task, Of course, these adjustments are exactly
ranking subjects from highest to lowest on this equivalent to computing "learning scores," or
test, and then assigning the subjects with odd- the differences between the initial and final per-
numbered ranks to one group and those with formance for one group, and comparing these
even-numbered ranks to the other group. In this differences to the difference between the initial
way the groups are almost exactly equal on the and final performance scores for the other group
first trial, and the problem is eliminated. (Other (Schmidt, 1971b). The use of "learning scores" or
matching techniques are available as well.) "gain scores" is not as prevalent as it was some
There are some drawbacks to using a pretest. years ago, largely for the reasons mentioned.
One is that the pretest provides some practice on
the experimental task, which then reduces the size Individual Differences in Learning
of the changes to be observed under the different
The problem just raised is similar in many
levels of the independent variable. This procedure
has the additional limitation that the first trials on respects to that of comparing the amount learned
a motor task are usually not strongly correlated by one person to the amount learned by another
with the last trials in the sequence-that is, with person. This is a common problem in the study of
the final trials on which the relative amount of individual differences in learning, as sometimes
learning is assessed. Such correlations are fre- it is of interest to determine whether or not some
quently as low as 0.20. If so, then the variable measure of an ability correlates with the amount
that is matched (initial performance) is not going that someone will learn (e.g., Bachman, 1961, and
to influence the "equality" of the groups on the many others). Naturally, in order to compute the
transfer trials to a very strong degree. correlation, one must achieve a measure of the
A second method is to use large groups that "amount learned" by each person. But we have
are randomly formed from some larger group. just argued that such measures are not meaning-
In this way the groups will be expected to differ ful if they are based on the differences between
due to chance effects; but if the groups are large the initial and final performance.
(e.g., 25 people per group), then the chance dif- A second situation in which these kinds of
ferences in the group means on the motor task measures are usually taken is in determining
should be relatively small. This procedure has the grades. Often, a measure of the student's initial
advantage that it does not give subjects experi- level of proficiency is taken at the beginning of
ence on the task before the independent variable instruction, and another measure of proficiency
is administered, but it has the disadvantage that is taken after the instruction has been completed;
chance differences can still occasionally occur. grades are then assigned on the basis of the dif-
More on this issue can be found in Schmidt ference between the initial and final performance
(1971a, 1972a). levels, presumably in terms of" amount learned."
Could you "correct for" the initial differences The fundamental problem with this technique
in performance (as seen, e.g., in figure 10.llb) by is that the difference between initial and final
subtracting the initial difference from the differ- performance is not an adequate measure of the
ences that occurred in the transfer trials? While "amount learned," for the reasons already dis-
arithmetically this can be done, such procedures cussed. Also, in philosophical and educational
produce nearly meaningless interpretations of terms, in most aspects of life we are judged on the
the relative amount learned. The problem is that basis of what we can do, not on the basis of how
the sensitivity of the scoring system to changes much we have improved. If someone received an
in the internal capability is different at various Ain physics because she could pass the examina-
places on the performance scale. Thus, subtract- tions at the end of the course, no one cares that
ing the initial differences in performance from the she could have passed the examinations on the
differences in performance at the end of practice first day of class. Evaluating improvement scores
will probably lead to an adjustment that is far in teaching or research settings is generally laced
too large. Because we have no idea how much with difficulty, and should be avoided.
 Motor Learning Concepts and Research Methods iii 345

"Rate" of Learning rarily, and which affect learning, allows the pro-
duction of more effective settings for instruction
We have argued that using scores from acquisi- in various motor tasks in sport, industry, therapy,
tion data to calculate measures of "amount" of and so on. Naturally, if the goal of practice in
learning leads to many problems in interpreta- such application areas is to maximize learning,
tion. A problem similar to those already noted variables that influence learning (as measured on
arises in the measure of "rate" of learning. The retention or transfer tests) should be emphasized
idea of a learning rate has been common in sport during practice, and variables that influence
folklore ("so-and-so is a fast learner"). Attempts performance only temporarily can be ignored.
to measure the speed of acquisition are often The topic of this relation between research and
based on the steepness or slope of the perfor- practice, as it pertains specifically to issues of
mance curve. For instance, look back at the plot learning, will arise numerous times throughout
of the two hypothetical subjects in figure 10.3. the remainder of the book.
It might be tempting to say that subject 2 is a
"faster" learner than subject 1 because the slope
of the curve is much steeper in the early stages Summary
of practice. However, such conclusions fall prey
to exactly the same problems as with trying to The study of motor learning is considerably dif-
measure the amount of learning. ferent from the study of performance in that the
focus is on the changes in performance that occur
Understanding Learning as a direct result of practice. Motor learning is
defined as a set of internal processes associated
and Performance Variables with practice or experience leading to a relatively
permanent change in the capability for skilled
The problem of identifying which of the many behavior, a state sometimes termed habit. Such a
independent variables are critical for learning, definition must be carefully worded to rule out
and which are relevant only for performance, is changes in behavior that are due to maturation
important not only for development and testing or growth, or to momentary fluctuations in per-
of theories of motor learning, but also for applica- formance attributable to temporary factors.
tion to a variety of practical situations. We treat In the typical motor learning experiment, two
these two issues briefly here but discuss them or more groups of subjects practice a task under
again throughout the remainder of the text. a different level of an independent variable. A
common method of data analysis involves per-
Importance for Theory formance curves, or plots of average performance
on each trial for a large number of subjects. These
Learning theories make predictions about how
curves can hide a great deal of important informa-
certain independent variables will affect learning.
tion about learning, however, such as individual
In such cases, experimental tests of the theory differences in learning or changes in strategies.
involve evaluations of, among other things, the They tend to characterize motor learning as a
extent to which one group of subjects learned slow, constantly evolving process requiring con-
more than some other. As we have argued, valid tinued practice, whereas other evidence suggests
tests of such questions ultimately rest on which that learning is often sudden, insightful, or even
of the two groups performs most effectively on "revolutionary." As a result, interpretations about
retention or transfer tests. Thus, such research
the nature of learning from performance curves
designs become the key way in which theories
must be made carefully.
of learning are tested in the laboratory. Many Learning experiments usually involve what is
examples of this kind of theory testing are pre- called a transfer (or retention) design, in which the
sented in the following chapters. groups of subjects practicing at different levels
of the independent variable are transferred to
Importance for Application a common level of that variable. These designs
A second practical outcome is that knowledge provide for the separation of the relatively perma-
about which variables affect performance tempo- nent effects (due to learning) and the temporary
346 Motor Control and Learning

effects of the independent variable. Those inde- in chapter 11 or 12, describe how an
pendent variables affecting performance "rela- experimenter might devise a double-
tively permanently" are called learning variables, transfer design to assess learning.
and those affecting performance only temporarily 2. Find a research article that illustrates one of
are called performance variables. the hypothetical learning and performance
In many situations, the performance scores are effects depicted in figure 10.7.
near a ceiling or floor, at which no changes can
occur because of task-imposed or biologically
imposed limitations on performance. In such situ- Web Resources
ations, a number of secondary-task methods can
be used, such as measures of latency, measures The Measurement in Physical Education and Exercise
of attention or effort, measures of retention, or Science journal is devoted to measurement issues
measures of generalizability. Even with all the in kinesiology.
methods, it is seldom possible to speak meaning- www.informaworld.com/smpp/title-content=
fully about the actual amount a person or a group t775653683-db=all
has learned or the rate of learning.
The Journal of the Learning Sciences is devoted to
the study of learning and education.
Student Assignments www.informaworld.com/smpp/title-content=
t775653672-db=all
1. Prepare to answer the following questions
during class discussion:
a. Motor learning is defined in terms of
Notes
four distinct characteristics. Use an
1 Some research examples presented in chapters 11
example of learning a real motor skill
and 12 will appear to show group differences on the
to illustrate these four characteristics. very first trial (e.g., Shea & Morgan. 1979). In these
b. In your own words, describe how cases, however, it is important to recognize that the first
you would explain to an athlete the data point on the graph does not correspond with the
distinction between performance and first trial but rather with the first block of trials, which
may represent the average of 5-10 or more trials. Thus,
learning. Why might this distinction
the learning variable is revealing rapid effect on the
be important to the athlete? performance of the groups, which we assume, were
c. Using an example of research discussed equated on trial 1.
 11

ime is a key constraint in many situations schools, in training for jobs in industry or the mili-
in which individuals are asked to learn (or tary, and in rehabilitation (e.g., see "Constraint-
relearn) motor skills. An insurance company may Induced Physical Therapy" on p. 350). Also, we
dictate that a maximum of nine rehabilitation stress those variables in which there is the greatest
sessions will be paid by insurance. A badminton theoretical interest. This emphasis also provides
practicum may involve three sessions per week a strong contribution to practical application,
for 14 weeks. A tool and die worker is given because well-established theories have many real-
one training session to learn to use a new piece world implications (Kerlinger, 1973). Generally,
of equipment. A microsurgeon takes a two-day the chapter is about attempts to understand the
course involving a new technique before surgery many variables that determine the effectiveness
on patients begins. In all of these situations that of the conditions of practice.
invoke specific temporal limitations on practice,
the implicit understanding is that practice should
be organized in a way that maximizes the amount The Law of Practice
of potential learning. To meet the demands of
these time constraints, the learning facilitator One practice variable dwarfs all the others in
must be aware of the variables, or the conditions terms of importance and is so obvious that it need
of practice, that influence performance and learn- hardly be mentioned at all-the amount of prac-
ing, and adjust them so that learning will be tice. All other things being equal, more learning
maximized. will occur if there are more practice trials. In fact,
This chapter is about these variables and tech- so pervasive is the effect on learning that this is
niques, with the focus on the major ones that are simply called the "law of practice."
important for motor learning. There are many One of the most frequently observed charac-
such variables, and we have confined the discus- teristics about the change in performance that
sion to those variables having the largest effects accompanies practice is that the improvements
(i.e., those that make the biggest difference) and in (average) performance are generally large and
those that are usually under the direct control of rapid at first and become systematically smaller
the experimenter or teacher. With this emphasis, as practice continues. Thus, whether the measure
the material relates rather closely to the design of performance increases or decreases (see figures
of instructional settings such as would be seen in 10.1and10.2) with practice, performance curves

II! 347
348 Motor Control and Learning

are usually negatively accelerated functions of good general, rough indicators of the relationship
practice (i.e., the "rate" of improvement changes between practice and performance. They are,
toward zero as practice continues). Although a after alt just performance curves-not learning
few particular tasks might not show this kind of curves, as we stressed repeatedly in chapter 10.
relationship, or certain performance phenomena
might distort it considerably (e.g., fatigue), as a Generality of the law of Practice
general rule the majority of the behaviors studied Fitts (1964), and later Newell and Rosenbloom
in motor learning seem to show practice curves (1981), pointed out that practice on numerous tasks
with this overall form. with widely differing movement measurements,
goals, and measures of performance-motor tasks
Time Scales for the law of Practice and verbal tasks alike-tended to follow this loga-
A negatively accelerated relation between per- rithmic power function. The first analysis of this
formance and practice trials has many of the kind was presented by Snoddy (1926). Subjects
general features of a very common equation that learned to draw figures while viewing only the
is termed a power function (or log-linear function), mirror image of their drawing hand, with practice
in which the time, T, to complete an action (where continuing over 100 days (one trial per day). The
performance is measured in terms of time) can data, plotted in figure 11.1 using a power func-
be expressed as tion, show a generally linear relationship, except
perhaps for the first few data points. This is an
(11.1)
example in which the score increases with practice.
where a and bare constants, and Pis some mea- Another example comes from Crossman (1959) in
sure of the amount of practice (e.g., number of a study of factory workers who made cigars using
trials). Here, as practice increases and P becomes a small hand-operated jig. In figure 11.2 is a plot
larger, the ratio a/P decreases, resulting in smaller of the average time for cigar production against
time to complete the action, T; the larger the con- practice time (a decreasing functiont also plotted
stant b, the more "rapid" are the decreases with as a power function. This plot is also generally
practice. Analogous power functions can also be linear, only flattening out near the region of high
defined for tasks with scores that increase with practice where the minimum cycle time of the
practice, such as the pursuit rotor or Bachman machine itself became a factor in performance.
ladder task (Bachman, 1966). For these the sign Fitts (1964) and Newell and Rosenbloom (1981)
of bis simply reversed. provide more examples and a fuller discussion of
One important feature of power functions such power functions in learning.
as that in equation 11.1 is that the plot of the loga- An important interpretation of these logarith-
rithm of performance (T) against the logarithm of mic relationships is that the rate of improvement
the number of practice trials (P) will yield a linear at any point in practice tends to be linearly related
function. For example, taking the logarithm of to the "amount left to improve" in the task. So,
both sides of equation 11.1 yields early in practice, when there is much learning left
to accomplish, the speed of improvement is very
Log (T) = -b (Log P) +Log (a) (11.2)
rapid as compared to that at the end of practice
Notice that, from the discussions of empirical when there is not so much "room for improve-
equations earlier in the book, equation 11.2 is ment" remaining. Furthermore, many of the
simply a special case of the standard equation experiments in support of the law of practice show
for a linear function, Y = a + b X, where here Y that improvements continue to occur for years,
is Log (T), a is the constant Log (a), and Xis Log even though these later gains may be very small.
(P). Therefore we can summarize equation 11.2 Consider, for example, the situation in Crossman's
easily by saying that the relationship between (1959, figure 11.2) data, where improvements can
Log (T) and Log (P) is linear, with an intercept of be seen even after seven years of practice, and
Log (a) and a slope of -b. Note again that many after 10 million cigars were rolled! Data like these
variables distort these performances (and hence suggest that learning is never really completed,
these kinds of relationships) so that these power even in the simplest of tasks. Statements like,
functions are not very useful for studying and "John learned to drive last week" lose a great deal
measuring learning; but these functions are still of their meaning when seen in this light.
 Conditions of Practice 111 349

40

!!! 20
gg
0 CD
,.... +
Ill

II
!0
~ 10
5
0
2 3 4 5 10 20 30 40 50 100
Trials (log scale)

FIGURE 11.1 Scores in a mirror-tracing task as a function of extended practice.

Reprinted from G.S. Snoddy, 1926, "Learning and stability," Journal of Applied Psychology 10:1-36.

30

20
'CD
iii
u
Ill
Cl
.2
~
~ 10
CD
g,
0

10,000 100,000 1,000,000 10,000,000

Number of items produced (log scale)

FIGURE 11.2 Completion time in making cigars as a function of extended practice.

Reprinted, by permission, from E.R.F.W. Crossman, 1959, 'Iii. theory of the acquisition of speed skill," Ergonomic 2: 153-166, reprinted by permission of the publisher (Taylor &
Francis Ltd, http://www.informaworld.com).

Although the power function has been used we cautioned in chapter 10, however, group data
to describe motor learning data sets for more can sometimes misrepresent the performance of
than 80 years, it may not necessarily be the opti- the individuals for whom the data have been
mal function to describe the law of learning. A averaged (see figure 10.3 as an example). When
reanalysis of many of these earlier data sets by the published data for each of the subjects were
Heathcoat, Brown, and Mewhort (2000) revealed analyzed separately, Heathcoat and colleagues
that the power function best described the aver- found strong evidence that an exponential function
aged data of all participants of the experiment. As provided a better fit for the law of learning than a
350 Motor Control and Learning

power function. Indeed, a replication of Snoddy' s practice trials and performance. As we discussed
original experiment (1926) revealed much better in chapter 10, this relationship may, but does not
fits for an exponential function than a power necessarily, provide a description of the progress
function (Stratton, Liu, Hong, Mayer-Kress, & of learning-that is, the underlying capability
Newell, 2007), leading these researchers to sug- for performance that is the goal of practice and
gest that many different time scales may be used learning research. Later in this chapter we discuss
to describe learning, depending on the interac- various conditions that alter the shapes of these
tions of the individual with different aspects of performance curves in practice, and we show
the task and practice-related factors (Newell, Liu, how these performance alterations may not be
& Mayer-Kress, 2001, 2009). closely related to the underlying changes in learn-
Regardless of which function best fits the data, ing. The law of practice is interesting and robust,
it is important to remember that the law of prac- but we should be cautious in using it to generalize
tice is just a description of the relationship between about the course of learning over practice.

Constraint-Induced Physical Therapy

In the fields of physical therapy and rehabilitation medicine, a relatively new approach has emerged
in the treatment of individuals following a stroke. The patients who have volunteered for these studies
ar€ typically many months (sometimes over a year) poststroke, when the recovery of motor function
in. these individuals is typicalfy considered to be finished. The new therapy consists of "constraining"
the use of the less affected limb (the "better" limb), for example by asking the patient to wear a sling
or a large oven mitt on the "good side" for many hours of the day. In this way the patient is "forced"
•,to use the affected limb to carry out activities of daily living.
·Experimental interventions of this type have been remarkably successful in facilitating further
recovery of function in the affected limb (see reviews in Wolf, Blanton, Baer, Breshears, & Butler,
2002;Taub, Uswatte, & Pidikiti, 1999). Furthermore, the motor recovery from these interventions has
been shown to be retained for at least two years (Wolf et al., 2006, 2008). There is some controversy,
however, over the reasons why this type of therapy is effective.
One interpretation (e.g., Taub et al., 1999) suggests that, following a stroke, patients learn strate-
gies that. tend to "replace" the goal-directed movements that would normally have been done by the
affected limb, Many of these strategies result in using the "better" limb more frequently, essentially
resulting in a reduced need to use the affected limb in the future. According to Taub, each success-
ful completion of an activity by means other than using the affected limb reinforces ("rewards") the
nonuse of that limb. Thus, constraint-induced therapy is seen as a method to overcome "learned
nonuse" by inducing the patient to use the affected limb.
An alternative view more closely resembles the principles of practice, suggesting that improvement
in motor skill is a function of an increasing accumulation of practice trials, coupled with movement-
produced feedback. By this rationale, the patient who compensates for the loss of function in the
affected limb by discontinuing use of that limb has essentially ceased to accumulate practice trials,
or is even reversing this trend (causing forgetting), which might make that limb's function even more
degraded than normal. Constraint-induced therapy is seen as a method for improving the acquisi-
tion of motor skill by forcing the accumulation of more practice trials (similar in many respects to
the definition of deliberate practice) (Wolf et al., 2002). Similarly, this notion is in keeping with the
view that, among perhaps many other things, physical, occupational, and speech therapy consist
of practice, whose principles are a part of the motor learning ideas we present here (Maas et al.,
2008; McNeil, Robin, & Schmidt, 1997; Schmidt, 1991 b; Winstein, 1991 ). Whatever the mechanism
at work here, the implications of constraint-induced therapy for the efficacy of therapeutic interven-
tion and theory development hold a great deal of promise.
 Conditions of Practice 111! 351

Deliberate Practice and how alterations affect performance and

learning. Remember to always keep in mind
What does the term "practice" really mean? the distinction between factors that influence
Ericsson, Krampe, and Tesch-Romer (1993) note temporary improvements in performance and
that the term "practice" can take on various those that have relatively permanent effects
meanings, and they advocate the term deliber- (i.e., on learning). This distinction is critical in
ate practice to denote the kinds of activities that a number of research areas discussed in this
instructors, teachers, and therapists use on a chapter and the next.
daily basis. Specifically, Ericsson and colleagues In the next sections we concentrate on the
define deliberate practice as activities that have
/1
various ways in which learning can be enhanced
been specially designed to improve the current when the learner is not physically engaged in
level of performance .... Deliberate practice practice of the actual task. Much of this discus-
requires effort and is not inherently enjoyable. sion-but not all of it-concerns those factors that
Individuals are motivated to practice because have been shown to operate before the practice
practice improves performance" (p. 368). But session begins-involving the preparation of the
research has not been conducted that compares learner for the upcoming practice sequence.
the long-term effects of years of deliberate prac-
tice versus other forms of practice (e.g., play). So,
it is unknown whether the laws of practice apply Motivation for Learning
to deliberate practice only, or whether there are
different functions for different forms of practice, Since practice is the most important factor in
and so on. learning, as discussed in the previous section,
In the remainder of this chapter we will dis- an important point is that motivation is a primary
cuss methods of deliberate practice that can be determinant for engaging in practice at all and
modified, structured, improved, and otherwise thus is important for learning. If the learner per-
changed so as to influence performance. Figure ceives the task as meaningless or undesirable,
11.3 provides a conceptual overview of these then learning of the task will probably be mini-
methods. Off-task practice considerations refer mal. If the level of motivation is too low, people
to those factors that are generally applied when may not be sufficiently motivated to practice at
the learner is not actively engaged in physical all, and very little, if any, learning will be the
performance, such as the effects of modeling result. Aside from this rather obvious conclusion,
(or observation). On-task practice conditions a number of reasonably complex determinants
generally refer to the methods by which the of learning exist as a function of motivational
physical engagement in practice can be altered, level.

FIGURE 11.3 Practice can include on-task situations, as well as conditions in which the learner does not physically inter-
act with the task.
352 Motor Control and Learning

Making the Task Seem Important kinds of goals (1) focus one's activities, (2) help to
regulate the effort directed toward these activities,
Before beginning the practice session, it is impor- (3) help maintain vigilance in attempting to reach
tant that the learner see the task as one that is the goals, and (4) serve as a referent against which
desirable to learn. Much of this kind of motiva- achievement can be compared.
tion appears already to be established in people Goal-setting studies in sport and exercise
because of culturally derived emphases on certain research, however, have not consistently sup-
activities. But in many cases it is important to ported Locke and Latham's conclusion that spe-
show why it would be useful to have a certain cific, "difficult" goals are the most effective ones
skill. A teacher once pointed out that the usual for performance. Problems specific to conducting
all-male games (e.g., tackle football) were fine research in sport and exercise in relation to goal
for playing among young males, but that a time setting may account for these inconsistent effects
would come in life when less violent recreational (see also Locke, 1991; Burton et al., 2001). Since
skills might be preferred. Such a comment empha- sport and exercise tasks are often intrinsically
sized the importance of learning these other activi- motivating and competitive, subjects may have
ties. Similarly, some might consider badminton a a vested interest in their outcome when they
rather leisurely game that is played in the back- serve as participants in a research study. For
yard on warm summer days-until, that is, they example, participants who are told not to set a
see a game involving highly skilled players. A specific goal or are told to use a vague goal may,
person might be much more interested in learning in fact, secretively set specific, perhaps" difficult"
these skills after seeing them performed by role goals, thereby masking the potential impact of
models in world-class competition. goal-setting variables. Thus, the smaller overall
effects of goal setting in sport and exercise studies
Goal Setting
do not necessarily mean that goal setting is not
Another frequently used motivational technique an important factor in the performance of motor
is goal setting, whereby performers try to attain skills (cf. Burton et al., 2001; Weinberg, 1994). The
goals that are set before they begin practicing. fact that the tasks in these studies lend themselves
This technique has been studied extensively in to intrinsic goal setting may overshadow the
research related to industrial and organizational experimenter's manipulations, making it difficult
psychology (for reviews see Locke & Latham, to detect the true effects of goal setting.
1985, 2006; Tubbs, 1986), and more recently in Kyllo and Landers' (1995) quantitative review
sport (for reviews see Burton, Naylor, & Holliday, (meta-analysis) of the literature suggests that
2001; Locke, 1991; Weinberg, 1994) and rehabilita- one small but consistent finding does appear
tion (Gauggel & Fischer, 2001; Playford, Siegert, in the sport and exercise research. The authors
Levack, & Freeman, 2009), suggesting some impor- found that specific, absolute goals of moderate "dif-
tant principles for performance and learning. ficulty" were beneficial to the performance of
Probably the most intuitively appealing and sport and exercise tasks. In addition, there was
most often used goal in all types of daily activi- evidence that setting short-term goals and setting
ties is the do-your-best goal. Encouragement such a combination of short-term and long-term goals
as "Try to get as many done as possible," "Do the facilitated performance as compared to setting
best job you can," and "Give it 100% out there" is long-term goals only.
typical of these types of goals. However, research One limitation of the research literature is
suggests that "do-your-best" goals are not as effec- that performance and learning effects often have
tive as other types of goals. Locke and Latham not been separated. Although most of the goal-
(1985) summarized the psychological research and setting literature is dominated by studies on
suggested that specific, "difficult" goals produce performance, there have been some studies that
stronger performance than either no goals or do address goal-setting effects on learning. For
vague, ambiguous goals such as "do your best." example, Boyce (1992) examined three groups of
Locke and Latham proposed four advantages subjects who set different goals while they were
of setting specific, "difficult" goals; these can be learning to shoot a rifle. One group was told to
considered analogous to the advantages of set- "do your best." Another group was encouraged
ting an agenda for an important meeting. These to set their own specific goals. A third group was
 Conditions of Practice 353

assigned specific goals that were made progres- teacher or instructor. An early study by Solley
sively higher on each practice day. The goal- (1952) demonstrated the long-lasting effects of ini-
setting procedures were established following a tial orientation. In this study, three groups of sub-
pretest and were applied on five days of practice jects practiced a two-step lunge to stab at a target.
sessions over a three-week period. A retention test Over six days of practice, the various groups were
assessed learning one week after the last practice encouraged to perform the action with an empha-
session. The results, shown in figure 11.4, revealed sis on either speed or accuracy or with an equal
that the group of subjects told to "do your best" emphasis on speed and accuracy. The results were
performed slightly better than the other groups quite dramatic. The group instructed to emphasize
on the first practice day, but worse on the other accuracy was the most accurate over practice; the
days. These differences were maintained in a group instructed to emphasize speed was the fast-
retention test, suggesting that the specific goal- est; and the group giving equal emphasis to speed
setting procedures had beneficial effects on both and accuracy performed at intermediate levels on
performance and learning. both measures. A strong feature of the study was
that it included transfer tests over six additional
days of practice, in which all groups were encour-
Verbal Information aged to give equal emphasis to speed and accuracy.
The effects seen during practice were maintained
Assuming that an individual is motivated to in transfer, especially for performance speed-the
learn, many believe it important to give the instructional "set" had a lasting impact on both
person some information about the task to be performance and learning.
learned before actually physically practicing the Instructions can also provide useful and
task. There are many ways to do this, of course, important information about the movement itself,
and some of these are discussed next. such as the initial positions of the limbs in rela-
tion to an apparatus or an implement used, the
Instructional Set stance, what to watch and listen for, and what to
One of the most common ways of giving students do. Perhaps more importantly, an overall "idea"
an initial orientation to the new skill is through or image of the movement can be conveyed that
verbal instructions, usually presented by the can serve as a representation for the first attempt.

45

40

35
...0
(I)

CJ
111
Cl 30
s:
~
0
tij 25

20

Pretest 2 3 4 5 Retention
Session

FIGURE 11.4 Effects of goal setting on learning a shooting task.

Adapted, by permission, from B.A. Boyce, 1992, "Effects of assigned versus participant-set goals on skill acquisition and retention of a selected shooting task;' Journal of
Teaching in Physical Education 11 (2): 227'.
354 Motor Control and Learning

Also, instructions can emphasize the ways in

which one can recognize one's own errors-for 7
example, "After the movement, check to see that
your arm is straight." These kinds of instructions
~6
can serve to stimulate the development of error-
detection capabilities (see chapter 13). "'~
Cl
!s
Focus of Attention w
en
~
In chapter 4 we discussed the effects of instructed cc 4
attentional focus on performance. Wulf and her
colleagues have conducted a number of experi-
3
ments using attentional focus instructions in
learning experiments and have demonstrated
Last trial of Retention Transfer
powerful effects that are both immediate and acquisition trials trials
long-lasting (for reviews see Wulf, 2007; Wulf &
Prinz, 2001). Most of these studies used novice
FIGURE 11.5 Effects of attentional focus instructions
subjects who learned various sport-related skills on stabilometer performance in acquisition, retention, and
or laboratory tasks. For example, in a study by transfer. RMSE is root mean square error so lower scores
Wulf, Weigelt, Poulter, and McNevin (2003), sub- are better.
jects learned to balance on a stabilometer (figure
2.Sc) while holding a tube with both hands. In
one experimental group (internal focus instruc- verbal instructions. Also, memory limitations
tions), subjects were told to keep their hands held (chapter 3) suggest that a learner can remember
horizontal; subjects in the other group (external only so many instructions, and few of these
instructions) were told to hold the tube horizontal; important points are assimilated on the first
and subjects in a control group were given no trials of practice. This problem seems even more
instructions. The results on the last trial of two critical when one is dealing with children, whose
days of practice (fourteen 90 s trials), four reten- attentional skills and memory capacities are much
tion trials one day later, and three transfer trials weaker than those of adults.
(without the tube) on the stabilometer balance One solution is to describe in words those
task are illustrated in figure 11.5. It is interest- aspects of the actions that are absolutely essen-
ing to note that in both the external and internal tial for the first trial or two, making sure that
groups, the instructions merely directed the learners achieve them in early practice. A dif-
subjects' attention to their hands, yet the effects ferent approach, used by coaches for years, has
were immediate and permanent: external focus been to use analogies to get the learner to think
of attention facilitated both performance and about how an action should be performed. For
learning as compared to internal focus. These example, basketball instructors may tell the
influences of attentional focus instructions on learner to finish the shot with a hand position
performance and learning have been replicated "like reaching into a cookie jar." Compared to
many times by Wulf and her colleagues, as well learning by explicit instructions, Liao and Mas-
as by others (e.g., Hodges & Franks, 2000). ters (2001) found that learning table tennis with
But as important as instructions are, they can analogy instructions resulted in less verbalizable
sometimes bias performance in undesirable ways knowledge, but transfer performance that more
(Hodges & Franks, 2001) or are overused in learn- strongly withstood the degrading effects of a
ing situations. Words alone are relatively crude secondary task. Similar findings have been found
descriptions of the complex kinds of movements for learning basketball skills (Lam, Maxwell, &
that a learner is attempting to achieve; just try to Masters, 2009).
describe the actions in pole-vaulting, or in tying Although more research in this area is war-
a bow-knot on your shoe, as examples. Only ranted, the effects of attentional focus and anal-
the most global, general aspects of the intended ogy learning may have similar bases. In both
movement are going to be transmitted through areas, instructions having the more positive
 Conditions of Practice m1 355

effects on learning (external focus and analogy Knowledge of Mechanical

learning) direct the learner's awareness away Principles
from how a movement should be regulated. And,
as we see in the next section, awareness appears to How much should learners know about mechani-
be unnecessary at all for motor learning to occur. cal properties underlying a particular movement
task? Will such knowledge, if provided before the
Learning Without Awareness learning of the task begins, be an aid to future
performance and learning? Early work by Judd
One of the assumed benefits of verbal instruc-
(1908) on dart throwing to targets submerged
tions is to make the learner aware of certain
under water gave initial indications that such
components of the task before practice begins.
information was useful. Judd taught his learners
But, is this even necessary? Some research sug-
the principles of refraction, whereby the light
gests that learning of some tasks can occur even rays from the submerged target are bent so that
though the subject is oblivious to the perceptual the target is not really where it appears, and this
regularities of the task. In a classic study in this provided initial advantages when the targets
area, conducted by Wickens and Pew (see Pew, were moved to different depths of submersion.
1974b), subjects practiced a pursuit tracking task, Similar results were obtained by Hendrickson
having to move a lever in order to try to match a and Schroeder (1941) in a task that required shoot-
perceptual input signal on a computer monitor. ing an air gun at underwater targets.
A trial comprised three 20 s segments. During Mechanical principles are also a part of
the first and last of these segments the perceptual instruction in sport skills, such as in swimming
signal was generated randomly by the computer, (propulsion) and billiards (ball spin, geometry).
and the random sequence was different on each However, explicit knowledge of mechanical prin-
of the 24 trials completed during each session ciples is not always necessary for performance of
on 16 days of practice. However, the perceptual certain tasks. In an often cited example, Polanyi
signal during the middle segment was always (1958) points to a champion cyclist who did not
the same. Improvements on the repeated, middle know the mechanical principles involved in the
segment were greater and occurred much faster maintenance of balance on a bicycle, implying
than on the random segments. More importantly, that such principles may not be critical for learn-
however, interviews with the subjects after day 11 ing the task. The implication here is consistent
revealed that none had any idea that the middle with the point made in the preceding sections-
portion of the trial was always repeating! It tha t at least some learning can occur in the
seems that the enhanced learning of the middle absence of explicit information (Krist, Fieberg,
segment had occurred even though the subjects & Wilkening, 1993).
were unaware that any part of the task had been
different from the others. These results have
been replicated in experiments in which the first Perceptual Learning
segment was repeated-again with none of the
subjects being aware of any regularities in the One of the features of expertise in many open
perceptual stimuli (see Magill, 1998). Studies have sporting events is the skill to use advance percep-
also shown that these effects persist in transfer tual information (reviewed in chapter 9). Based
tests, again without the subjects' awareness (Wulf on this evidence, another prepractice technique
& Schmidt, 1997). The result of experiments in is to expose the learner to the environmental
which learning occurs in the absence of explicit stimuli that will be experienced in the task. In
awareness or purpose by the learner has sparked this way, the temporal and spatial regularities
considerable research interest and debate, lend- can be presented, and the performance of the
ing support to the argument presented in the task with the actual stimuli may be more effective
previous section of fundamental differences in when the performer then begins to respond to the
the role of awareness in retention and transfer environmental stimuli physically (i.e., when the
characteristics of motor skills (e.g., Gentile, 1998; entire task is performed). As examples, a base-
Verdolini-Marston & Balota, 1994; see Hodges & ball batter can watch the flight of a curve ball a
Franks, 2002b; Masters, 2008, for reviews). number of times before swinging at one; and a
356 Motor Control and Learning

race car driver can walk through a road racing same, with the live or archived model providing
course before driving it. information about the task to be learned and
Adams and Creamer (1962) used a technique perhaps some essential details about technique.
like this in a laboratory situation. Subjects were Shrewd entrepreneurs have seized upon the
to learn a tracking task that involved moving a simplicity of this idea in marketing videotapes
lever to follow a stimulus dot in a regular sine- of expert athletes demonstrating their talents.
wave pattern. Before practice on the task began, But, while these specific techniques may seem to
some subjects were asked to watch these stimuli have intuitive appeal, there is reason to doubt the
and to respond by pressing a button every time learning benefit they provide (Druckman & Bjork,
the stimulus reversed direction, attempting to 1991). Nevertheless, there is mounting evidence
anticipate its movement. After considerable expe- that observation is an effective method for learn-
rience with this task, subjects were transferred to ing motor skills (for reviews see Ferrari, 1996;
the task in which the same stimuli were used; but McCullagh & Weiss, 2001; Maslovat, Hayes, Horn,
now, instead of pressing the button, they were & Hodges, 2010; Scully & Newell, 1985). In the fol-
required to make lever movements with the hand lowing sections we discuss some of the important
to follow the dot. Subjects who had experienced issues regarding observational learning.
the perceptual pretraining were more accurate People observe for different reasons. While one
in tracking than were subjects who had not had person may go to a concert solely to be entertained,
this experience. Similar findings were produced another may be there to observe the musicians
by Trumbo, Ulrich, and Noble (1965) when sub- play, perhaps trying to pick up skills to practice
jects learned to name each stimulus position in later. An aspect of this latter situation-how and
a regular series. When subjects had learned to what information is learned from observation-is
anticipate the order of stimulus positions through our interest in this section. Much of the early work
this naming activity, they were more effective in credited Bandura with the theoretical framework
responding to those stimuli in a tracking task (see for the development of this research (see Adams,
also Schmidt, 1968, for a review). 1987). However, some argue that Bandura's theory,
Perceptual training techniques represent a which was developed to explain the acquisition of
good off-task training option when physical social behaviors, is not appropriate for understand-
practice is either not possible or not practical. ing the learning of motor skills (e.g., Maslovat et al.,
Early research studies in which perceptual skills 2010), and recent advances in research specifically
in sport were trained provided promising results involving motor skill observational learning have
(Abernethy, Wood, & Parks, 1999; Farrow & provided new insights. Research on perceptual
Abernethy, 2002; Williams & Grant, 1999). These learning (see previous section) provides evidence
studies indicate that there is potential for learning that acquisition of environmental regularities is
from perceptual inputs, especially if the learner an important aspect of skilled performance. And
attends to and responds to them in some way the discovery of the mirror neuron system in the
rather than merely watching them passively. brain has elicited much excitement and research
Moreover, perceptual learning may be enhanced regarding the possibility that specific neural
by methods that attract the observer's attention to mechanisms provide the foundation for learning
the most relevant, advance perceptual informa- through observation (Rizzolatti & Craighero, 2004;
tion (Hagemann, Strauss, & Cafial-Bruland, 2006; Rizzolatti & Fabbri-Destro, 2010). These develop-
Jackson & Farrow, 2005). ments in research and theory have resulted in a
rapid increase in the amount of research conducted
Observational Learning on observational learning in recent years.

What Is Learned?
Learning from a model (observational learning)
is a specific kind of perceptual learning. One One of the earliest studies on modeling effects
important way to use modeling is to demonstrate showed quite clearly that movement strategies
the skill so that learners can observe the elements could be learned by observation. The task used by
of the action directly. Another variant is to use Martens, Burwitz, and Zuckerman (1976) involved
videotapes or photos of skilled performers. At a trying to move a ball rolling on top of two rods
superficial level, these procedures seem to be the by varying the distance between the rods. Two
 Conditions of Practice 357

strategies for success were modeled. The "creep- Participants in a study by Mattar and Gribble
ing" strategy was a conservative approach to the (2005) watched a video of a model who learned to
task whereby the distance between the rods was manipulate a robot arm in an environment where
adjusted slowly in order to move the ball. This the movement path was deflected away from the
strategy typically produces consistent, but only target by an invisible force field provided by a
moderately successful results. In contrast, the robot (see figure 2.16). Observers who watched
"ballistic" strategy involved a rapid and more vio- the model perform in the same force field environ-
lent adjustment of the rods. This strategy typically ment as they would later perform the task were
produces quite variable levels of achievement; more effective than no-observation controls, even
yet when scores are high, they can be very high. when attention was engaged in a secondary task.
When given the opportunity to perform the task, Timing is another type of information that can
observers of these two different modeling strate- be successfully learned through observation, and
gies tended to imitate deliberately the actions various methods for presenting temporal informa-
they had observed. Similar beneficial effects for tion have been used (Adams, 1986; Doody, Bird,
learning a three-ball juggling task by imitating & Ross, 1985; McCullagh & Little, 1989; Meegan,
the actions of the model were reported by Hayes, Aslin, & Jacobs, 2000; Zelaznik, Shapiro, &
Ashford, and Bennett (2008). Newell, 1978; Zelaznik & Spring, 1976). For exam-
Information that can be represented spatially ple, Zelaznik and colleagues provided one group
can be modeled quite readily, especially if the of subjects with the recorded sounds of another
spatial attributes are presented in a static, discrete subject making a correct timing movement prior
manner. For example, the acquisition of sign lan- to any practice. After the presentation of the
guage is a frequently used task for observational modeled information, this group performed the
learning because the actions are discrete and also timed movement more accurately than another
associated with a verbal label (e.g., Caroll & Ban- group who had not listened to the sounds. The
dura, 1990; Steffens, 2007; Weeks, Hall, & Ander- "listening group" could even improve slightly in
son, 1996). Spatial sequences represent another the task without any knowledge of results. The
type of task that appears to be readily learned interpretation of these findings is that the listening
through demonstration (e.g., Heyes & Foster, 2002; experience provided the subjects with a reference
Kelly, Burton, Riedel, & Lynch, 2003). Modeling of correctness, and the reference allowed an evalu-
of dynamic skills is also effective. For example, ation of the auditory feedback produced by the
a study of females with no prior dance training movement and subsequent adjustments on the
revealed more benefit after subjects watched vid- upcoming trials (see also Blandin & Proteau, 2000;
eotaped ballet sequences than after they looked Wrisberg & Schmidt, 1975). Timing information
at a series of still photographs of the dance (Gray, can also be successfully modeled when auditory
Neisser, Shapiro, & Kouns, 1991). The investiga- and visual information is presented in a spatial
tors concluded that the modeled information con- timing task (e.g., Vogt, 1995).
tributed to learning the qualitative features of the
ballet routine. These conclusions are supported by Model Skill Level
experiments that modeled skiing actions (Whit- One of the implicit assumptions regarding the
ing, Bijlard, & den Brinker, 1987) and gymnastics use of videos that provide sport instruction is
rope skills (Magill & Schoenfelder-Zohdi, 1996). In that highly skilled performers are more effec-
one practical study in this area, medical students tive than unskilled models. This assumption
performed surgical techniques more accurately may be true, depending on how the unskilled
after watching a skilled surgeon perform the model demonstrates the task to be learned. The
surgery, compared to students without the ben- evidence suggests that the assumption is not true
efit of observation (Custers, Regehr, McCulloch, when people observe unskilled models learn a
Peniston, & Reznick, 1999). This finding suggests motor task. For instance, Pollock and Lee (1992)
that even very fine spatial information can be suc- studied the effectiveness of demonstrations of
cessfully learned through demonstration. a computer video game to groups of individu-
Watching another's motor behavior can have a als who had had no prior practice on the task.
more subtle influence on the observer, sometimes Subjects either watched an expert perform the
perhaps even without the observer's awareness. task or watched a novice who was learning the
358 Motor Control and Learning

task for the first time (a learning model). After a from observing" the performance, but also from
/1

series of modeling trials, the observers in both observing the processing operations of the model
groups showed a substantial benefit of having in the attempt to improve performance, which
watched their respective models (as compared to becomes increasingly important as the difficulty
a no-model control group). However, there was of the task is elevated (Laguna, 2008).
no advantage to having viewed the expert model The research method used by Adams (1986)
compared to the learning model. Similar findings was extended by McCullagh and Caird (1990),
were reported by McCullagh and Meyer (1997) who directly compared the effectiveness of learn-
using a weightlifting task. ing models and expert models on Adams' task.
A more surprising finding in this research is Three observation groups were compared. One
that, under some circumstances, the provision of a group had repeated exposures to a tape of a perfect
learning model can result in stronger observational execution of the timing goal. Two other groups
learning than an expert model. This research area watched a tape of a model who was learning the
was initiated by Adams (1986), who used learn- task; one group also received the model's KR and
ing models to demonstrate the performance of a one did not. As illustrated in figure 11.6, the larg-
manual timing task. Adams found that observa- est effects were found for those who observed
tion alone was insufficient for learning this task. the learning model and also received the model's
However, considerable learning was seen if the KR (open squares). These subjects improved their
model's feedback from the experimenter (termed performance consistently over the acquisition
knowledge of results, KR) was also presented to the period, in the absence of any KR about their own
observer (Blandin, Lhuisset, & Proteau, 1999). performance, and both retained their performance
In this dynamic observation environment, the levels and transferred to a novel timing goal better
observer can gain information (a) from the model than either of the other observation groups.
about the movement performed (both visual These findings suggest an important applica-
and auditory), (b) from the augmented feedback tion to modeling real-world tasks. Novice athletes
presented to the model (as KR), and (c) from are likely to gain relatively little insight from
seeing the success of the model's attempt to use watching experts unless they are specifically cued
that feedback on the next performance of the regarding where to watch and what to look for
task. In this way, the observer benefits not only (Janelle, Champenoy, Coombes, & Mousseau,

500

I 400

e
Qj 300
$
:I
0
tn
~ 200

100

2 3 4 5 6 Immediate Delayed Delayed

retention retention transfer
Trial blocks

FIGURE 11.6 Effects of model skill level and availability of model's knowledge of results (KR) on learning.
Adapted, by permission, from P. McCullagh and J.K. Caird, 1990, "Correct and learning models and the use of model knowledge of results in the acquisition and retention of a
motor skill," Journal of Human Movement Studies 18: 107-116.
 Conditions of Practice 359

2003). While viewing professional golf on tele- Experimental assessment of mental practice effects
vision, for example, we likely get the greatest usually requires several different groups of sub-
learning benefit when experts make mistakes. jects, at a minimum (Goginsky & Collins, 1996): All
The mistakes occur so infrequently that the com- subjects are given a pretest on a task to be learned,
mentators usually replay the action and point out followed by the experimental manipulation, then
exactly what went wrong-what movement error a posttest on the learning task. The mental prac-
resulted in the flubbed shot. In other words, the tice manipulation often entails covert rehearsal of
"model" demonstrated an incorrect action, which the task, sometimes involving strategies such as
was accompanied by KR that identified the error. imagery. In this case, however, the learning that
Thus, the issue of essential importance in this can be attributed only to mental practice effects
research may not be the skill level of the model at cannot be inferred just from a retention test.
all, but rather about what type of information is Rather, one must demonstrate that performance
being demonstrated-errors or perfect templates on the posttest exceeded performance in a control
of an action. The research findings suggest that group that did not perform intervening practice
we learn more from observing mistakes than we or that performed practice on an unrelated task.
do from correct performances. In addition, mental practice is usually compared
to a third condition in which a group physically
Scheduling Observations practices the task for the same amount of time as
Although this part of the chapter is about "off- the mental practice group. Some experiments also
task" practice considerations, perhaps the most include combination conditions with alternation
successful strategy in the use of models is to between trials of mental and physical practice. Of
intersperse the demonstrated information with course, many experiments use other variations of
physical practice trials performed by the observer. these mental practice manipulations (see reviews
Similar to the conclusion from the previous by Feltz & Landers, 1983; Feltz, Landers, & Becker,
discussion, this scheduling of observation and 1988; Lotze & Halsband, 2006).
physical practice has the advantage of engaging
the learner more in the active problem-solving Is Mental Practice as Effective
process of learning than would be the case if all as Physical Practice?
demonstrations were presented prior to physical
practice. This scheduling also has the advantage of A nice demonstration of all these various prac-
providing the performer with some rest between tice conditions and their effects was provided
physical trials, which, as we discuss in a later sec- in a complex study by Hird, Landers, Thomas,
tion, is also important for learning. The literature and Horan (1991). Twelve groups of subjects
is quite consistent in showing that interspersing participated in the experiment. Six groups were
physical and modeled trials results in better learn- asked to learn a pegboard task, inserting pegs of
ing (retention and transfer) than a single block of different colors and shapes as rapidly as possible
observations prior to physical practice of the task into squares cut in a board. The other six groups
(Deakin & Proteau, 2000; Granados & Wulf, 2007; performed the pursuit rotor task. For each task,
Shea, Wulf, & Whitacre, 1999; Shea, Wright, Wulf, subjects performed a pretest, seven sessions of
& Whitacre, 2000; Weeks & Anderson, 2000). In training (on separate days), and a posttest. During
this sense, interleaving information gained from the training sessions the 100% physical practice
off-task observation with on-task practice appears group performed eight trials on the task while
to optimize the benefits gained from observations. the 100% mental practice group covertly prac-
ticed the task for the same amount of time. Three
other groups involved combinations of practice,
Mental Practice consisting of two, four, or six trials of physical
practice combined with six, four, or two trials of
One of the most frequently used and advocated mental practice (i.e., 75% physical practice [P],
off-task methods to promote learning is mental 25% mental practice [M]; 50P:50M; and 25P:75M
practice, in which the performance of a task is groups). The control group performed an unre-
mentally rehearsed, often using imagery tech- lated task (the stabilometer) for the same amount
niques, in the absence of overt physical practice. of time during these training sessions.
360 Motor Control and Learning

The difference in performance between the Hypotheses About Mental Practice

pretest and posttest for each group in Hird and Effects
colleagues' study is presented in figure 11.7. 1
The sets of findings for the two tasks are remark- Why then, is mental practice effective for learning
ably similar. The groups given mental practice a motor skill? Certainly, one of the components
(100%M) were more effective than the no-practice of mental practice involves learning the cognitive
(control) groups, but not nearly as effective as elements in the task; that is, learning what to do
the groups given the same amount of physical (Heuer, 1985). Given the requirement of rehears-
practice (100%P). In addition, the results for the ing mentally, the learner can think about what
combination groups showed that learning was kinds of things could be tried, can predict the
enhanced with higher proportions of the train- consequences of each action to some extent on the
ing trials spent in physical, compared to mental, basis of previous experiences with similar skills,
practice (e.g., compare the 75P:25M groups with and can perhaps rule out inappropriate courses
the 25P:75M groups in figure 11.7). of action. This view suggests that not very much
The findings of Hird and colleagues (1991) motor learning is happening in mental practice,
have been replicated (Allami, Paulignan, Brovelli, the majority being the rapid learning associated
& Boussaoud, 2008) and concur with the reviews with the cognitive elements of the task. Such a
of the mental practice literature conducted by view fits well with data from Minas (1978, 1980),
Feltz and Landers (1983; Feltz et al., 1988; Lotze & who used a serial throwing task in which subjects
Halsband, 2006). The results suggest that whenever had to throw balls of different weights and tex-
possible, physical practice is preferable to mental tures into the proper bins. The main finding was
practice for learning. However, when physically that mental practice contributed to the learning of
practicing a task is not possible, as when an indi- the sequence (the cognitive element) but did not
vidual is away from a clinical rehabilitation set- contribute very much to learning the particular
throwing actions (motor elements).
ting, then mental rehearsal is an effective method
Another view, however, suggests that there
for augmenting learning (Dickstein & Deutsch,
is more to mental practice than the learning of
2007; Mulder 2007).

25 5

~ 20
c:
al
.c:
u
'iii i 15
~ ca
~
c.
g.
t/J
.s ;!!:.. 10

Ia.. 5

100%M 100%M
25P:75M 50P:50M _ _ __ J ~--- 25P:75M
Practice condition

FIGURE 11.7 Effects of various combinations of physical and mental practice on pegboard and pursuit rotor tasks.
Adapted, .by permission, from J.S. Hird et al., 1991, "Physical practice is superior to mental practice in enhancing cognitive and motor task performance" Journal of Sport and
Exercise Psychology 13(3): 286-287. '
 Conditions of Practice Iii 361

the cognitive elements in a task. Some suggest Distribution of Practice

that the motor programs for the movements are
actually being run off during mental practice but
Our discussion of on-task conditions of practice
that the learner simply turns down the "gain" of
begins with one of the variables that instructors
the program so that the muscular contractions are
and therapists have under their control: the sched-
not visible. Research on so-called implicit speech,
uling of periods of work (i.e., time spent in actual
in which subjects are told to imagine speaking a
practice) and rest (i.e., time not practicing the task).
given sentence, shows patterns of electromyo-
This scheduling can be considered within the con-
graphic activity from the vocal musculature
straints of a short time frame, such as the amount
that resemble the patterns evoked during actual
of work and rest during a 45 min therapy session.
speech. One possibility is that very small forces
Or the scheduling may be considered in terms of a
(not sufficient to cause movements) are produced
longer time scale, as when one chooses the length
and the performer receives Golgi tendon organ
feedback about them (chapter 5), as the Golgi and frequency of sessions per week. The ques-
tion of importance concerns whether or not the
tendon organs are extremely sensitive to small
loads. Another possibility is that the "move- frequency and length of rest periods have an effect
on learning the skill being practiced in the work
ments" are sensed via feedforward and corollary
periods. What is the best way to distribute the time
discharge (i.e., "internal feedback"), generated
spent in work versus the time spent resting-or
when the motor programs are run off (chapter
5). Yet another possibility discussed earlier in simply, what is the best practice distribution?
this chapter is that planning a movement (which
Defining "Massed"
should be part of mental practice) is, in itself,
beneficial to learning. and "Distributed" Practice
Most recent hypotheses about the effects of Research on practice-distribution effects has
mental practice have focused on the specific frequently used the terms massed practice and
role of imagery. Many researchers now agree distributed practice. In one sense, "massing" means
that imagined actions share similarities with the to put things together-in this case, running
actual movements being imaged. For example, work periods very close together with either no
performance times are similar for imaged and rest at all or very brief rest intervals between
physically performed trials of the Fitts reciprocal periods of work. By default, distributing practice
tapping tasks with different indexes of difficulty means spacing these periods of work apart with
(Cerritelli, Maruff, Wilson, & Currie, 2000; Decety longer intervals of rest. The labels are not truly
& Jeannerod, 1996; Kohl & Fisicaro, 1995; Stevens, satisfactory, however, because researchers often
2005). Similar effects are also observed in grasp- use these terms to describe the two extremes of
ing tasks (Prak, Paulignan, & Jeannerod, 2001) practice distributions within a particular experi-
and when a mass either is loaded or is imagined ment, and because many experimenters use more
being loaded to a limb (Papaxanthis, Schieppati, than two distribution conditions (e.g., Ammons,
Gentili, & Pozzo, 2002). Studies involving brain 1950; Bourne & Archer, 1956). Thus, these terms
mapping techniques (chapter 2) also point to must be considered relative to the context of other
similar activation regions in the brain when move- conditions within any particular experiment and
ments are produced and imagined Geannerod, relative to the context of other experiments.
2001; Jeannerod & Prak, 1999). Together with The vast majority of the research on distri-
observations that apraxia patients sometimes fail bution-of-practice effects has been conducted
to inhibit imagined movements (without aware- using continuous tasks, for which the work
ness; Schwoebel, Boronat, & Branch Coslett, 2002), period might be 20 or 30 s in duration. The task
the findings suggest that imagery is a process most commonly used for this research was the
by which actions are programmed as in normal pursuit rotor tracking task (figure 2.5). However,
movements but are inhibited from being executed. tasks such as mirror tracing, the Bachman ladder
According to this view, at least some learning can (figure 2.8), and inverted-alphabet printing were
be attributed solely to the motor programming also popular in the early research studies of
process, in the absence of movement execution. this area. We first discuss the effects of practice
362 Motor Control and Learning

distribution using continuous tasks. Only a few at least part of the decrement in performance
studies have been done using discrete tasks; displayed by these groups was due to temporary
however, the findings are quite different from fatigue, not all of the performance depression
those of studies using continuous tasks and are could be attributed to differences in the relatively
presented later. permanent development of skill. So, how much
was due to learning?
Distribution-of-Practice Effects To assess this issue, Bourne and Archer gave all
on Performance of their subjects a 5 min rest period following the
last acquisition trial. After this rest period, sub-
Many experiments were done in the 1940s and
jects performed a common transfer test in which
1950s on practice-distribution effects (for reviews,
all groups were shifted to a massed schedule-all
see Donovan & Radosevich, 1999; Lee & Geno-
trials were performed with 0 s rest between 30 s
vese, 1988). Even though these experiments
work periods. The rationale was that if muscular
involved wide differences in methods (such as
fatigue was responsible for all of the differences
the length of work and rest periods, the number
between groups during the acquisition trials, then
of trials), the results are remarkably similar. Put
the groups should be similar in performance after
simply, the findings converge on a very straight-
the dissipation of the fatigue. This was not the
forward statement about the effect on perfor-
case, as can be seen in the right side of figure 11.8.
mance: Given constant periods of work, short rest
Several items in these transfer data are notewor-
periods degrade performance relative to performance
thy. The most important is that substantial differ-
with longer rest periods.
ences were maintained between the groups after
Findings from a study by Bourne and Archer
the rest period, with initial transfer performance
(1956) are typical of the performance effects seen
being increasingly more skilled for groups that
in experiments on practice distribution. The task
had longer periods of rest between work periods
was pursuit rotor tracking. Five different groups
during the acquisition trials. This finding suggests
of subjects were compared; all groups had work
that the practice distribution had a relatively per-
periods of 30 s. In one group (the 0 s rest group),
manent effect, which is supported by many other
subjects practiced continuously for 21 trials, with
studies in the literature (Lee & Genovese, 1988).
no rest at all. For the other four groups, a period of
Another item worth noting in these data is that
rest was interspersed between each work period.
the differences between the groups on the first
One group had rest periods of 15 s, and the other
transfer trial (trial 22) are smaller than the differ-
three groups had rest periods of 30, 45, or 60 s.
ences between groups on the last acquisition trial
Bourne and Archer found that longer rest
(trial 21). Thus, some of the practice-distribution
periods resulted in more effective performances.
effect seen in the acquisition data was due to the
Looking closely at the left side of figure 11.8, one
temporary, detrimental influence of fatigue. Still,
can see that a systematic separation of the various
the differences due to changes brought about by
distribution-of-practice groups had emerged by
learning remained large on the transfer trials. The
about trial 7 and that these differences became
last item to notice is that massing the transfer trials
larger with further practice. Many other examples
of effects like these have been documented, and also had a depressing effect on performance. How-
reviews by McGeoch and Irion (1952), Bilodeau ever, even after nine transfer trials with no rest (i.e.,
and Bilodeau (1961), Lee and Genovese (1988), on trial 30), the groups that had initially practiced
and Donovan and Radosevich (1999) describe with some rest between trials still out-performed
details of this literature. the group that had practiced with no rest.
We have used the Bourne and Archer (1956)
Distribution-of-Practice Effects study to illustrate the effects of practice dis-
tribution on performance and learning. It is a
on Learning particularly good example of this effect because
For tasks such as the pursuit rotor, practice with more than two distribution groups were used and
little or no rest would likely cause muscular because a transfer design was used to separate
fatigue to develop; this fatigue could be expected the temporary from the permanent effects of the
to depress performance, and that seems to be evi- practice variable. However, several conclusions
dent by the results in figure 11.8. Thus, because drawn from this study require further discussion.
 Conditions of Practice 1111 363

35

30

i.. 25

-
as
c:
0
CD
E
:;:; 20

-..
CD
CJ)
as
c:
CD
I,)

CD
D.. 15

10

3 5 7 9 11 13 15 17 19 21 5 min rest 22 24 26 28 30
Acquisition trials Transfer trials

FIGURE 11.8 Distribution-of-practice effects on a pursuit rotor task in acquisition and transfer. Trials were 30 s in duration,
and separate groups received either 0, 15, 30, 45, or 60 s of rest between practice trials. Transfer trials were performed with
O s rest between trials.
Reprinted from LE. Bourne and E.J. Archer, 1956, "fime continuously on target as a function of distribution of practice," Journal of Experimental Psychology 51: 27.

Length of the Retention Interval about the learning difference from the Bourne
One complicating factor about the Bourne and and Archer study.
Archer experiment is that a 5 min rest period Do the Learning Effects "Wash Out"?
following continuous practice may not have been The Bourne and Archer data show that perfor-
long enough to allow the temporary influence mance differences in transfer began to converge
of muscular fatigue on performance to dissipate by trial 30 (after nine transfer trials). The conver-
(Ammons, 1988; Lintern, 1988). Thus, the trans- gence of effects in transfer has been argued by
fer trials may have been influenced by the same some to cast doubt on the "relative permanency"
temporary effects that influenced acquisition of the learning effect. An important study by
performance (e.g., fatigue). Several studies have Adams and Reynolds (1954) further calls this
addressed this issue using longer rest intervals issue into question. In this study, distributed
following practice (e.g., one day in Adams, 1952; practice was defined as 30 s of work with 30 s rest.
10 weeks in Reynolds & Bilodeau, 1952) and have Massed practice involved the same trial duration
shown that the learning effect remained even but with only 5 s of rest. One group received 40
when these temporary effects had surely dis- trials under distributed conditions. Four more
sipated. These findings support the conclusion groups received initial practice for 5, 10, 15, or 20
364 Motor Control and Learning

trials, respectively, under massed conditions; they subjects performed on consecutive days (Monday
then rested for 10 min, and finally transferred to to Friday) for seven weeks. The distributed group
the distributed-practice condition for the remain- practiced three times per week for 12 weeks.
der of the 40 practice trials. Adams and Reynolds Results at the end of the 34th day of practice and
found that when the various massed-practice on a retention test performed three months later
groups were shifted to distributed practice, they showed both performance and learning benefits
caught up (though not entirely) within a few for the distributed group.
trials to the level of performance of the group Similar findings were reported by Baddeley
that had practiced entirely under distributed- and Longman (1978) for postal workers who were
practice conditions. A small flaw in the design, training to use a keyboard. In this study, separate
however, makes these effects difficult to interpret. groups of postal workers trained for 60 to 80 h
The problem is that the groups that transferred using one of four schedules: Work periods were
to distributed-practice conditions received the conducted either once or twice per day, and the
benefit of a 10 min rest. The distributed group, duration of each work period was 1 or 2 h. The
which may have experienced some temporary data for the practice period and for retention
fatigue effects, did not have such a rest. Thus, it tests performed one, three, and nine months
is difficult to know whether or not the differences later showed that the condition that massed the
observed were temporary or permanent. practice the most (2 X 2) resulted in the lowest
A clever design by Ammons (1950) helps to performance and learning (see figure 11.9).
clarify this issue. Groups received rest periods Although the other three groups did not differ
that ranged from 0 s and 20 s to up to 12 min in these retention tests, the effects of the "most
between each 20 s trial on the pursuit rotor task. distributed" group (1 x 1) are likely diminished
One group even had 24 h between each of the because practice for this group was stopped after
36 practice trials. A 20 min rest period followed a total accumulation of 60 h, as compared to the
the 36th practice trial, after which subjects per- 80 h of practice for the other three groups. These
formed an additional 36 transfer trials with no data appear to suggest that there is some gener-
rest between trials (many more transfer trials alizability of the results obtained in experiments
than had been used by Bourne & Archer [1956]). of relatively short duration to studies involving
By the end of this transfer period, only small dif- practice and retention over much longer periods
ferences remained between the groups. However, of time (see also Shea, Lai, Black, & Park, 2001).
Ammons (1950) asked subjects to return to the lab
Total Practice Time
for another set of transfer trials one day later. The
differences that had been seen on the first transfer From the previous sections, it would appear that
test-and apparently washed out by the transfer it is not beneficial for learning to mass trials in a
trials-were "restored" after this additional rest practice session. But there is another important
period. These data are strong indicators that variable that coexists with massing-the total
practice distribution has large effects on tempo- time involved in practice. Recall that in the experi-
rary performance levels and relatively permanent ments presented so far, the number of practice
influences on learning. trials was held constant; and because the amount
of time between practice trials was different for
Distributing Practice Over the massed and distributed conditions, the overall
a Longer Time Scale practice time was allowed to vary. That is, a group
Perhaps of more direct significance to instruc- receiving massed practice will have a shorter total
tors and therapists are the effects of practice practice period than will an equivalent group
distribution when practice takes place on a much with distributed practice.
longer time scale. A few such studies have been Consider the Baddeley and Longman (1978)
performed, and the results are generally similar study just described. Although the group that
to those of the studies done in a single session. practiced for 2 h per session twice per day (2
In a very early investigation of this type, right- X 2) showed the poorest acquisition and retention
handed subjects were asked to throw javelins performance, their practice period was completed
with their left arm (Murphy, 1916). All subjects in one-half the time used by two groups (1 X 2
practiced on 34 separate days. Massed-practice and 2 X 1) and one-quarter of the time used by
 Conditions of Practice 111 365

90

80

c
.E
Ui 70
..
~
0
Ui
>-
~
u
..
I!:! 60
0
0

50

40 44 48 52 56 60 64 68 72 76 80 3 9
Hours of practice Retention (months)

FIGURE 11.9 Training postal workers to use a keyboard under different practice-distribution schedules (1 x 1 refers to
sessions conducted once per day of 1 h duration each; 2 x 2 refers to two sessions per day of 2 h duration each, etc.).
Data from Baddeley and Longman 1978.

the 1 X 1 group. Additional training for this 2 dangerous, but many tasks used in sport (e.g.,
X 2 group would likely have resulted in improved giant swings on the horizontal bar) and industry
performance and learning. (e.g., work with a hydraulic paper cutter) entail
The issue. of practice distribution and total considerably more opportunity for serious injury
practice time involves a trade-off Distributed if errors are made. And most certainly for people
practice results in the most learning per time in in rehabilitation, whose motor coordination has
training but requires the most total time to com- already been affected, the risk of injury is of vital
plete. Massed practice results in reduced benefits concern to the therapist. Thus, caution should be
per time in training but requires the least total used in designing training regimens in situations
time. Thus, what is most effective for learning may in which factors such as fatigue could put the
not always be the most efficient, at least in terms learner at risk.
of time required to optimize learning.
Safety Issues Discrete Tasks
Finally, it should be clear that massing has strong The amount of evidence about discrete tasks is
effects on performance of many tasks and that far less than for continuous tasks. Carron (1967,
the risks of injury in dangerous tasks are likely 1969) used a peg-turn task in which the subject
to increase with massed practice. The labora- moved 44 cm from a key to grasp a peg in a hole,
tory tasks described here are not particularly turned the peg end-for-end to reinsert it into
366 Motor Control and Learning

the hole, and then returned to the key again as test, so that practice under distributed conditions
quickly as possible. This movement was discrete in acquisition resulted in more learning, regard-
and required a movement time (MT) of from 1,300 less of whether the transfer-test conditions were
to 1,700 ms, depending on the level of skill of the distributed or massed. This was essentially the
performers. Carron had subjects learn this task same as had been found with the other continu-
under two conditions: distributed (the amount ous tasks (see previous section). However, for the
of rest between trials was 5 s) and massed (the discrete task, there was a slight tendency for the
amount of rest between trials was only 300 ms, massed condition to be more effective in acquisi-
with a 5 s rest every 10 trials). Carron found no tion. In a delayed transfer test performed under
effect of the massing conditions on performance massed conditions, there was an advantage to
of the task while the massing was present. When the group that had massed practice in acquisition
he tested the subjects 48 h later as a measure of compared to distributed practice. This provides
learning, he found that the subjects in the massed at least one example indicating that massed
condition actually performed slightly faster than practice can be more effective for learning than
the subjects in the distributed condition (1A30 distributed practice.
vs. 1,510 ms), but it is probably more reasonable A different interpretation of the distribution-
to say there were no real differences. For this of-practice effect can be made when considered
discrete task, massing appeared to be neither a over a longer-term time scale. For example, in a
performance variable nor a learning variable, golf putting study by Dail and Christina (2004),
contrary to the rather strong effects of massing novice subjects performed 240 practice putts of
found for continuous tasks. 3.7 m distance. Subjects in the massed group per-
Lee and Genovese (1989) investigated this formed all of their trials in one session, whereas
apparent continuous-discrete difference directly, the distributed group practiced 60 trials on each of
in parallel experiments employing very similar four consecutive days. All subjects returned either
timing tasks. For the continuous task, there was 1, 7, or 28 days later for retention tests. The findings
a tendency for subjects in the distributed condi- from the Dail and Christina study, presented in
tions in acquisition to perform more effectively figure 11.10, revealed small but consistent advan-
than those in the massed conditions. This effect tages for distributed practice by about halfway
carried over into the delayed (seven days) transfer through the practice trials. More importantly, these

180

160

140

120
~
0
0
t/)
100
Cl
c:
=:l
D..
80

60

40

20

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 R1 R7R28
Trial blocks of 10 putts

FIGURE 11.10 Effects of massed and distributed practice in a golf putting task.
Reprinted with permission from Research Quarterly for Exercise and Sport, Vol. 75, pg. 151, Copyright 2004, by the American Alliance for Health, Physical Education, Recreation
and Dance, 1900 Association Drive, Reston, VA 20191.
 Conditions of Practice !!ii 367

performance advantages were maintained on each complication with this type of study, however, as
of the retention tests, conducted up to one month subjects in the different groups practice different
later (see also Shea, Lai, Black, & Park, 2000, experi- tasks; thus what has been practiced and what is
ment 2). Thus, it would appear that the effects of assessed in retention cannot be equated. But this
distributed practice on performance and learning does not pose a problem for results such as those
in continuous and discrete tasks might have more we see in studies conducted by Shea and Kohl
similarities than previously thought. (1990, 1991).
Subjects in the Shea and Kohl experiments
learned to generate a goal force by squeezing
Variability of Practice a handgrip that was connected to a force trans-
ducer. In one experiment (Shea & Kohl, 1991,
Another factor that has been shown to affect experiment 1), subjects performed 100 trials to
learning is the amount of variability in a practice acquire a criterion force of 150 N. One group
sequence. In one sense, this is obvious. Many ("Criterion only" in figure 11.11) received only
tasks have inherent variability, such as fielding these acquisition trials. Another group ("Cri-
ground balls in baseball or steering a car down terion + variable") received the same number
an unfamiliar road (so-called open skills; see chap- of acquisition trials on the criterion task but, in
ter 2). An important part of learning such tasks addition, practiced goal forces that were ±25
is acquiring the capability to cope with novel or ±50 N relative to the criterion task (i.e., 100,
situations; practicing under constant (unvarying) 125, 175, and 200 N). Notice, however, that this
situations would probably not be appropriate. But variable-practice group not only had the same
in another sense, the effect of variable practice is amount of specific practice as the criterion group,
not so obvious, especially when the task involves but also practiced at tasks that surrounded the
closed skills, for which the environmental condi- criterion task-which confounds the role of the
tions are always similar (e.g., archery, bowling). variable practice with the amount of additional
Here, because the criterion task to be learned is practice. So, Shea and Kohl also included a third
always performed under the same environmental group of subjects ("Criterion + criterion") that
conditions, it would seem that practice under practiced the criterion task, as well as performing
these exact conditions would be most effective additional practice trials on the criterion task, so
for learning. Yet the evidence suggests that varied that the total number of practice trials was equal
practice may be important in closed tasks as well. to the total practiced by the variable group.
Much of the research on variability of practice Performance on the criterion task for these
has been conducted to test certain predictions of groups in acquisition and in a retention test one
schema theory (Schmidt, 1975b; see chapter 13 for day later is presented in figure 11.11. The crite-
more on schema theory). One prediction was that rion + variable practice group performed less
transfer to novel tasks would be enhanced after accurately on the criterion task throughout most
practice in variable, as compared to constant, of the acquisition period in comparison to the
practice conditions. We discuss only a few of these other two groups. However, after a rest period
studies; reviews of many more of these experi- of no practice, subjects in the criterion + variable
ments are available (Lee, Magill, & Weeks, 1985; practice group performed more accurately in a
Shapiro & Schmidt, 1982; Van Rossum, 1990). retention test on the criterion task than both the
criterion-only group and the criterion+ criterion
Variability-of-Practice Effects group. These findings indicate that practice at
in Retention tasks that were similar to (and that surrounded")
/1

the criterion task actually facilitated its retention.

One way to obtain an indication of the effect of
practice variability is to assess retention perfor- Variability-of-Practice Effects
mance after a period of time following the acqui-
sition session (chapter 10). A few experimenters
in Transfer
have done this by comparing the relative impacts In one of the first studies investigating practice
of constant and varied practice on retention of variability, McCracken and Stelmach (1977) had
the tasks that were practiced. There is a design subjects move their right arm from a starting key
368 Motor Control and Learning

g 15

......0
Q)
.....c:
!'CS
.....
I/)
c: 10
0
(.)

s::I
0I/)
.c
<( 5

2 4 6 8 10 12 14 16 18 20
Acquisition trials Retention

FIGURE 11.11 Effects of variable-practice conditions and a criterion-only practice condition during acquisition and retention
(performance plotted only for the criterion task).
Reprinted with permission from Research Quarterly for Exercise and Sport, Vol. 62, pg.190, Copyright 1991, by the American Alliance for Health, Physical Education, Recreation
and Dance, 1900 Association Drive, Reston, VA 20191.

to knock over a barrier, with a 200 ms goal MT In a transfer-test phase, the two groups per-
from initiation to barrier contact. The distances to formed a novel (50 cm) distance, both immedi-
the barrier could be changed in different condi- ately after training and after a two-day interval.
tions (15, 35, 60, and 65 cm), with a constant 200 With this design the authors evaluated the
ms goal in the practice phase. Table 11.1 presents effect of variable versus constant practice on
the details of the experimental design. A Constant the performance of a variation that had never
group was actually made up of four subgroups, been performed previously. This transfer design
each of which had practice at only one of the bar- addresses an effect of learning quite different
rier distances for 300 trials. The Variable group, from that studied in the retention design by
on the other hand, had the same number of trials Shea and Kohl (1991) discussed in the preced-
as the Constant group (i.e., 300), but these trials ing section. In that research, Shea and Kohl
were varied in that all four barrier distances were assessed how well one common task, practiced
practiced in a random order (75 trials of each). 2 by all the groups, was retained as a function of

Subgroup a 15 cm only 50 cm
Subgroup b 35 cm only 50cm
Subgroup c only cm
Subgroup d only cm
Variable 60, 65 cm cm
Adapted from McCracken and Stelmach 1977.
 Conditions of Practice !Ill 369

the other tasks that had also been practiced. In the transfer task. Would such effects also be seen
the McCracken and Stelmach study, the primary if the novel transfer task was outside the range of
research interest was the effect of varied versus previous experience?
constant practice on the capability to perform a Catalano and Kleiner (1984) used a timing
novel task. task in which the subject was to press a button
The results are shown in figure 11.12, where when a moving pattern of lights arrived at a
the absolute errors are plotted for the trials at coincidence point. Using a design much like that
the end of the acquisition phase, as well as for of McCracken and Stelmach (1977), they had a
the trials on the two transfer-test phases. In the Variable group practice at speeds of 5, 7, 9, and
original-practice phase, the Constant group had 11 mph; a Constant group (with four subgroups)
less absolute error than the Variable group. This practiced at only one of these speeds. Then, on a
finding is similar to the results shown by Shea subsequent transfer test, all subjects transferred
and Kohl (see figure 11.11). The critical contrasts, to four novel light speeds that were outside the
however, are in the transfer-test phases, when the range of previous experience (i.e., 1, 3, 13, and
task was novel for both groups. In immediate 15 mph). The performance of the two groups on
transfer, the order of the groups was reversed, these transfer tests is shown in figure 11.13. The
with the Variable group now having less absolute absolute errors were smaller for the Variable
error than the Constant group. This trend per- group than for the Constant group, and the dif-
sisted into delayed transfer, 48 h later, but with ferences were present even when the "distance"
the difference between groups being consider- from the range of previous experience was quite
ably smaller. Thus, it appeared that variability large. Variable practice appeared to increase the
in practice (during the original-practice phase) 11
applicability" of the learning that occurred in
allowed the subjects to learn the task more effec- acquisition, contributing to the performance
tively, permitting them to perform a new version of novel variations of the task that were well
of it on the transfer phase with less error than the outside the range of the stimuli experienced in
Constant group (see also Wrisberg & Ragsdale, the acquisition phase. One cannot argue that
1979). Variable practice seemed to be important the Variable group was more effective simply
in generating a capacity to perform a novel ver- because the subjects had experienced the range
sion of this task. But this novel task (50 cm) was of speeds involved in the transfer test, as neither
clearly within the original range of experience of group had experienced them. In other words,
the Variable group (i.e., from 15 to 65 cm), and variable practice seemed to increase generaliz-
one could argue that the Variable group had more ability, an important criterion for motor learning
practice than the Constant group at tasks closer to as discussed in chapter 10 (see Roller, Cohen,
Kimball, & Bloomberg, 2001, for a different dem-
onstration of generalization following variable
practice).

Other Factors Influencing Effects

of Practice Variability
When adults are used as subjects, there is rea-
sonably strong evidence that increased practice
variability is beneficial for learning (as measured
on novel transfer tests), and basically no evidence
that variable practice is detrimental to learning
(Shapiro & Schmidt, 1982). However, a number
of studies show very small effects, and others
Last 30 Immediate Delayed
show essentially no effects, casting some doubt on
acquisition trials transfer transfer the "strength" or generality of these effects (Van
Rossum, 1990). Overall, in practical settings it is
FIGURE 11.12 Performance in a ballistic-timing task as a reasonably safe to say that attempts to make the
function of variability in practice conditions. practice more variable for learners will result in
370 Motor Control and Learning

80
ti)
.§.
......0 70
Q)

....::I
Q)

Ci 60
t/l
.c
ctl
c:
ctl
Q)
:ii: 50

40

3 13 15
Transfer-test velocity (mph)

FIGURE 11.13 Mean absolute timing error during novel transfer-test velocities as a function of the variability in practice con-
ditions in acquisition.
Reproduced and adapted from figure 2 with permission from authors and publisher from: Catalano. J.F., & Kleiner, B.M. Distant transfer in coincident timing as a function of
variability of practice. Perceptual and Motor Skills, 1984, 58, 851-856. ©Perceptual and Motor Skills 1984.

greater learning and generalizability. However, of variability of practice may be accounted for
in the following paragraphs we discuss some of by the nature of the task. For example, a study
the issues that complicate this general statement. by Shea, Lai, Wright, Immink, and Black (2001)
showed that the relative timing of a closed task was
Age of the Learner
learned more effectively with practice conditions
The effects of practice variability seem to depend that promoted movement consistency (e.g., con-
on the nature of the learners. Certainly the most stant practice) compared to practice conditions
obvious classification is that of children versus that promoted variability. In contrast, learning
adults. In their review of the literature on practice the absolute timing of the task was better under
variability, Shapiro and Schmidt (1982) noted variable- than consistent-practice conditions.
that the advantage for variable versus constant These recent findings support the intriguing
practice for children was strong in nearly every possibility that some aspects of the same task may
study conducted. For example, using a strictly be learned more effectively through constant-
closed throwing skill with young children, Kerr practice conditions and other aspects through
and Booth (1977, 1978) found in two experiments variable-practice conditions (see Wulf & Shea,
that variable practice was more effective than 2002, for more discussion).
constant practice when subjects were transferred
to a novel version of the task. Even more surpris- Scheduling Variable Practice
ing was the finding that for learning this novel We mentioned earlier that the effects of variable
variation of the task, practice at variations of practice in adults have not always been consis-
the task approximating the novel task was more tent-some studies showing positive effects and
effective than was practicing the novel task itself! others showing no effects. A review of those
Practice variability appears to be a powerful vari- studies showing no effects by Lee and colleagues
able in children's motor learning (see also Green, (1985) revealed an interesting pattern of find-
Whitehead, & Sugden, 1995; Wulf, 1991). ings. Many of these experiments had structured
the variable-practice sessions such that most or
Nature of the Task all of the practice on any single variant of the
Similar to the practice-distribution effects, it task was conducted together, in what is called a
appears that some of the differences in the effects blocked practice sequence. Although we will have
 Conditions of Practice Iii! 371

much more to say about the effects of random and Why should variable practice be more effective
blocked practice in the next section, the conclu- for children? One idea is that children are less
sion drawn by Lee and colleagues (1985) was that experienced at motor skills than are older (adult)
for variable practice to be most effectively utilized subjects, so the rules (schemas) that the children
(relative to constant practice), the order should acquire in laboratory settings have already been
be randomized, rather than blocked. achieved by the adults in their earlier experiences
with motor tasks. Also, when the task used in the
Interpreting Variability-of-Practice research is simple, it is possible that the adults
Effects already have at their disposal the rules (sche-
mas) necessary to perform novel tasks whereas
Most of the studies on variability have been done the children must learn some of the rules in the
in the context of schema theory (chapter 13). The experimental setting. Here, then, variable practice
basic premise is that with practice, people develop is more effective for children than for the adults
rules (called schemas) about their own motor because the children have considerably "more to
behavior. Think back to the ideas about the gener- learn" than the adults.
alized motor program (chapter 6), indicating that a
set of parameters must be applied to the program
in order for it to be performed. Schema theory Contextual Interference
proposes that subjects learn a rule in the practice
sequence. The rule is a relationship between all The preceding section focused on practicing a
the past environmental outcomes that the person number of task variations, compared to practice
produced and the values of the parameters that on only one task variation, as measured in reten-
were used to produce those outcomes. This rule is tion and transfer tests (i.e., tests of learning). In
maintained in memory and can be used to select a this section we assume that variable practice is
new set of parameters for the next movement situ- more useful for learning, and ask the question
ation-even a novel variation-that involves the whether or not it makes a difference how the
same motor program. Knowing the rule and what variable practice is scheduled.
environmental outcome is to be produced, the
person can select the parameters for the program Blocked Versus Random Practice
that will produce it. The schema theory is related to Research on the scheduling of practice for mul-
variability in practice because the theory predicts tiple tasks was popularized in the motor skills
that learning the rule will be more effective if the area with the study by Shea and Morgan (1979),
experience is varied rather than constant. although a few isolated studies of practice-
Another important finding from the literature scheduling effects had been published earlier
on variability in practice is that the occurrence (e.g., Pyle, 1919). The Shea and Morgan study
of learning during the acquisition phase was was influenced considerably by the ideas of
revealed by performance on a novel version of William Battig (see "William Battig and Contex-
the task in transfer. This was true regardless of tual Interference"), and together, their work has
whether the novel version was inside (McCracken made a substantial impact on research in motor
& Stelmach, 1977) or outside (Catalano & Kleiner, learning and practical application to real-world
1984) the range of variation experienced in the skills. One of the main issues was the finding that
acquisition phase. As we will point out in chapter variables that made the organization of practice
13, such evidence suggests that what was learned more difficult" during acquisition degraded
/1

was not some particular movement, but rather performance during acquisition (not surpris-
the (generalizable) capability to produce any of a ingly), but also made performance in retention
variety of movements of this type. These results and transfer tests more effective. This latter
are explained well by schema theory, in that the effect was surprising and attracted considerable
variable practice produces a rule (or schema) for attention from researchers. Battig's notion was
selecting parameters of the generalized motor that contextual interference-which he defined
program (e.g., for throwing), and this rule can as interference generated due to the context in
be used for any novel movement using the same which the skills were being practiced-produced
motor program. decrements in performance during practice, but
372 Motor Control and Learning

made the learning of these tasks more effective. notion was that, by randomizing the order in
This advantage for learning showed up in later which several tasks were practiced, high levels
tests of retention and transfer. Shea and Morgan of contextual interference would be generated,
(1979) operationalized Battig's notion of contex- and the learning advantage of this interference
tual interference in terms of the order of practice would then be seen later in retention and trans-
among several tasks during acquisition. Their fer tests.

William Battig and Contextual Interference

Research on practice schedules owes a huge debt of gratitude to William F. Battig. Throughout a
distinguished career, this cognitive psychologist maintained an interest in memory and learning,
conducting studies using both. verbal and motor tasks. Early in his research, Battig found that fac-
tors that make a task more "difficult" for the subject to perform actually enhanced remembering and
transfer. For example, requiring (vs. not requiring) learners to pronounce nonsense "words" (e.g.,
XENF), whose letters corresponded to individual finger movements, made performance on another
version of the finger task more effective (Battig, 1956).
Battig interpreted these and related findings in terms of the principle that "intertask facilitation
is produced by intratask interference" (Battig, 1966, p. 227). lntratask interference referred to the
performance decrement caused by attempting to keep multiple items in working memory at one
time (e.g., the interference between the "word" pronunciations and the finger movements). By inter-
task transfer, Battig was referring to the beneficial transfer of learning to other motor tasks. These
findings ran counter to intuition, as many researchers believed that transfer to other tasks would
be strongest if the first task had been learned under the most optimal conditions for performance
during practice .
. But the field of psychology was not prepared to consider such radical ideas, perhaps because
the concepts ran so counter to existing theories of memory and learning. Little attention was paid to
Battig's ideas, even though he continued to publish more demonstrations of these counterintuitive
findings (e.g., Battig, 1972; Hiew, 1977). A responsive chord was finally struck with the publication
of Battig's expanded ideas on contextual interference. In this paper (Battig, 1979), he presented a
wider framework that conceptualized the findings he had accumulated over the years. These ideas
were expanded shortly thereafter to motor skill learning (Battig & Shea, 1980), where Battig's influ-
ence has made a very important mark.
Battig identified two important sources of interference that could arise during practice. One factor
related to the order in which multiple items were studied or practiced. If the same task was practiced
repeatedly, then only this one task needed to be held in working memory, and interference should
be low. However, if practice involved frequent switching among multiple tasks, then interference
should be hi9her. This source of interference has been the object of considerable study and is the
focus of the. present discussion on contextual interference. The other source of interference was
the nature of the material to be practiced. If the items (or motor tasks) were quite similar, then the
interference arising during practice would be high because of the increased confusion among them.
Items or tasks that were quite different or distinct would cause lower interference. Battig showed
thatthe presence of these factors both (a) degraded performance in acquisition and (b} facilitated
performance in retention, transfer, or both.
Above all, the most important element of the contextual interference arising from a set of tasks or
items to practice was how the learner responded to the interference. Battig suggested that subjects
respond to situations of high or low interference with correspondingly high or low levels of elaborative
and distinctive processing (which are effortful). These ideas have been expanded for motor behavior
and motor learning by John Shea and his colleagues; today, they represent an important account
of the contextual-interference effect.
 Conditions of Practice 111 373

In their study, Shea and Morgan (1979) used variable. The findings are very clear. Blocked
two kinds of practice sequences in acquisition, practice resulted in an immediate performance
and learners practiced three different tasks. advantage during acquisition compared to
Blocked practice is a sequence in which all the trials random practice (blocks 1-6, during acquisition,
on one task are done together, uninterrupted by in figure 11.14). And, though the performance
practice on any of the other tasks. On the surface, difference was reduced over blocks of trials, there
a blocked practice sequence seems to make good remained an advantage favoring the blocked
"common sense," in that learners can concentrate group at the end of practice.
on improving one task before moving on to the The performances in the retention tests are
next task. For random practice, on the other hand, presented on the right side of figure 11.14.
the same task is rarely repeated on consecutive Consider first the tests given under random
trials. Notice that, in both practice sequences, the conditions, seen as the solid lines in the figure;
same number of trials is performed on each task, the first letter in the label gives the acquisition
the only difference being the order in which the condition (R or B), and the second letter gives
various task variations are presented. The task the retention condition (R or B). The group that
variations involved rapid arm movements to pro- received random practice in acquisition (R-R) far
duce three different movement patterns, and were outperformed the group with blocked conditions
evaluated by measures of reaction time (RT) to in acquisition (B-R) in both the 10 min and 10-day
initiate and the movement time (MT) to produce random retention tests. That is, the performance
the pattern; total time (RT + MT) was also used. advantage seen during acquisition was reversed
There were 54 trials in acquisition, 18 on each of when learning was assessed in these retention
the tasks. Retention and transfer tests were given tests. Now examine the data for subjects in the
after 10 min and 10 days, administered under blocked retention test (dotted lines). Here, the
both random and blocked conditions for different group with random conditions in acquisition
subgroups of subjects. (R-B) outperformed the group with blocked
The Shea and Morgan results are shown in conditions in acquisition (B-B), but not by as
figure 11.14, here with total time as the dependent large an amount as was seen for the R-R and B-R

2.6

2.4

2.2

2.0
~
I- 1.8
==+
I- 1.6
a:
1.4

1.2

Blocks of nine trials

Acquisition

FIGURE 11.14 Blocked and random practice effects during practice and in various conditions of immediate and delayed
retention.
Adapted, by permission, from J.B. Shea and R.L. Morgan, 1979, 'Contextual interference effects on the acquisition, retention, and transfer of a motor skill:' Journal of
Experimental Psychology: Human Leaming and Memory 5: 183. Copyright© 1979 by the American Psychological Association.
374 m Motor Control and Learning

groups. In summary, regardless of the schedul- 2004; Overduin, Richardson, Lane, Bizzi, & Press,
ing in the retention trials, learning was always 2006), and error-detection capabilities (Sherwood,
enhanced when practiced under random condi- 1996), to list just a few.
tions. Random practice degraded performance Many studies of CI effects have used rela-
in acquisition but facilitated learning. Similar tively simple laboratory tasks, leading some to
findings were seen in transfer tests to novel pat- suggest that the potential value of this research
terns. These findings strongly supported Battig's for practical situations would be greater if the
ideas that contextual interference during practice findings were replicated in more complex tasks
would facilitate retention and transfer-that is, (e.g., Wulf & Shea, 2002). For example, Tsutsui,
would facilitate learning. Lee, and Hodges (1998) showed that random
These effects of blocked versus random prac- practice facilitated the learning of new bimanual
tice on learning have been termed contextual- timing patterns-tasks that are difficult to acquire
interference effects (or CI effects), based on Bat- because of the inherent stability of competing in-
tig's original notions. However, notice that the phase and anti-phase patterns (see chapter 8). And
CI manipulation is just one of many ways in Ollis, Button, and Fairweather (2005) found that
which contextual interference could be gener- random practice facilitated the learning of knot-
ated or manipulated (or both) during practice tying skills, regardless of the complexity of the
(for reviews see Brady, 1998, 2004; Lee & Simon, knots. In contrast, Albaret and Thon (1998) found
2004; Magill & Hall, 1990). that random practice facilitated learning only for
a version of a drawing task that had the smallest
Generalizability of number of component parts. No CI differences
Contextual-Interference Effects were found for the complex version of the task.
Despite some uncertainty about the role of
Issues about the generalizability of CI effects
task complexity, the typical CI effect has been
might be rephrased to ask the question, "How
replicated in a number of sport tasks (which are,
much faith should I put in the implications aris-
ing from the Shea and Morgan study?" Should arguably, more complex than laboratory tasks),
random and blocked differences be expected to such as badminton (Goode & Magill, 1986; Mem-
emerge under a variety of different conditions, mert, Hagemann, Althoetmar, Geppert, & Seiler,
using different tasks, for different subjects, and so 2009; Wrisberg, 1991; Wrisberg & Liu, 1991), rifle
on? Research suggests that, while there is overall shooting (Boyce & Del Rey, 1990), volleyball skills
rather wide generalizability, certain conditions (Bartoli, Robazza, Durigon, & Carra, 1992), kayak-
tend to influence the size of the CI effect. ing (Smith & Davies, 1995), snowboarding (Smith,
2002), and baseball batting (Hall, Domingues, &
Task Influences Cavazos, 1994). Moreover, it is important to note
The original Shea and Morgan (1979) experiment that CI effects have also been found in many other
used a laboratory task in which subjects were tasks as well, such as learning logic operations
required to make rapid patterns of arm move- (Carlson & Yaure, 1990), automatic teller opera-
ments by knocking over small wooden barriers tions (Jamieson & Rogers, 2000), foreign language
in response to a stimulus light. These findings vocabulary (Schneider, Healy, & Bourne, 2002),
have been replicated a number of times using artwork identification (Kornell & Bjork, 2008), and
similar task requirements (e.g., Del Rey, Liu, handwriting in children (Ste-Marie, Clark, Findlay,
& Simpson, 1994; Lee & Magill, 1983b; Shea & & Latimer, 2004).
Wright, 1991; Wright, 1991). Other laboratory In contrast to these replications of CI effects in
studies have revealed similar results using tasks "complex tasks" are a number of studies in which
that emphasize the timing of actions (e.g., Lee & no differences were found, including investiga-
Magill, 1983b; Proteau, Blandin, Alain, & Dorion, tions using tasks such as dart throwing (Meira &
1994; Wulf & Lee, 1993), perceptual anticipa- Tani, 2001; Moreno et al., 2003), volleyball skills
tion (e.g., Del Rey, 1989; Del Rey, Wughalter, & (Jones & French, 2007), and gymnastics cart-
Whitehurst, 1982), force regulation (Enebo & wheels (Smith, Gregory, & Davies, 2003). How-
Sherwood, 2005; Shea, Kohl, & Indermill, 1990; ever, in most of these studies it was only rarely
Shea, Shebilske, Kohl, & Gudadgnoli, 1991), force the case that blocked practice resulted in more
field adaptation (Osu, Hirai, Yoshioka, & Kawato, learning than random practice (e.g., Shewokis &
 Conditions of Practice 1111 375

Klopfer, 2000; see also Wulf & Shea, 2002, and a per week for six weeks. The batting sessions
related discussion later in this section). In general, involved practice in which the pitcher threw 15
random practice has almost always been found fastballs, 15 curves, and 15 change-ups. Groups of
to be as beneficial as, if not superior to, blocked batters received these pitches in either a blocked
practice for learning. or a random order over the entire six-week
period. They also performed two transfer tests
Subject Influences in which pitches were delivered in both random
In the preceding section we presented evidence and blocked orders. The results are presented in
that variability-of-practice differences were larger figure 11.15, including the results for a control
in children than in adults. The evidence relating group that did not receive the extra batting prac-
· to CI effects is not quite as clear, however, as tice. The control group performed more poorly
some studies have shown these effects in children on the transfer tests than did either practice
(e.g., Pollock & Lee, 1997; Ste-Marie et al., 2004; group, suggesting that the extra batting practice
Wulf, 1991) whereas others have not (e.g., Brady, was beneficial regardless of the order in which
2004; Del Rey, Whitehurst, & Wood, 1983; Jarus & .the pitches were thrown. However, the most
Goverover, 1999; Pigott & Shapiro, 1984). There interesting finding was the observed CI effect in
is some evidence that the magnitude of CI effects practice and transfer. Notice in figure 11.15 that
may also depend on experience. Del Rey and her the performances of the blocked group on the
colleagues have shown, for example, that transfer blocked transfer test and of the random group
in an anticipation task after random practice is on the random transfer test were nearly identi-
facilitated more for subjects with experience in cal to their respective performances in the eighth
open skills than for novices (Del Rey, 1989; Del practice session. It was when performance was
Rey et al., 1982). assessed on the common transfer tests that the true
A particularly interesting finding was reported value of the practice sequences in learning came
by Hall and colleagues (1994), whose subjects through, as random practice facilitated transfer
were college-level baseball players and thus under both orderings of pitches. Thus, it would
already quite skilled at the task. All subjects appear from this study that even highly skilled
performed two extra batting practice sessions athletes benefited from random practice.

27

24

J!l 21
:E
:2
0Ill
-...
0
Q)
.c
18

E 15
::I
z

12

Practice session Transfer order

FIGURE 11.15 Blocked and random practice effects in baseball batting.

Reproduced and adapted from Figure 1 with permission of authors and publisher from: Hall, K.G., Domingues, D.A., & Cavazos, R. Contextual interference effects with skilled
baseball players. Perceptual and Motor Skills, 1994, 78, 835-841. ©Perceptual and Motor Skills 1994.
376 Motor Control and Learning

Other Practice Schedules Morgan (1979), under either random or blocked

There are a wide variety of ways in which practice sequences. Results for both acquisition and reten-
could be scheduled, given X tasks and N trials tion replicated the Shea and Morgan findings.
per task. Having all of the trials for one task But Al-Ameer and Toole also added two groups
performed in a drill-type sequence represents an that performed small, randomized blocks of trials, in
extreme scheduling manipulation. Alternatively, which a subject would practice one task for two or
never performing two consecutive trials on the three trials, then randomly switch to another task
same task might be considered the opposite and practice that for two or three trials. This mod-
extreme. A serial practice order might be con- erate CI condition, involving randomized blocks
sidered "moderate," relative to these extremes, of either two or three trials, facilitated acquisi-
because it combines the predictability of blocked tion performance (relative to random practice),
practice with the nonrepetitiveness of random and was just as beneficial to learning as random
practice (Lee & Magill, 1983b). In one study, how- practice. Landin and Hebert (1997) also found
ever, as shown in figure 11.16, the performance of that serial alternation of blocks of trials was more
the serial group was nearly identical to that of the effective than either blocked or random practice
random group, leading to the suggestion that the in the learning of basketball shooting skills.
lack of repetitiveness of random practice-and These findings are important, as they suggest
not its unpredictability-may be the key factor that it may be possible to reduce the acquisi-
that both degrades acquisition performance and tion performance decrement normally seen
facilitates learning. with random practice without sacrificing the
An important question is whether or not a long-term learning benefit as a consequence.
practice schedule that represents some "middle The results are also important for applying the
ground" in terms of repetitiveness of practice findings from these scheduling experiments to
might be beneficial to both performance and tasks involving daily activities. One obvious
learning. Studies by Pigott and Shapiro (1984) drawback with a completely random schedule is
and Al-Ameer and Toole (1993) support this pos- that constantly switching from one task to another
sibility. In the Al-Ameer and Toole study, subjects may be impractical. For example, consider tasks
practiced a task similar to that used by Shea and that involve training a new worker on specific

1.8

1.6

1.4
~

~
~ 1.2
+
l-
a:
1.0

0.8

Acquisition trial blocks Retention

FIGURE 11.16 Comparison of blocked and random practice with a serial practice schedule.
Adapted, by permission, from T.D. Lee and RA Magill, 1983, "The locus of contextual interference in motor-skill acquisition;· Journal of Experimental Psychology: Learning,
Memory, and Cognition, 9: 739. Copyright© 1983 by the American Psychological Association.
 Conditions of Practice ill 377

job-related skills. If these tasks are performed in which the opportunity for contrasting the differ-
separate locations in a plant, it is logical to do at ent tasks is minimized because of the repetitive
least some blocked practice before switching to nature of the schedule. The advantages shown
a new task. The findings previously discussed by random schedules in retention and transfer
regarding a randomized-blocks schedule suggest result from more meaningful representations of a
that this condition combines the positive effect given movement task and more elaborate distinc-
in acquisition performance of a blocked practice tions between the various task versions (see also
with the beneficial learning effect of a random Kornell & Bjork, 2008).
practice schedule. Verbal reports from subjects involved in
these experiments provide one line of evidence
Hypotheses in support of the elaboration hypothesis. Post-
on Contextual-Interference Effects experiment interviews indicated that subjects in
Before the first CI studies were published, learn- the random condition understood the tasks in a
ing researchers seemed to be quite satisfied with qualitatively different way than did subjects who
the general understanding that any practice performed blocked practice (Shea & Zimny, 1983).
variable that promoted effective performance Compared to those who had blocked practice,
in acquisition would also promote learning. subjects in the random group reported a much
The findings of Shea and Morgan (1979) caused larger number of elaborate mental representa-
many researchers to become much less comfort- tions for distinguishing the shapes of the various
able with this general idea. How could a variable movement patterns (e.g., noting that one pattern
that slows improvement and retards the overall was essentially a mirror image of another, or that
level of performance in practice be so potent in a given pattern was the only one with a reversal
facilitating retention? This performance-learning in direction; Shea & Zimny, 1983). In contrast,
paradox generated new thinking and debate, not subjects in the blocked group reported that they
only about why these scheduling effects occur, tended to run the movements off without much
but also with regard to the motor learning pro- thought, more or less "automatically." Using
cess in general. Several hypotheses have been a concurrent verbal report protocol, Zimny
advanced to explain the advantages of random (reported in Shea & Zimny, 1988) found that
over blocked practice for learning. Though they subjects who were engaged in random practice
may seem to present competing views, they made comments about specific tasks, as well as
probably have much in common to say about the between-task comparisons, about twice as often
learning process. as those engaged in blocked practice (see also
Del Rey & Shewokis, 1993). These verbal report
Elaborative Processing Hypothesis data support the enhanced contrastive value of
One of these hypotheses, proposed by Shea and random practice as predicted by the elaborative
colleagues, holds that random practice forces the processing hypothesis.
learner into more elaborative conceptual process- A different type of support for the elaborative
ing of the tasks to be learned (Shea & Morgan, view was provided in a study in which physical
1979; Shea & Titzer, 1993; Shea & Zimny, 1983, practice trials were interspersed with three imag-
1988). During a random schedule, practice on ery practice trials (Gabriele, Hall, & Lee, 1989,
one task is usually followed by practice on a experiment 1). For two groups, these imagery
completely different task. Thus, the preparation trials were performed on the task that had just
for action before movement and the evaluation been practiced physically (blocked imagery). In
of performance afterward may be quite different the other two groups, subjects imaged the three
from the preparatory and evaluative process- tasks, but these had not been performed on the
ing that was completed on the previous trial. preceding physical practice trial (random imag-
According to the elaborative processing view, ery). Regardless of whether the physical trials
random practice promotes more comparative and were practiced in a blocked or random order,
contrastive analyses of the actions required to random imagery facilitated retention more than
complete these tasks. As a result, the represen- blocked imagery, supporting the view that this
tation of each task following random practice contrastive processing during practice was ben-
is more memorable than in blocked practice, in eficial for learning.
378 Motor Control and Learning

A more direct experimental manipulation of program (GMP) must first be retrieved from
elaboration was examined in studies by Wright long-term memory; it must be parameterized and
(1991; Wright, Li, & Whitacre, 1992). Using an otherwise readied for execution in the response-
arm movement task similar to that employed by programming stage of information processing.
Shea and Morgan (1979), four groups of subjects Presumably only one action plan can be in work-
engaged in blocked practice. One group per- ing memory at a time.
formed no additional processing, while subjects In blocked practice, "action plan A" is con-
in the other groups performed selected cognitive structed for task A and remains available in work-
activities between practice trials. After each prac- ing memory for future trials on task A. However,
tice trial, subjects in two of the groups were asked since tasks are ordered intermittently in random
to describe verbally the order of movements of practice, "action plan A" is no longer useful for
one of the tasks-either the task just completed the next task (say task B). Therefore, for the next
or one of the other tasks. In the fourth group, trial, "action plan A" must be abandoned and
subjects were asked to make specific comparisons "action plan B" must be prepared. When the
between the task just performed and one of the random ordering determines that task A is to
other tasks. The prediction was that processing in be performed again, the previously constructed
this last condition would be most like the process- "action plan A," which was abandoned from
ing used by subjects in random practice. The pre- working memory, must be reconstructed. Accord-
diction was supported, as subjects in the intertask ing to the hypothesis, the value of a practice
processing condition were elevated in retention trial depends on the reconstructive processing
as compared to the other three groups. Interest- that was undertaken. Remembering the "solu-
ingly, the additional processing in the other two tion" (the action plan) from a previous trial (as
groups with intervening cognitive descriptions in blocked practice) promotes performance in
did not improve retention at all, suggesting that acquisition, but does not promote the kind of
the qualitative nature of the processing was more processing that facilitates learning as measured in
important than the quantity. retention and transfer. In contrast, random prac-
A more recent technique to study the elabora- tice causes a short-term "forgetting" of the action
tion hypothesis was introduced by Lin, Fisher, plan when a different task must be produced.
Winstein, Wu, and Gordon (2008; see also Lin, This is detrimental to acquisition performance
Winstein, Fisher, & Wu, 2010). They provided but beneficial to retention and transfer because
transcranial magnetic stimulation (TMS; see it forces the subject to undertake reconstructive
chapter 2) between blocked and random practice processing. Notice that this view of the CI effect
trials that was meant to prevent subjects from is based on the same logic as a key explanation of
conducting elaborative processing. The typical the spacing effect in memory (see "When Forget-
CI effect was found for the control, blocked, and ting Improves Remembering").
random practice groups. However, the random One prediction of the reconstruction view that
practice advantage was eliminated when TMS has been empirically tested in a learning para-
was applied between random practice trials. Lin digm relates to the cause of short-term forgetting.
and colleagues provided a strong rationale to sup- The idea here is that on a given task, any activity
port the argument that TMS prevented subjects in between practice trials that causes short-term
the random group from undertaking elaborative forgetting should promote learning. Note that
processing, thereby eliminating the CI effect. this prediction is different from the elaborative
processing view, which suggests that distinctive-
Forgetting and Reconstruction ness increases as the similarity of different tasks
Hypothesis in working memory increases. Several studies
A different explanation for the CI effect was pro- have examined this prediction; the evidence,
posed by Lee and Magill (1983b, 1985). According though not strong, has been generally positive in
to the reconstruction view, the action planning that support of the reconstruction hypothesis (Lee &
occurs just prior to a practice trial is influenced Magill, 1983a, 1987; Magill, 1988; Young, Cohen,
by what has been done in the previous trial. In & Husak, 1993).
chapter 6, we presented the notion that in order A different approach to examining the recon-
to produce an action, the generalized motor struction hypothesis was used in a study by Lee,
 Conditions of Practice mll 379

when Forgetting 1mfj~oves~R~tff ff.

One of the puzzles aboutworkin~ memorfisthe cl.lr!ous stfite~~llfl~~
(Cuddy & Jacoby, 1982).This statelljent $Ountls~t.~angebecal.l~t~ ·
chapter 14) is usually thought of asa~ductiohi.ritn~tapability;tdiem . . . •. ·. ·.·
helps remembering" sounds ".ke riopsense. Ho~ever. the eivi.\:leb~e ·.. ·~~p·
effect in verbal memory suggestslh(lt~Qis stilt~tr)et)t is rotas bi~a_rre•as·lf;$e~w·s1:1
ments, subjects are typically givenatong.li~~of2W~rds that tre\f are~sK~dt~ · ·· ·
time later on a memory .test The list often·c6rl;l~ti.~es wordspreserite€f ~r)lyf
presented more than once. For the'wqrds preis~~.t~!:l more tha6 011~e. tf:l~y:a
immediately (a zero "lag" condition) or atotf"ler.tif(~~· Witry as'~al! o~t~r~4:
ing between repetitions. The spacinfl effect/~f~~s~othe fihdjngfhat.lat~(\f
been repeated with long lags ismore effectlvet~~~recall'ofword~repec:tle~
Larry Jacoby (Cuddy & Jacoby,1982;Jaco~~·it9:78;Ja.c~by&pa119ls,•
that forgetting helps memory because the:pro~e8$ing undert1,11<en dltrl~ ·....· . }s de
what is remembered about the .material from the last processing of it iH.n!il inforr)'.iati~
bered well, then the material to be studied willnotbe fully processed On its s'S'Cdnd;'Pl.'. . ·. ··.···· < . . ,.
If the information has been forgotten,·• then the material will be more fully .pr<:>ce~ae:a,ofj9~·~~a~~.,~ .. :
The critical issue for Jacoby. is thatthe value of a repetition lies in thejlegree te \'\l~r: ·
full processing of the information on each presentation. Processing inform?tiotlis si ·. ·
a problem. The results of the processing co.nstitute a solution, rtJU(;:h Uke the s ...
multiplying two numbers together "in your head.'' If the same problem arises soori aft:
has been determined, then the mental arithrneti~ need not be undertaken ageli!l iti,9r0·er
the problem because the solution is readily ava.ilable in working merrmry. How~yet1 .;i~.~~
has been forgotten, then full processing.rnuf?t be undertakeninor9ert~solvE)t!J~pt~e
Memory, according to Jacoby, is aproduct of the processi119 activities. He ~uhlrl;larjzea:t!Jl~f(~~
when he wrote, ''The. means by Which a .solution is obtained· influenCE)S· $t.1bseqUf:{nt, ret~. •. •' . . .,. •. ...../~;. ;i':): •
formance: subsequent retention suffers when the solution is rernemberE):a... (Jaqoby, 19:78~ . . .:': . ,,, .
':o~ «~1'.f' ;-<v>~-,:,;_-;;-:::!>;,:;:;:;'+;,-'.~~~'>':,'.' F~ '.'.;'.'.~ ,,,~ >;~:~,: ', , _, -

Wishart, Cunningham, and Carnahan (1997). An the timing requirements, presented three times
important component of the prediction relates to before each action.
the information in working memory when a trial The absolute constant error ( ICE I) results are
is practiced. Instead of trying to induce short- presented in figure 11.17. Two points of interest
term forgetting, Lee and colleagues attempted are noteworthy. First, even though practice was
to introduce into working memory the informa- conducted in a random order, the performance
tion necessary for the upcoming trial's action of the group given modeled information was
plan by means of a model. If the action plan was excellent. In fact, this random plus model group
provided by the model, then the problem-solving outperformed the blocked group on the very
activity normally undertaken during random first block of practice trials. And, second, there
practice would be unnecessary, and the beneficial was a strong negative influence of the model in
effects of random practice would be eliminated immediate and delayed retention tests. The learn-
or at least reduced. The experimenters used a ing advantage normally seen following random
timing task, with subjects making patterns of key practice was eliminated by the presence of the
presses on a computer keyboard. The random and modeled information during practice. These find-
blocked practice groups performed in acquisition ings were replicated and extended by Simon and
and retention as expected. A third, random plus Bjork (2002). In their study, the modeled informa-
model group also practiced in a random order; tion either matched the action plan requirements
however, the computer generated a visual map for the upcoming trial (as in Lee et al., 1997) or
of the task along with an auditory template of was inappropriate for the upcoming trial. The
380 Motor Control and Learning

200

180

! 160
......0 140
Q)

'E 120
.srn
s:: 100
0
0
.l!!
:s 80
0
rn
.Q
<(
60

40

20

2 3 4 5 6 Immediate Delayed
Acquisition trial blocks Retention

FIGURE 11.17 Elimination of the random practice decrement in practice and the random practice benefit in retention with a
guiding model.
Reprinted with permission from Research Quarterly for Exercise and Sport, vol. 68, pg. 103. Copyright 1997 by the American Alliance for Health, Physical Education, Recreation
and Dance, 1900 Association Drive, Reston, VA 20191.

information provided by the inappropriate model when only one GMP is practiced (Giuffrida, Shea,
was detrimental to acquisition performance but & Fairbrother, 2002; Lai & Shea, 1998; Lai, Shea,
beneficial for retention, providing further evi- Wulf, & Wright, 2000; Shea, Lai, Wright, Immink,
dence for the reconstruction hypothesis. & Black, 2001). This latter result is inconsistent
Another issue that is central to the reconstruc- with the reconstruction view because it predicts
tion view deals with the nature of the planning that no differences between random and blocked
activity. The hypothesis is that, for random prac- practice should result. When only a rescaled action
tice, the action prepared for one task is inappro- plan is required, the same relative timing can be
priate for a completely different task, and thus a maintained in memory on subsequent trials and
new preparation is required. Experimenters have little reconstructive process should be necessary
shown, for example, that more attentional capac- (Magill & Hall, 1990; Lee et al., 1992).
ity in general (Li & Wright, 2000) and motor pro- Finally, Trachtman (2003) has used a paradigm
gramming time in particular (Immink & Wright, that examines the benefit from the action plan-
1998, 2001) are required to prepare an action plan ning per se. She used so-called "abort" trials, in
for an upcoming trial in random practice than in which the action is planned and readied but at
blocked practice. Measures of functional magnetic the last second is not executed. According to the
resonance imaging also show greater activity in reconstruction view, if the number of executed
the brain's planning regions in random practice trials is held constant, additional abort trials
(Cross, Schmitt, & Grafton, 2007). Research sug- should be beneficial to learning because of addi-
gests also that random practice is more beneficial tional reconstructions provided. Trachtman' s data
than blocked practice if task variations require fail to provide any evidence for the benefits of
learning different GMPs, or parameter learning abort trials, however, and thus offer no support
when all tasks belong to the same GMP (Hall & for the reconstruction view.
Magill, 1995; Lee, Wulf, & Schmidt, 1992; Wulf
& Lee, 1993). However, the evidence also sug- Evaluating the Hypotheses
gests that blocked practice is more beneficial On balance, there appears to be considerable sup-
than random practice for relative timing learning port for both the elaborative processing account
 Conditions of Practice 381

and the reconstruction view of the CI effect, distinction discussed earlier in this and the preced-
although neither is able to explain all of the find- ing chapter. Here we have a situation for which the
ings that now exist on the topic. One of the prob- conditions in acquisition that make performance
lems in comparing these hypotheses is that there most effective (blocked practice) are not the most
are few situations in which different predictions effective for learning-an important general con-
can be contrasted. Thus, the hypotheses should sideration for those designing workable practice
not necessarily be seen as competing predictors of sessions (Bjork, 1994; Schmidt & Bjork, 1992).
the CI effect, but perhaps rather as complementary Although the application for these ideas is strongly
theoretical views about the ways in which learners implicated (Dempster, 1988; Goettl, 1996), much
comply with the processing operations encour- work remains to be done on these issues with more
aged under different practice and task conditions. complex tasks (Wulf & Shea, 2002) and various
training settings before we can be confident about
Other Hypotheses
how to effectively apply these principles.
Of course, it is possible that both the elaboration
and the reconstruction views are wrong. Wulf and Two Caveats
Schmidt (1994a) argued that random practice is
If, indeed, beneficial effects of random practice
beneficial for learning because it makes response
feedback (KR) less useful during practice-which, are found to be useful for training settings, then
another problem will need to be considered. In
as we will discuss in chapter 12, is often an effec-
their research discussed earlier, Simon and Bjork
tive strategy for learning. Others have suggested
(2001, 2002) asked their subjects during practice
that the detrimental effect in retention occurs
and just prior to retention to predict their level
because of greater retroactive inhibition: The early
of success in performance on the delayed reten-
tasks practiced by the blocked group are more dif-
tion test. If subjects were aware that performance
ficult to recall because of the interference caused
effects are misleading as predictors of retention,
by practicing the other tasks in the interim (Del
we would expect subjects in a blocked practice
Rey et al., 1994; Shea & Titzer, 1993). Another view
group to realize that their expected performance
downplays the role of cognitive factors and sug-
in retention would be poor. However, as illustrated
gests instead that CI effects reflect a "difficulty"
in figure 11.18, Simon and Bjork (2001) found
arising from an interaction among factors that
that this was not the case at all. Rather, blocked
include the learner, the task, and the environ-
practice resulted in overestimated levels of reten-
ment (Davids, Button, & Bennett, 2008; Ollis et
tion-error in retention was more than double what
al., 2005). Lastly, some researchers have mod-
had been predicted just prior to the retention trials
eled these effects using a connectionist approach
(compare the open and filled circles in retention).
(Horak, 1992; Masson, 1990; Shea & Graf, 1994).
This finding underscores a persistent problem, that
performance effects are often mistaken for learning
Practical Implications
effects (Bjork, 1994). As a result of this problem, it
Regardless of the theoretical explanation for these is difficult to encourage students, instructors, and
curious effects, it is clear that they are present in other training program organizers to use practice
both laboratory and practical situations, that they conditions that appear to be ineffective for learning
lead to relatively large effects in learning, and that during the training period. The implications are
they should have important practical implications that some explanation about the learning process
for the design of learning environments in sport, may be critical before the effects of CI and other
industry, and therapy. The "traditional" methods practice-related phenomena can be successfully
of continuous drill on a particular action (i.e., employed in training settings.
practicing one skill repeatedly until it is correct) But, let's consider one more conundrum before
are probably not the most effective way to learn. ending this section. Presumably, if learners were
Rather, the evidence suggests that practicing a poor judges of their own learning, then allowing
number of tasks in some nearly randomized order them to regulate their own practice schedules
will be the most successful means of achieving the would probably result in considerable blocked
goal of stable retention and transfer. Of course, practice, and very little random practice. That
these findings highlight the learning-performance seems to be true for some individuals. However,
382 Motor Control and Learning

20
18

~
16
0~
~ 14
(.)
Q)
12
'S
0
UI
..c
10
<(
8
6
4

2 3 4 5 6 Ret 2 3 4
Predicted performance Actual performance
Trial blocks

FIGURE 11.18 Effects of blocked and random practice for actual and predicted levels of performance and retention.
Adapted, by permission, from D.A. Simon and R.A. Bjork, 2001, "Metacognition in motor learning;• Journal of Experimental Psychology: Learning, Memory, and Cognition 27: 910.
Copyright© 2001 by the American Psychological Association.

some tend to choose a more random schedule reasons exist for breaking down tasks. One is to
and some a series of "mini-blocks" of trials; still eliminate the burden of repeating the simpler
others choose a progression from blocked to parts of the entire task, and another applies when
random practice (Keetch & Lee, 2007; Wu, 2007). the task is very complex and cannot be grasped as
However, regardless of the specific nature of the a whole. Examples are numerous, such as prac-
schedule that individuals choose, the very fact ticing separately the arm and leg strokes in Red
that they have self-regulated their own practice Cross swimming methods and practicing specific
appears to provide a boost to learning when stunts in gymnastics that later become part of a
compared to yoked controls and experimenter- larger routine. The ultimate test of whether or not
determined schedules. These findings offer a these methods are effective is the amount of trans-
tantalizing suggestion that perhaps the optimal fer from practice of the parts to the performance
variability in learning multiple tasks occurs when of the whole task. It seems obvious that if practice
it is self-determined by the learner (see also the is given on the part, it would certainly transfer
section of self-regulated augmented feedback highly to the whole task, as the part would seem
in chapter 12). Adaptive, or learner-contingent, to be identical to one element of the whole. The
schedules have also been explored recently as a problem with this idea is that practice on the part
means to tailor the schedule to the needs of the in isolation may change the motor programming
learner (Huang, Shadmehr, & Diedrichson, 2008; of the part, so that for all practical purposes it is
Choi, Qi, Gordon, & Schweighofer, 2008; Simon, no longer the "same" as it is in the context of the
Lee, & Cullen, 2008). We predict that the success total skill. The research suggests that whether or
of these types of practice schedules will be deter- not part practice or whole practice is effective
mined when the subject's long-term (learning) depends largely on the nature of the task (Wight-
interests are tailored to the schedule, rather than man & Lintern, 1985).
to the interests of immediate performance gains.
Serial Tasks
Part Versus Whole Practice Seymour (1954) conducted extensive research on
industrial tasks that are serial in nature. One task
A common technique for teaching motor skills is consisted of a series of elements to be performed
to break them down into smaller parts. Several on a lathe. Some of the elements were difficult,
 Conditions of Practice 111! 383

requiring a great deal of practice to master; some in sharp contrast to the situation for serial tasks,
were easy and could be accomplished on the first in which discrete parts are sequentially organized.
try. Seymour found that if the difficult parts were Also, in continuous tasks the parts must often be
practiced separately, without any correspond- coordinated with each other; and breaking into this
ing practice on the less difficult parts, there was pattern of coordination to practice a part might
considerable transfer of the part to the whole not appear to be highly effective, as it is the coor-
task. Similar findings were produced by Adams dination between the parts that must be learned.
and Hufford (1962), who studied part practice Swimming strokes have this characteristic, as the
of various discrete actions involved in aircraft arm strokes, breathing, and kicking actions must
flying, and Mane, Adams, and Donchin (1989), be coordinated to form an effective whole.
who studied multicomponent video games. Briggs and Brogden (1954) and Briggs and
An interesting variation of the part-task Waters (1958) used a lever-positioning task that
technique was used by Dubrowski, Backstein, required positioning in two dimensions (forward-
Abughaduma, Leidl, and Carnahan (2005) to backward, left-right) simultaneously and continu-
train medical students in a surgical bone-plating ously, much like the positioning of the "joystick"
task. This particular task involves multiple steps that controls the motions of an airplane. They
that are usually undertaken in a specific serial found that practice on the separate dimensions
order. Practicing the parts of the task in the same alone transferred to the whole task, but that this
order as required for surgery resulted in the practice was less effective than practicing the
most effective learning. However, Dubrowski whole task for the same period of time. It is pos-
and colleagues found that part practice was also sible that the most effective way to learn such
effective, but only if repetitions of the parts were tasks is to practice the whole, unless the task is
practiced in a random order-blocked practice highly complex or contains rather trivial elements.
severely degraded learning. Another situation that produces slightly dif-
In some cases, transfer from part practice can ferent findings involves those tasks in which the
be greater than 100%, meaning that the benefits parts interact while they are being performed
afforded by some amount of practice on the parts simultaneously. As an example, one complex
in isolation can be greater than those obtained task involves the operations necessary to take a
with an equal amount of time devoted to the helicopter from the ground into flight. Accord-
ing to Zavala, Locke, Van Cott, and Fleishman
whole task (see also Newell, Carlton, Fisher, &
(1965), the operator must handle four separate
Rutter, 1989; Wightman & Sistrunk, 1987). One
controls simultaneously. The first is a cyclic pitch-
way of viewing these effects is that, in part-task
control stick, which is really a stick control for two
practice, the learner does not have to spend time
dimensions in one (roll and pitch). When moved
on the parts of the task that already have been
in a particular direction, it causes the helicopter
mastered, and efficiency is improved because
to tilt in that direction. Thus, it can be used to
practice can be devoted to those parts of the task
control roll (side to side) and pitch (nose up or
that need it most. The task can be reassembled
down) simultaneously. Second, a collective pitch
in many ways, of course, but an efficient method
lever is mounted to the left of the pilot, and up-
is backward chaining, in which the last element in and-down movements of this lever control the
the sequence is systematically preceded by earlier vertical component of the flight. Third, a throttle
and earlier parts until the whole chain is com- is located as a twist grip on the pitch control just
pleted (Wightman & Lintern, 1985; Wightman & mentioned; it controls the engine speed in the
Sistrunk, 1987). Mere practice on a part isolated same way the accelerator in an automobile does.
from the sequence does not appear to be as useful Fourth, antitorque pedals under the pilot's feet
for transfer in serial tasks (Sheppard, 1984). control the pitch of the small propeller at the tail
of the helicopter, thus controlling its direction
Continuous Tasks and compensating for the torque produced by
For continuous tasks, in which the behavior con- the overhead rotors.
tinues more or less uninterrupted (as in walking or The problem for the person learning this task
steering a car), the parts that can be isolated occur is that these components interact strongly. That
at the same time as other parts. This is, of course, is, when the throttle control is used to speed up
384 111 Motor Control and Learning

the rotor or the pitch control is adjusted, there this finding was a study by Kurtz and Lee (2003),
is a tendency for the helicopter to tum in the who found that independently practicing the
direction opposite the rotation of the rotor. This motor parts of the rhythm while simultaneously
must be counteracted by an appropriately graded listening to both auditory streams was effective in
foot-pedal movement to maintain the proper learning the rhythmic structure of the whole task.
heading. But also, the helicopter will increase its This finding suggests that part-task practice may
tendency to roll and attempt to dive; both of these be effective in learning a continuous task that has
motions must be counteracted by the adjustment a perceptual component.
of the cyclic-control stick. Thus, when the lift of
the rotors is increased in the attempt to take off, Discrete Tasks
three other adjustments must be made simultane- Can we apply this evidence about part versus
ously to prevent it from turning upside down. whole practice to discrete tasks whose MTs are
The amount of control change in these three very short (e.g., less than 1 s)? Probably not, as
dimensions depends on the amount of lift that is the evidence gives a different picture in these
imparted to the helicopter via the other pitch con- situations. For example, Lersten (1968) had
trol. These control dimensions are said to interact subjects learn a hand movement task (the rho
because the setting of one of them depends on the task) that required a rapid movement with two
setting applied to the others. components. The subject grasped a handle and
It would be tempting to take this highly com- rotated it in the horizontal plane through 270°
plex task and break it down into separate parts. until it hit a stop, whereupon the subject was to
But this breakdown seems to sidestep the most release the handle and move forward to knock
important problem for the learner: how to coordi- over a barrier. This was to be done as quickly as
nate these actions. Any one of the dimensions can possible (MTs were about 600 ms). Thus, a circu-
easily be performed separately, but this practice lar component was followed by a linear compo-
would not seem to be very effective for learning nent, both of which were practiced separately by
the total task. In general, the limitations of part- part-practice groups. Another group of subjects
to-whole transfer methods probably depend on practiced only the whole task. Lersten found
the extent to which the parts of the task interact that practice on the circular component alone
within the whole task. As for the helicopter, it transferred only about 7% to the performance
would seem that an effective way to learn the task of the circular phase in the context of the whole
would be to practice in a ground-based simulator, task. Other conditions produced no transfer to
where all dimensions would be learned together the whole task. Even more important, Lersten
without the fear of an accident (see more on simu- found that the practice on the linear component
lators in chapter 14). alone transferred negatively (about -8%) to the
Although experimental investigations of part- whole task. That is, practicing this linear com-
whole practice in learning to control a helicopter ponent in isolation produced less transfer to the
have not been conducted, the conclusions just whole task than not practicing at all! Overall, the
mentioned are certainly supported by research findings suggest that practicing these isolated
on polyrhythm learning. Recall from chapter 8 components of the whole task produced essen-
that polyrhythms are complex bimanual timing tially negligible transfer to the performance of
sequences in which the rhythms tapped by the the whole task.
two hands have different metrical structures (the
left hand tapping every 600 ms with the right Sequential Parts
hand tapping every 400 ms is an example of a 2:3 How can these findings be explained? First,
polyrhythm that has a simultaneous tap every even though the task Lersten used was "serial"
1,200 ms). Studies have shown that part-task in nature, it must be seen as quite different from
training, in which the rhythms of each hand are the serial tasks used by Seymour (1954) and
practiced separately, results in virtually no trans- others reviewed previously. One clear difference
fer to the performance of the polyrhythm (Klapp, relates to the overall MT-Seymour's tasks had
Martin, McMillan, & Brock, 1987; Klapp, Nelson, durations in the order of minutes, and Lersten's
& Jagacinski, 1998; Summers & Kennedy, 1992; task lasted for only about 600 ms. It seems reason-
see Summers, 2002, for a review). An exception to able to assume that Lersten's task was governed
 Conditions of Practice 111 385

by a single motor program that contained the program, with the program containing instruc-
instructions for both the circular and the linear tions for both hands (Heuer, Schmidt, & Ghod-
components, as well as the instructions for the sian, 1995). Practicing the movements of one hand
transition between the two (timing the release of in isolation probably results in the development
the handle, for example). If so, then practicing of a different program than practicing that" same"
the circular part in isolation would result in the movement in the context of a total two-handed
subject's practicing a program different from that skill. For example, data from Konzem (1987) sug-
involved in the circular part within the context gest that the motor program to make a V with the
of the whole skill, because the isolated circular right hand is probably different from the program
part did not entail a handle release. In contrast, required to make a V with the right hand and "/
the serial tasks that Seymour (1954) used might with the left hand simultaneously. The principles
be thought of as a series of programs that have underlying coordination of the separate limbs in
been strung together. Practicing one of them in an action are not well understood, and more work
isolation is the same as practicing that program on this is needed; but it seems clear that breaking
in the context of the total skill, so part-to-whole down a task into its components will not always
transfer should be high from this perspective. result in large part-to-whole transfer.
These ideas suggest that the major determinant
of whether or not part-to-whole transfer will be Lead-Up Activities
effective is the extent to which the movement is
A closely related question for the teaching of skills
governed by a single program. If the movement
concerns the use of so-called lead-up activities. In
is very fast, it will almost certainly be governed
these situations, certain simpler tasks are thought
by one motor program, and it should be prac-
to be in some way fundamental to the learning of
ticed as a whole. Second, if the movement is
more complex tasks, so the simpler tasks are for-
slower but there is a "break" in the action that
mally taught as a part of the procedure for learn-
is easily adjusted, it is possible that the move-
ing the more complex task. These procedures are
ment is governed by more than one program.
often used in gymnastics as the instructors talk
An example is the break between the toss action
of a progression of subtasks leading eventually
and the hit action in a tennis serve; the toss seems
to the complex goal action. The question can be
programmed, but then there is a feedback-based
thought of as one of transfer from the lead-up task
break so that the hit program can be adjusted to
to the goal movement. SU:ch activities might have
the exact location and timing of the ball toss. In a
springboard dive, the takeoff and tuck are prob- the disadvantage of being, by necessity, different
ably programmed, but the timing of the "untuck" from the goal action, and the motor transfer from
movements might be feedback based, determined them could be very small.
by visual or vestibular information. These tasks On the other hand, lead-up activities may
could probably be split into their component have many positive aspects. Recall from our dis-
parts for separate practice, and part-to-whole cussion earlier in the chapter that one effective
transfer would probably be higher. "off-task" practice activity involves perceptual
training, such as watching videos of opponents
Simultaneous Parts in open-skill sports. In this case, learning to use
A second situation occurs when the parts of the advance perceptual cues to anticipate a temporal
task are simultaneous, rather than serial as in or spatial event can facilitate the motor response
Lersten's task. Many examples exist, such as process. Research shows that perceptual learning
playing the piano (left and right hands) or any in the absence of overt motor responses can be an
other task for which one part of the body has to effective training technique.
be coordinated with another. Transfer research Additionally, in many tasks (e.g., stunts in gym-
on these questions is nearly nonexistent, though, nastics) there is a strong element of fear. Lead-up
and the decisions about part-whole transfer are activities, being simpler and less dangerous, may
mostly speculative. serve a useful role in reducing fear responses
First, it seems from the data presented in chap- that can be detrimental to learning a more
ter 8 that rapid, discrete two-handed simultane- complex movement. This fear-reduction aspect
ous movements must be controlled by a single is borne out in studies using "desensitization
386 Motor Control and Learning

techniques," whereby people are taught to which the learner was to make an elbow move-
eliminate phobic responses (e.g., fear of snakes, ment having a complex spatial-temporal pat-
or heights) by performing lead-up activities that tern (see figure 6.21), with MTs of 3 or 4 s. Three
bring them closer to the target fear (e.g., progres- of his groups are of specific interest here. One
sively more "realistic" snakes, eventually leading group practiced the task and received terminal,
to an actual snake; see Bandura, 1969, or Bandura, kinematic feedback (knowledge of performance,
Blanchard, & Ritter, 1969). Also, many lead-ups KP) after each trial. In addition, after the last in
are designed with a particular action in mind. In a block of 15 trials, subjects in this group were
gymnastics, again, the "kipping" action (a force- shown a plot of their last trial in combination with
ful, timed extension of the hip) is thought to be a template of the goal pattern. Another group was
involved in a large number of skills. The idea is given concurrent, real-time visual feedback of the
that learning to kip in one simple lead-up activ- movement together with a perfect template of the
ity possibly will transfer to the "same" action in goal movement. In a third group, the movement
a more complex and dangerous activity. device moved by the subject was mechanically
controlled such that deviations from the target
path were physically restricted. Practice was
Guidance conducted over three days, which also included
a transfer test on the third day during which all
A technique frequently used in teaching and in subjects performed the task with no augmented
rehabilitation involves guidance, whereby the information.
learner is physically assisted through the task The results, presented in figure 11.19, were
to be learned. Guidance refers to a variety of very dramatic. As can be seen, the guidance
separate procedures, including physically push- device restricted errors so that performance was
ing and pulling the learner through a movement nearly perfect throughout the entire practice
sequence, preventing incorrect movements by period. The concurrent feedback was also quite
means of physical limitations on the apparatus, or effective in reducing performance error during
even verbally "talking someone through" a new practice, but clearly not as effective as the guid-
situation. These guidance procedures tend to pre- ance device (although much more so than the
vent the learner from making errors in the task. terminal feedback). But Armstrong's results
What does the evidence on guidance suggest? showed that the guidance effects provided only
Much of the early research on guidance by Hold- temporary boosts to performance. As can be seen
ing (1970; Holding & Macrae, 1964; Macrae & on the right side of figure 11.19, the transfer trials
Holding, 1965, 1966), Singer (1980; Singer & Pease, were performed most accurately by the terminal-
1976; Singer & Gaines, 1975), and others (reviewed feedback group and very poorly by both the guid-
by Armstrong, 1970a), using various tasks and ance- and concurrent-feedback groups. In fact, the
guidance procedures, showed considerable latter two groups performed the transfer trials
positive effects of guidance procedures on perfor- at almost the same level of performance as had
mance during acquisition. We should remember, the terminal KP group early in practice. Together
however, that guidance almost certainly would with other results (e.g., Domingo & Ferris, 2009;
usually have strong effects on performance during Schmidt & Wulf, 1997; Singer & Pease, 1976),
the trials in which it is administered. By its very Armstrong's findings raise considerable doubt
nature, a guidance technique is a method that will about the benefits of guidance techniques as
prevent or severely limit the occurrence of errors. learning aids.
Of course, as we have discussed previously, per- There is research however, suggesting that
formance gains during acquisition may not repre- some guidance can be beneficial when inter-
sent relatively permanent changes attributable to spersed with active practice trials (Hagman,
learning, and the important question is whether 1983; Sidaway et al., 2008; Winstein, Pohl, &
such performance gains will survive in a transfer Lewthwaite, 1994) or when used with complex
test when the guidance is removed. tasks (Tsutsui & Imanaka, 2003; Wulf & Toole,
One of the definitive studies in this area was 1999). Still other research has used guidance that
performed by Armstrong (1970b). He compared is less physically restrictive than many tradi-
various forms of physical guidance in a task for tional types of physical guidance. For example, a
 Conditions of Practice 111 387

14

...0 12
......
..
CD
CD
::I
0..-.
10

II) 0
.c ...
ca ca 8
'tJ 0
.l!! II)
ca CD
... J:
Clo 6
CD C:
:E ~
CD
Cl 4
...ca
CD
cs:>
2

1 2 3 4 5 6 7 8 9 10 11 12 13 14 1 2
Blocks of trials Transfer

FIGURE 11.19 Comparison of physical guidance with terminal- and concurrent-feedback effects on acquisition and transfer.
Adapted from T.R. Armstrong, 1970, "Training for the production of memorized movement patterns," Technical Report 26: 15.

computer-operated steering mechanism allowed ballpark" so that later refinements can be made.
subjects in a study by Marchal Crespo and Second, guidance may be most effective for tasks
Reinkensmeyer (2008) to manipulate a wheel- that are very slow in time (e.g., matching some
chair around a virtual environment under three perceived state such as a force or a position). Pre-
training conditions: no guidance, fixed guidance, sumably guidance gives the learner an indication
or a guidance-as-needed condition. This latter of the nature of the state that is to be matched, so
condition permitted subjects to perform the task that movements may be adjusted to match this
without restriction until a tolerance for error in state later. Direct and indirect lines of evidence,
performance was reached, at which time the com- however, suggest that guidance will be less effec-
puter provided haptic guidance as it physically tive for tasks that are rapid and ballistic or for
corrected the tracking device. This condition tasks that involve the learning of motor programs
also eased the tolerance limits as proficiency on (e.g., Armstrong's spatial-temporal pattern).
the task improved over trials. 3 Marchal Crespo A final aspect of guidance that is rarely studied
and Reinkensmeyer found that providing some is the prevention of injury and reduction of fear.
guidance without eliminating all errors provided As with Marchal Crespo and Reinkensmeyer 's
positive transfer performance when all guidance "guidance-as-needed" robot, providing guid-
was removed, as contrasted with the fixed-guid- ance that is relatively "loose" until the individual
ance and no-guidance conditions. These findings produces an error can also be an effective way to
offer strong promise for the use of computers help people learn a dangerous task. Gymnasts use
and robots in the training of functional motor manual assistance and spotting belts regularly to
skills (Reinkensmeyer & Patton, 2008) and in ensure that a mistake will not result in a serious
rehabilitative settings (Reinkensmeyer, Emken, fall. Similar "techniques" are used by parents
& Cramer, 2004). when their children begin to ride a bicycle with-
These various findings permit a number of out the training wheels. An adult running down
tentative generalizations about guidance. First, the street beside the child, either with a guiding
guidance may be effective in early practice when hand or without, is an easily recognizable image
the task is unfamiliar to the learners. Much of the of "guidance as needed." Such procedures pro-
apparent contribution of guidance procedures vide a great deal of confidence for learners and are
is involved in getting the movement "into the highly effective in reducing the fear and potential
388 Motor Control and Learning

disruptions in skill learning. The key feature is than massed practice when the retention trials
that guidance is minimal, while fear of injury were conducted in a distributed fashion (which
is nearly eliminated. Even so, there remains the is consistent with specificity predictions), but also
problem of the guidance serving as a "crutch," so when the retention trials were massed (Bourne &
that when it is removed there will be a marked Archer, 1956). Also, recall that random practice
reduction in the skill level. Gymnasts say that it produced superior retention performance relative
is difficult to perform a risky new skill "out of the to blocked practice under conditions in which
spotting belt" for the first time, as the performer retention trials were either blocked or randomly
knows that now a mistake could cause a serious ordered (e.g., Shea & Morgan, 1979). Both of
injury. Even so, it would seem that there is no these findings are opposite to a strict specificity
substitute for practicing the skill on your own as prediction.
soon as it is safe to do so. What is happening here? We suggest that these
different practice effects are related to different
types of specificity phenomena that emerge as a
Principles of Practice function of the interaction between conditions of
Specificity practice and the conditions of retention (or trans-
fer). We discuss different types of specificity next.
Many of the issues addressed in this chapter
reveal a common dilemma for those designing
Sensory and Motor Specificity
practice settings-deciding how to establish Some motor learning researchers suggest that the
performance conditions in the acquisition phase internal representations for skills that are devel-
that will best prepare the learner for the condi- oped with practice are very specific to the condi-
tions under which the learning will be applied. tions of practice. These effects are illustrated nicely
The general hypothesis-that we should attempt in a series of studies by Proteau and his colleagues
to match those conditions in practice with those in which subjects aimed a stylus at a target (Pro-
expected in the "test" performance-is an old one, teau, 1992, 1995). Subjects in different conditions
based on empirical evidence and common sense. practiced this task with KR for varying numbers
In motor behavior research it has been called the of trials (ranging from very few trials to several
specificity of learning hypothesis (e.g., Barnett, Ross, days of practice); they were then asked to perform
Schmidt, & Todd, 1973), stemming from Henry's transfer trials without KR. Various practice condi-
(1958/1968) work on individual differences (see tions were used that manipulated the amount of
chapter 9). The view holds that because skills are inherent visual feedback the subject was able to
very specific (i.e., generally uncorrelated with gather before, during, and after the movement.
each other), changing the conditions under which These conditions ranged from full vision of the
a task is performed will require a substantial shift subject's arm, the stylus, and the target, at one
in the underlying abilities. Therefore, because extreme, to conditions in which no visual feed-
practicing a task under one set of conditions back was provided. Various visual conditions
and then performing it as a criterion task under have been included in other experiments (e.g.,
different conditions would require a shift in abili- Blandin, Toussaint, & Shea, 2008; Elliott, Lyons,
ties, the conditions in practice and "test" can be & Dyson, 1997; Proteau, 2005; Proteau & Isabelle,
equated whenever possible. However, note that 2002; Soucy & Proteau, 2001; Tremblay & Proteau,
similar ideas exist in the psychology of learning, 1998). In general, these studies show that, after
originating with Thorndike's theory of identi- practice, if the transfer test has required subjects
cal elements (Thorndike, 1906; see also Healy, to perform without visual feedback or vision of
Wohldmann, Sutton, & Bourne, 2006). the target only, then the groups that perform the
Although this straightforward principle is very best are the ones that learned the task with the
robust-that practice conditions should match least amount of vision during practice. Typically,
the conditions expected later in retention or the weakest performance is by the group that had
transfer-it is worthwhile to recall evidence pre- the most vision available during practice.
sented earlier in the chapter that seems to violate These findings may not seem too surprising, as
this specificity effect. For example, distributed- we know that vision tends to dominate all other
practice conditions were superior for retention sensory modalities when it is available. Thus,
 Conditions of Practice l!l 389

when practicing with vision the subject may The evidence for context specificity in motor
come to rely on its availability to support perfor- learning is not abundant, although it does appear
mance, and then will suffer when performing in to be consistent with the general set of findings in
the absence of vision. However, Proteau and his cognitive psychology. For example, subjects in a
colleagues took their research one step further, study by Wright and Shea (1991) learned sequences
showing that when vision was added in transfer, of key-press patterns, with the computer monitor
performance deteriorated considerably for the providing stimulus cues specific to each pattern
groups that had performed in the absence of vision (i.e., the information about which keys to press was
during practice (Proteau, Marteniuk, & Levesque, presented in different colors, shapes, and positions
1992). Findings such as these have led Proteau and on the screen and was accompanied by auditory
colleagues to suggest that learning involves a sen- cues specific for each pattern). Performance in
sorimotor representation that integrates the motor retention was maximized when the cues were
components with the sensory information avail- matched with the same patterns as had been prac-
able during practice. This representation results in ticed, leading to the conclusion that the stimulus
specificity during transfer such that performance information provided a context that was learned
is optimized when the conditions during transfer as part of the representation for the movement
require the same sensorimotor representation that sequence (see also Wright & Shea, 1994).
was learned in practice. A possible relation of this research to practical
Although the topic is not often discussed in experience is the so-called home advantage in sport.
relation to motor learning, there is evidence from A finding that appears in many team sports is that
exercise physiology studies that show a similar a higher proportion of wins (or points) is achieved
effect. For example, there are generally large when playing at home than when playing on the
specificity effects when training and performance road. This finding has been well documented in
comparisons involve the same types of exercise the literature: It has been remarkably consistent
(e.g., isometric and concentric exercise), the same for many years, across the various major team
ranges of motion, and to a lesser degree, the sports, and is found at both the college and
same movement velocity (Morrissey, Harman, & professional levels (Carron, Loughhead, & Bray,
Johnson, 1995; Sale & MacDougall, 1981). Physi- 2005; Courneya & Carron, 1992). Several potential
ological specificity effects have also been found hypotheses for the home advantage seem to be
by experimenters who manipulated levels of ruled out, such as effects of travel, crowd size,
arousal in practice and retention (e.g., Movahedi, and crowd aggressiveness. But it appears that
Sheikh, Bagherzadeh, Hemayattalab, & Ashayeri, one factor that cannot be ruled out is related to
2007), suggesting that this practice for this type of the idea of context specificity-that certain factors
specificity effect might represent a way to offset related to the court or field on which the game is
some of the test performance detriments due to played (and on which the home team practices)
"choking" (Oudejans & Pijpers, 2009). provide a home advantage. Perhaps the contex-
tual information provided by the surroundings
Context Specificity of the practice area constitutes a small advantage
A logic similar to that in Proteau's research when games are played in the same venue. This
hypothesis must be viewed quite cautiously,
methods has been used to examine specificity
however, as the evidence that lends support to it
of practice conditions in a more general way.
is not strong (e.g., Pollard, 1986).
This research has been done often in psychology
and reveals a kind of "mixed bag" of specificity
Transfer-Appropriate Processing
effects. For example, various environmental fac-
tors that compose a study context (room tempera- The specificity effects presented in the preced-
ture, color, etc.) have an influence on remember- ing sections seem to provide some guidelines
ing the information that has been learned. When for establishing determinants of the effective-
a person attempts to recall the information later ness of practice, when considered in light of the
(e.g., in an exam), the same contextual informa- conditions under which retention or transfer
tion, if present, can serve as cues to help retrieve will be conducted. However, trying to anticipate
the information (Davies & Thomson, 1988; Smith the conditions of retention or transfer, and then
& Vela, 2001). matching practice conditions to them, is often
390 Motor Control and Learning

difficult if not impossible in the real world. A suggests that it is the similarity of the underly-
rather different kind of specificity in learning has ing processes (not simply the conditions) between
to do with the processing that a learner undertakes acquisition and transfer that will be the critical
during practice. determinant of the effectiveness of practice. In
The concept of "transfer-appropriate process- these cases, the most effective practice conditions
ing" was first suggested by Morris, Bransford, are those that require subjects to practice and
and Franks (1977; Bransford, Franks, Morris, & learn the same underlying processes that will be
Stein, 1979; see also Lee, 1988). The idea is that the ultimately used in the retention or transfer test.
effectiveness of practice activities can be evalu- That is, practice will be most effective if it fosters
ated only in relation to the goals and purposes of the processes most appropriate to performance on
the transfer or retention test. That is, conditions the transfer test. Sometimes, of course, when the
in practice are said to be effective to the extent superficial environmental conditions described
that they engage processing that is appropriate for by the specificity of learning hypothesis are the
performance on the retention or transfer test. We same in practice and transfer, the underlying
can evaluate "relative amount learned" only with processes are the same as well. But often this is
respect to some particular transfer task or par- not the case.
ticular transfer conditions; acquisition conditions In relation to all the practice variables noted
that might be effective for one transfer test might previously, though the superficial conditions in
be ineffective for another because they stress dif- acquisition versus transfer may differ, the gain
ferent kinds of information-processing activities. provided by learning some new appropriate
Evidence of transfer-appropriate processing processing capability overshadows any switch in
was seen in a number of instances in this chapter. conditions, so that the overall result is improved
Distributed practice is more effective than massed performance on the transfer test. This hypoth-
practice for retention under both distributed and esis of processing specificity does not identify
massed retention trials. Variable practice can be the nature of the appropriate processes learned
more effective for retention of a specific task than in acquisition, however, and these still must be
specific practice on that task alone. Random prac- discovered by research.
tice is usually more effective than blocked practice
for both random and blocked retention orders.
And we also saw that observational learning can Summary
be enhanced by watching a learning model (a
model who is actually learning) as compared to This chapter deals with the major independent
an expert model (e.g., McCullagh & Caird, 1990; variables that affect the learning of motor skills and
figure 11.6). The transfer-appropriate processing thus those variables that influence the design of
view explains this latter finding as follows: The instructional programs. Of most importance is the
observer sees how the learning model attempts to amount of practice itself, as illustrated by the law
perform the task, receives information about the of practice. But one can do a considerable amount
results of the model's performance, and sees how of learning apart from physically practicing a
the model uses that information to make adjust- motor skill. Much of this learning involves the per-
ments on the next attempt. In other words, the former's trying to figure out what to do. Methods
observer is drawn into the same problem-solving that engage the learner in information-processing
process that she will encounter when actually activities and encourage problem solving will
performing the task (Adams, 1986). In contrast, likely benefit learning, especially when used later
observing an expert engages the observer in a in on-task practice. Procedures involving mental
kind of processing that will be very different from practice, perceptual training, and observational
the processing involved in the trial-and-error, learning have all been shown to be effective "off-
problem-solving activities one performs when task" methods that facilitate learning.
attempting to learn the motor skill. The structure of practice also has very impor-
The notion of transfer-appropriate processing tant influences on learning. Distributed practice
addresses more than just the contextual or inci- facilitates performance and learning more than
dental similarities between practice and reten- massed practice does, although these effects seem
tion or transfer situations. Processing specificity to be stronger for the learning of continuous tasks.
 Conditions of Practice 11 391

Practice sequences in which the task conditions b. Describe an "alternative" practice

are deliberately varied from trial to trial are schedule that combines the benefits
slightly more effective than constant-practice con- to performance and learning that are
ditions for adults and far more effective for chil- typical of blocked and random prac-
dren. Randomly ordered practice is detrimental to tice.
performance as compared to blocked practice, but c. Describe the learning of two industrial
facilitates retention and transfer. Decisions about tasks, one that would be expected to
whether to break down a task into its component benefit more from part practice and
parts for practice, or whether to practice the task another that would be expected to
as a whole, depend entirely on the nature of the benefit more from whole practice.
task. If practicing the parts means changing the
task itself, then whole practice will probably be 2. Find a research article that examines the
more effective. And guidance can be a useful effects of mental practice in the perfor-
aid in some situations, but overuse of guidance mance or learning of a sport skill.
techniques can also be detrimental to learning.
We have emphasized often that the value of Web Resources
practice sessions must be assessed in tests of
retention and transfer. Complicating this evalu-
ation of learning is the fact that performance Distribution of practice in baseball:
in these tests is influenced by the relationship http://baseballtips.com/practice.html
between the nature of practice conditions and the
nature of the retention and transfer conditions. Some principles of practice for sailboat racing:
Specificity in learning suggests that the sensory, www.sailingscuttlebutt.com/news/04/ras/
motor, contextual, and processing activities of
the retention and transfer tests influence to a
considerable extent the "value" that we attribute Notes
to certain practice conditions.
1Although we are not in favor of change (pretest-
posttest) scores to measure learning (see chapter 10),
Student Assignments here these scores are similar to absolute retention scores
because there were no differences between groups in
the pretest. Thus, the change scores are essentially an
1. Prepare to answer the following questions absolute-retention score with a constant subtracted.
during class discussion: 2 Note that these variable-practice trials were con-
a. Describe one task that might be ducted in a random order. This factor will become an
expected to benefit from distributed important variable, as discussed in the next section.
practice and another task that might 3 Note that this guidance-as-needed condition provides

be expected to benefit from massed a type of bandwidth augmented feedback that is faded
practice. over trials, discussed in greater detail in chapter 12.
 12

ne of the most important features of prac- tion produced by the movement, such as the
tice is the information learners receive way it feels, sounds, and looks. After the action
about their attempts to produce an action. Some is completed, information is available regarding
of this information is inherent in the movement the result that the movement produced in the
production; we have examined this kind of environment (e.g., the actions of a ball that has
sensory information in chapter 5; we can also been struck) and, for a brief time, a memory for
consider information that is presented in an the how the movement felt, sounded, and looked.
"augmented" form from the instructor, therapist, This latter class of information is usually termed
or coach. This chapter deals with this latter form movement-produced feedback, or simply feedback.
of information. The term "feedback" can be further subdivided
into two broad classes: inherent feedback (some-
times called "intrinsic" feedback) and augmented
Classifications and (sometimes called "extrinsic") feedback.
Definitions
Inherent feedback
Consider, as the broadest class, all the various People can gain information about many aspects
kinds of sensory information that people can of their own movements through various sen-
receive, including all those sources that have to sory mechanisms. These forms of information
do with the many diverse aspects of our lives. are inherent to the individual during the action,
Of course, not all such information is related to and result from the movement's execution. For
our movements: the sound of wind in the trees example, you know that an error was made in a
as we walk through a forest is not relevant in basketball shot because you saw that the ball did
this respect. Of the sources of information that not go into the basket. Also, the stinging sensa-
are related to our movements, we can speak of tions as you land on your back in a pool after a
those available (a) before the action, (b) during the faulty dive inform you that something probably
action, and (c) after the action. Before the action, went wrong. Just about every movement we
sensory information signals the position of your can make has associated with it certain sources
limbs, the sight of a ball flying toward you, the of inherent feedback that provide a basis for
nature of the environmental setting, and so on. evaluating those movements. Such feedback is
During the action, you receive sensory informa- usually rich and varied, containing substantial

393
394 Motor Control and Learning

information regarding performance. Depending mation from a buzzer when your car's engine
on the nature of the movement and the source exceeds a certain temperature-information that
of inherent feedback, sometimes the performer is not normally available during driving. Aug-
knows that something has gone wrong before mented information can be provided verbally, for
the movement is even completed. The informa- example in the presentation of one's time after a
tion provided as the movement is executed is 100 m race or the set of scores after a gymnastics
sufficiently useful that the movement outcome or ice skating routine. Even though these various
can often be predicted even before it occurs. At forms of information are not strictly verbal, they
other times the nature of the movement and the are in a form that is capable of being verbalized.
source of feedback are such that the evaluation of A number of useful dimensions for augmented
the movement must occur after it is completed. feedback are summarized in table 12.1. First, one
In many situations, inherent feedback requires can distinguish between concurrent and terminal
almost no evaluation at all; one sees that the bat feedback. Concurrent feedback is delivered during
missed the ball or one can feel the fall while walk- the movement (e.g., the information about engine
ing on an icy sidewalk. Thus, some errors seem speed that the racing driver receives from the
to be signaled immediately and clearly. But other tachometer), while terminal feedback is postponed
aspects of inherent feedback are not so easily until after the movement has been completed
understood, and perhaps the performer must (e.g., the gymnast's score). Another dimension
learn to recognize their occurrence and evaluate of augmented feedback is the time at which it is
what the feedback means. Examples might be delivered; it can be either immediate or delayed by
the gymnast learning to sense whether or not the some amount of time. The feedback can be verbal
knees are bent during a movement, or a patient (or capable of being verbalized) or nonverbal (e.g.,
with a recent hip replacement who is learning to a buzzer indicating that the car's engine is too hot).
put partial weight through the leg while walking Also, the performance can be sampled for a period
with canes. It is thought that inherent feedback is of time, with the accumulated feedback indicating
compared to a learned reference of correctness, the average performance for the past few seconds;
with this reference acting in conjunction with the or the feedback can be distinct, representing each
feedback in an error-detection process. Without moment of the performance (e.g., feedback from
such a reference of correctness, many forms of a speedometer). (See Holding, 1965, Annett, 1969,
inherent feedback probably cannot be used to and Singer, 1980, for additional dimensions.)
detect errors. These various dimensions of augmented
feedback should be considered independent
Augmented Feedback of one another. For example, if the augmented
In contrast to inherent feedback, augmented feed- feedback is terminal, it could be either verbal or
back is information provided about the action that nonverbal, and it might be delayed or immediate.
is supplemental to, or that augments, the inherent These dimensions, then, should be thought of as
feedback. For example, you can receive infor- separate descriptors of augmented feedback that

Concurrent: Presented during the movement Terminal: Presented after the movement
Immediate: Presented immediately after the relevant Delayed: Delayed in time after the relevant action
action
Verbal: Presented in a form that is spoken or capable of Nonverbal: Presented in a form that is not capable of
being spoken being spoken
••O•••••O•••O••O•••O•M•MOoooooO•o•••••••••o ............................ ;...
Accumulated: Feedback that represents an accumula- Distinct: Feedback that represents each performance
tion of past performance separately
,
.•....•..... .•.
Knowledge of results {KR): Verbalized (or verbaliz- Knowledge of performance {KP): Verbalized (or ver-
able) postmovement information about the outcome of balizable) postmovement information about the nature
the movement in the environment of the movement pattern.
 Augmented Feedback 111 395

define most kinds of feedback commonly used. such as the behavior of a particular limb in a
~omplex movement. And it can refer to processes
Knowledge of Results
m the body about which the subject is normally
One of the important categories of augmented unaware, such as blood pressure or the activity
feedback is termed knowledge of results (KR). of a particular motor unit-often referred to as
Essentially, KR is verbal (or verbalizable), ter- biofeedback (Basmajian, 1989).
minal (i.e., postmovement) feedback about the
outcome of the movement in terms of the environ-
mental goal. It forms one combination of the vari- Research on Augmented
ous possible dimensions of augmented feedback Feedback
(verbal-terminal) shown in table 12.1. Examples
are seen when the instructor says "You were 2 m
How do scientists conduct research to understand
off target that time" or a computer screen presents
the symbolic information "long 12" (meaning that feedback and learning? What forms of feedback
are useful in motor learning, and how are these
the movement was 12 units too long). Knowledge
of results can be highly specific, or it can be very forms of feedback most effectively presented to
general. Knowledge of results can also contain the learner? A major problem for such research
a rewarding component, such as "very good." is that, in most natural situations, it is difficult to
It is important to be clear about the use of the control the information received by a performer,
term KR. First, note that KR is about movement so the situation is not easy to study. For example,
outcome in terms of an environmental goal ("You there are many sources of feedback in the task of
missed the ball"). KR is not feedback about the shooting a basketball, and it is difficult to know
movement itself ("Your elbow was bent"). Usu- which sources are being used at any one time
ally this distinction is easily made; in shooting a and how they are being used. A typical strategy
basketball, for example, the goal and the move- used by many researchers in motor behavior is
ment to produce it are clearly separable. But often to alter the environment or the task (or both) so
these two aspects of feedback are difficult to that minimal feedback information is provided
distinguish-for example, in a situation in which to the subject, and then provide augmented
the goal of a movement is the form of the move- feedback information artificially (in the form
ment itself, as in a gymnastics move. Occasionally, of KR or KP) so that the effects can be studied
other terms are used for KR as defined here, such directly. This technique usually involves experi-
as information feedback (Bilodeau, 1966), extrinsic ments with tasks that are artificial and novel, but
feedback, or reinforcement (which implies a reward). a basic understanding of the functioning of error
Despite these inconsistencies, the tendency is to information can result just the same.
use the term KR as we have defined it here: verbal,
terminal, augmented feedback about goal achievement. Paradigms
(See the review by Salmoni, Schmidt, & Walter, for Augmented-Feedback Research
1984, for additional distinctions.) Although many definitions exist (Kuhn, 1962),
Knowledge of Performance a paradigm often refers to a standardized way of
gaining knowledge through research. The study
As already mentioned, an additional kind of feed-
of KR variables1 in motor learning research was
back information concerns the movement pattern
directly influenced by research in experimental
that the learner has made (e.g., "Your elbow was
psychology, and these traditions remain today
bent"). Gentile (1972) called this type of feedback
(see "Origins of the KR Paradigm"). Seldom
knowledge of performance (KP) to distinguish it from
stated explicitly is the assumption that the (aug-
KR as defined previously (see table 12.1). Knowl-
:n:ent~d) ~provided in these artificial learning
edge of performance is probably more related
situations is fundamentally like the (inherent)
to the feedback that instructors give to their
erro_r i~formation a person would normally
students, being directed toward the correction of
receive ma more natural setting. Is it correct to
improper movement patterns rather than just the
say that the information "You moved 2 cm too
outcome of the movement in the environment.
far" in a blindfolded linear-positioning move-
Also, KP can refer to aspects of the movement
ment works fundamentally in the same way as
about which the subject is only vaguely aware,
396 m Motor Control and Learning

the information received by observing visually a should remember that the principles might not be
shot missing the basket in basketball? Certainly quite the same in these two situations.
different processes are involved, but it is entirely The dominant paradigm for understanding
possible that the use of the error information is the functions of feedback information in learn-
the same in both situations, in that the informa- ing is a legacy from the historical influences of
tion provides a basis for changing the movement experimental psychology (discussed in "Ori-
on the next attempt in order to make it more gins of the KR Paradigm"). The KR paradigm
accurate. If this assumption is correct, then this frequently uses a movement task that is very
general method provides a way to come to an simple; the most common task used in early
understanding of the way in which inherent research investigations was the linear-positioning
feedback works to produce learning in natural task, for which the person must learn to move a
environments. slide or a lever to a given position, usually while
The other side of the argument is that such blindfolded. In such tasks, the subject cannot
research, using tasks that are so simple and arti- evaluate performance outcome without some
ficial, may have little to tell us about the ways supplemental information because of the removal
in which the rich and varied sources of inherent of the most potent source of inherent information
feedback work in more natural settings. For now, (vision). If the instruction is to move 20 cm, the
our assumption will be that the study of KR is one subject cannot know for certain whether a given
means to understanding the operation of inherent attempt to move that distance was correct or
feedback in natural environments. However, you not on the basis of inherent information. True,

. ··· ... ·.. • .·~~t~g~.Thorndike)Guthrie,Tolman, Hull, and others during

· · aJramework.forresearch that remains today. One of the
..;;~~nditioningparadigm~ influenced largely by Thorndike.
•ietea.thatif an.animal's behavior was followed quite soon
~lidtt~fit1l~rsfretjuentty·underthese conditions•in the.future. In theory,
· 'fu~d IretWe~n\the situation .and the behavion The association was
'ife~e~t~~l~· 0r~.i[Jfotcetj'~ (by the reward). Thus, the reward was
bfr~nce ?fl~arnin.g, .•..•· •·.• ..
· ··. lrig.j:iatadigm is the assessment of learning by means of
f#1P~vlov torefert~ the apparent elimination of the learned
rri:~nt~l.learnipg Paradigm during a period of time when the
reinforc~et;·Strength of the conditioned response is mea-
.·e;continued hehavior in the absence of the reward.
·• gtudy Of ~otot behavior began with Thorndike
\·rri.rli~Etss~qjectsdrew lines of 3,4, 5, and 6 in.
•.·.. ·.· ·.. ;sessi()p~~aw petiormance improve steadily with
:·~ft\§lr~E.l~·:ge~e.riqration in performance. In Thorndike's
n~: tijei,C~~(l~~ti.on betwe~~ a stim.ulus (the movement
· .·. q'f:emr(lt), arl~ .f<FI Was. viewed as i.nstrumental in
.no~l<iR friars. was to• stuciy. the ,;strength" of the bond
'1,,://·) ·C,>~·,;:,:'~\J,;',,,' ·\~ ',''

~awi~d~~~~tirt,e~fwas not to study the laws of motor

E~lity O:f .t\Js. L~w 9t E:ffect using a motor task. For our
i$.1 t~tn~rribered f()~ introducing the KR/no-KR paradigm
· st::Jeqrfi~C1.l1¥ii;i::numan ..motor learning. This influence
, rrici1kfiE1.1~9aei~s {ct Ada{l1s, 1978).
 Augmented Feedback 111 397

the feedback from the limb is present to signal processing the KR and planning the next move-
the movement details, but the individual likely ment. The sum of the KR-delay and post-KR-
does not have the reference of correctness against delay intervals is termed the intertrial interval.
which to evaluate this source of inherent feed- Usually the intertrial interval is on the order of
back. In some sense, the feedback has not been 10 to 20 s, but of course these intervals can be
"calibrated" to the environment. With this kind practically any length to serve the purposes of a
of task, one can study the use of feedback or KR particular experimental situation.
by "augmenting" information to the subject in
Learning Versus Performance Effects
a systematic way. The most elementary of these
experiments might involve the contrast between In the typical KR paradigm, the variables (such as
providing KR and withholding KR altogether. A amount of KR; absolute and relative frequency;
more refined experiment might manipulate the precision; length of, and activity during, the KR-
time of presentation of the KR, the way in which delay, post-KR-delay, and intertrial intervals) are
the KR is presented (e.g., on a computer monitor typically manipulated over a series of acquisition
or verbally, by an experimenter), or the qualita- trials, just as we have discussed in chapter 11.
tive aspects of the KR (e.g., imprecise or precise). After these trials, all the conditions of the par-
In this way, experiments that vary the nature of ticular KR manipulation (preferably involving
the feedback given to the learner can be done in separate groups of subjects) are transferred to a
the same ways as experiments about any other common condition of KR for additional perfor-
independent variable. Thus, the task used must mance trials. By far the most common transfer test
allow control over the relative usefulness of the is a series of no-KR (or "KR withdrawal") trials.
sources of inherent feedback. Although other paradigms have been used, the
no-KR transfer test has a long history of use in
Temporal Placement of KR experimental psychology, upon which much of
Many of the experiments on KR and motor learn- motor behavior research in this area is based (see
ing are structured so that the temporal relation "Origins of the KR Paradigm").
among the events in a trial is closely controlled. Salmoni and colleagues (1984) provided a strong
These events are shown in figure 12.1. The subject argument in support of the typical paradigm, in
performs movement 1 (M1); then, after a period of which a KR variable is manipulated during prac-
time called the KR-delay interval, the KR for that tice trials and the effects of that manipulation are
trial (KRi) is delivered by the experimenter. The evaluated in a common, no-KR transfer test. The
period of time from the presentation of KR until authors argued that the two phases of the typical
the next movement is termed the post-KR delay, KR paradigm permitted a direct comparison of
during which it is presumed that the person is the effects of a KR variable on performance and

FIGURE 12.1 Temporal placement of events in the knowledge-of-results (KR) paradigm. M1 refers to movement trial 1. KR 1
refers to the augmented feedback provided about results of movement trial 1.
398 Motor Control and Learning

learning (see chapter 10 for more explanation). and many motor behavior workers chose not to
Making a distinction similar to other distinctions use them. However, with the use of computing
between learning and performance (chapters technology and increased emphasis on biome-
10 and 11), they argued that a KR variable that chanical techniques, researchers have examined
exerted an influence only while being manipulated KP as a source of error information much more
was a performance variable. A KR variable that frequently. For now, we will assume that the
exerted an influence after the manipulation was mechanisms involved when the learner receives
withdrawn (and after the temporary influences any type of augmented feedback are essentially
had dissipated) was a learning variable. the same. That is, we assume that what the learner
Several arguments support the preference for does with these various kinds of information is
the no-KR transfer test over other transfer tests in identical, the major distinction being that these
the assessment of KR effects on performance and different kinds of information refer to different
learning. One argument is that learning can be aspects of the movement. Thus, for example,
addressed in a more steady state under no-KR than the principles that have been discovered for KR
under KR trials, since continued improvements in would be applicable to situations when KP would
performance are unlikely to occur in the absence be given. This could be incorrect, of course, but
of KR (Salmoni et al., 1984). A further contention until evidence appears to the contrary, we think
is that a series of no-KR trials provides a more the assumption is reasonable.
consistent estimate of performance capabilities
and thus a more reliable account of learning effects,
since performance is stabilized more in the absence
Evaluating the Effects of
than in the presence of KR (Rubin, 1978). Another Augmented Feedback
argument is that the use of a no-KR test of learn-
ing is consistent with many practical applications: In this part of the chapter, some of the fundamen-
Augmented information supplied during training tal principles of augmented feedback for motor
or rehabilitation is often unavailable when "real" learning situations are presented. A number of
performance is required (e.g., in a game situation conclusions can be drawn from the literature,
or when a patient is away from the clinic). probably because this area has received a great
deal of study in motor skills research (for reviews
Potential Applications see Adams, 1987; Magill, 2001; Salmoni et al.,
of Augmented-Feedback Research 1984; Swinnen, 1996; Wulf & Shea, 2004). Also, the
effects found are very robust and large relative to
The vast majority of the research on augmented
those of other variables considered. First, we dis-
feedback and motor learning has involved infor-
cuss a basic question: whether or not augmented
mation about movement outcome (KR). For a
feedback is a variable affecting performance,
person who has had extensive practice at a sport
learning, or both. Then we discuss the research
or occupational activity, it would seem far more
variables related to KP, and finally, we present
effective to provide information about the patterns
the rather large and complex set of effects of KR
of movement the person made-defined earlier
variables on performance and learning.
here as KP. Why is there a focus on KR (move-
Most of us probably suspect that KR has impor-
ment outcome) when KP (movement pattern) is
tant effects on both performance and learning, so
what will probably be most useful for applica-
it is perhaps not crucial that we document these
tion? Probably the most important reason is that
effects. But we have been fooled by our intuitions
in experiments on KR, the movement outcome
before, so we will review briefly some of the criti-
can usually be measured easily and corrections cal evidence on this issue.
on the next trial can be measured. But when the
experimenter wants to give KP, there is more Augmented feedback Is
difficulty in measuring the pattern of movement
and then noting how the pattern changed on
a Learning Variable
the subsequent trial. Until late in the previous Using the paradigm described in the previous
century, these procedures were tedious (using section, Bilodeau, Bilodeau, and Schumsky (1959)
film analysis, strip-chart records, and so on), employed a linear-positioning task with four
 Augmented Feedback 111 399

groups of subjects. One group had KR after the phase. But another possibility is that KR had
first 19 of the 20 acquisition trials, and a second affected performance only temporarily, perhaps
group received no KR at all in the 20 trials. Two through some kind of motivational or "energiz-
other groups received KR for two and six trials, ing" process. Thus, it could be that when these
respectively, before having KR withdrawn for temporary effects of KR are allowed to dissipate
the remainder of the 20 practice trials. The main with rest (as with fatigue effects), the temporary
findings are shown in figure 12.2, where absolute effects of KR will vanish and performance will
error is plotted as a function of trials for these regress to the original level (see chapter 10).
four groups. The group that had KR provided Bilodeau and colleagues provided a partial
after trials 1through19 showed an initial sharp answer to this question when they transferred
decrease in error, followed by a more gradual their no-KR group to the KR conditions for an
decrease. On the other hand, the group that had additional five trials. In the right portion of figure
no KR at all showed essentially no change in 12.2, the absolute errors on these five trials are
performance over the 20 practice trials. For the plotted together with those for the first five trials
remaining two groups, improvement occurred of group 19. The size of the errors, as well as the
on trials that followed the administration of KR, pattern of change with trials, was practically iden-
but the improvement ceased when KR was with- tical for these two sets of trials. That is, the no-KR
drawn, with slight decrements in performance group in this transfer condition performed nearly
thereafter. the same as group 19 at the beginning of their
Did KR affect the learning in this task? As practice trials. Thus, we can say that the 20-trial
with any other variable that could affect learning no-KR practice sequence for this group did not
or performance, these data can be interpreted produce any learning at all, and consequently that
in at least two ways. First, we could conclude KR is a learning variable. And, KR is not just a
that the 19-trial KR group learned more than the variable that affects learning; rather, when KR is
no-KR group, as evidenced by the fact that they not present in such situations, learning does not
performed more effectively during the practice occur at all. While Bilodeau and colleagues' study

14

12

"E 10
.§.
...0
...... 8
GI
....:IGI
0
Ill 6
.c
as
c
as
GI 4
:a:
2

5 10 15 20 1 2 3 4 5
Trials

FIGURE 12.2 Absolute errors in a linear-positioning task as a function of knowledge of results (KR). (The group numbers
indicate the number of presentations of KR received before KR withdrawal; group O switched to a KR condition shown at the
right, where its performance is compared to group 19's first five trials replotted from left.)
Reprinted from E.A. Bilodeau, l.M. Bilodeau, and D.A. Schumsky, 1959, "Some effects of introducing and withdrawing knowledge of results early and late in practice,'' Journal of
Experimental Psychology 58: 143.
400 Motor Control and Learning

uses a kind of transfer design, it does not use the but weaker effects on learning (e.g., Szalma, Han-
typical transfer procedures we recommended cock, Warm, Dember, & Parsons, 2006).
earlier (chapter 10). Similar conclusions, however, Another temporary effect of KR is related to
have come from a number of other studies in its informational properties, whereby KR informs
which relative amount learned was evaluated on the subject of the errors that have been made and
no-KR transfer tests (Bennett & Simmons, 1984; then indicates what to do next. Thus, KR provides
Newell, 1974; Trowbridge & Cason, 1932). something like guidance for the learner. In chapter
Knowledge of results does not always have such 11 we presented evidence that guidance is very
dramatic effects on learning motor skills, though, effective for performance when it is present but
and the reasons often depend on the availability that all or part of the beneficial effect can disap-
and usefulness of inherent feedback. For example, pear when the guidance is removed (e.g., Arm-
KR had only minimal effects on performance and strong, 1970a; see figure 11.19). In an analogous
learning for a tracking task in which KR was or way, then, KR (acting as guidance) might provide
was not provided after each trial (Archer, Kent, & strong informational support for performance
Mote, 1956; Bilodeau, 1966). Similar effects have when it is being administered, with the benefits
been found for learning an anticipation-timing disappearing as soon as the KR is removed or the
task (Magill, Chamberlin, & Hall, 1991). task conditions are changed (Salmoni et al., 1984).
Does this mean that augmented informa-
tion about errors is somehow not important for Untangling the Learning Versus
learning these tasks? Probably not. Rather, while Performance Effects
practicing the task, subjects are able to detect their
own errors through the inherent feedback (visual From the previous sections we have seen that
in these cases) provided during the normal course variations in KR can have powerful effects on
of the trial. This visual information probably performance when KR is present, but there is
serves the same function as the verbal KR did good reason to question whether such effects are
in the linear-positioning experiment described always "relatively permanent" to the extent that
earlier. This observation is in accord with the idea they can be thought of as learning effects. The
that the presentation of information about errors scientific problem is to distinguish the variables
to the learner is more effectively studied in situ- that produce transient performance changes from
ations in which learners are unable to evaluate those that produce relatively permanent changes.
accurately their inherent feedback to detect errors. Transfer designs used as discussed in chapters 10
and 11 provide a good way to make this distinc-
Augmented feedback Is tion in experiments on KR. However, except for
a few studies (e.g., Annett, 1959; Griffith, 1931;
a Performance Variable McGuigan, 1959; Trowbridge & Cason, 1932),
The evidence clearly points to (temporary) per- early feedback researchers did not take this learn-
formance effects of KR in addition to the learning ing-performance distinction seriously, apparently
effects we have just described. For example, KR assuming that any variation of feedback that
can be motivating, or "energizing," for the learn- affected performance was automatically a learn-
ers. Some early research shows that when KR ing variable. As we will see, there are many situa-
is provided, subjects report that they are more tions in which this assumption is simply incorrect.
interested in the task, they seem to put more effort
into practice, and they persist longer after the KR
is removed, in comparison to practicing without Knowledge of Performance
KR (Arps, 1920; Crawley, 1926; Elwell & Grindley,
1938). In relatively boring situations such as vigi- We begin our analysis of augmented-feedback
lance tasks, in which subjects are asked to spend variables by looking at studies of information
hours monitoring a display for the appearance of that is provided to learners about the patterns
a threatening object (e.g., in airport security moni- of actions they make. It was Gentile (1972) who
toring), KR about the subject's performance has an termed these kinds of feedback "knowledge of
"alerting" (or energizing) effect, and it can act to performance." Many forms of KP are possible;
counteract sleep loss (Poulton, 1973). All these phe- they may range from rather casual comments
nomena exert strong influences on performance, about performance, made by a teacher or coach,
 Augmented Feedback 111 401

to complex feedback generated by computer in a of video feedback than did studies using undi-
simulator and delivered to the learner online in rected viewing.
computer-aided instruction. Some of these kinds The benefits of cued or directed viewing of
of KP are discussed in the following sections. video feedback were shown clearly in a study by
Kernodle and Carlton (1992). Subjects practiced
Video Feedback throwing a sponge ball with their nondominant
It would certainly seem reasonable to think of arm. After each throw, they were provided with
analog or digital video feedback as a powerful KR regarding the distance thrown (subjects closed
mode in which to present KP. From a motor skills their eyes on ball release, making the augmented
viewpoint, a video will contain a record of the feedback more important for learning) and were
entire performance, and the individual can detect shown video feedback of the trial just completed.
errors directly and attempt to correct them on One group of subjects was provided only KR,
the next trial. However, for all the logic leading while another group watched a video replay
to the use of video feedback, as well as its use of their own performance, with no additional
in many sport situations, the research evidence augmented information. Previous research, how-
suggests that this method of presenting KP, by ever, had shown that combining verbal KP with
itself, is rather ineffective. Rothstein and Arnold other forms of augmented feedback can be quite
(1976) and Newell (1981) have reviewed this beneficial to learning (Wallace & Hagler, 1979).
work, finding that numerous experiments fail So, another group received a verbal cue to watch
to show positive effects of these techniques for one particular aspect of the movement during the
motor learning. Some evidence even suggests that video feedback (e.g., "Focus on the hips during
video feedback might actually hinder learning the throwing phase"). A final group, before
(Ross, Bird, Doody, & Zoeller, 1985). One sug- watching the videotape, was given additional
gestion is that video feedback might provide too augmented feedback in the form of specific error-
much information, especially if the skill is complex correction information (e.g., "Rotate the hips from
and the viewer does not know which of the many left to right during the throwing phase"). Figure
details are important. In support of this notion, 12.3 illustrates the subjective ratings of throwing
Rothstein and Arnold pointed out that studies performance (or form) during no-feedback trials
using cuing, in which subjects were directed or on five transfer tests over a four-week period.
taught to examine certain aspects of the display The results were clear: The strongest learning
during a viewing, showed more positive effects effects were seen when the video feedback was

en 5
c:
~
E4
.e
~u 3
GI
:c
::J
(/) 2

2 3 4 5
Test session

FIGURE 12.3 Throwing performance under various conditions of videotape replays.

Data from Kernodle and Carlton 1992.
402 ili1 Motor Control and Learning

accompanied by error-correcting cues, although skilled workers and then used this pattern as a
considerable gains were achieved with the atten- "gold standard" for providing feedback about
tion-focusing cues as well. The video feedback foot action to new employees. Such kinematic
alone was no better than simply providing KR. feedback greatly facilitated training; in as few
Similar results were obtained when measures of as 10 weeks of practice, new trainees could be
distance thrown were analyzed. brought to the level of employees who had nine
It is important to remember that video can months of experience. Knowledge of performance
also be used to present the performance of a about the most effective patterns of actions-not
model (chapter 11). However, in both uses of easily observable without additional measure-
videos as forms of augmented information, ments of the fine details of foot movements, and
research has shown that they are most effective not easily verbalizable-was apparently critical
when supplemented with additional, attention- to the establishment of proper actions in the new
directing augmented information. Practically performers.
speaking, information provided in videos is most A key feature of kinematic feedback is that it
effective for learning when it is augmented by an informs the subjects about some aspect of the
instructor who can direct the learner's attention movement pattern that is otherwise difficult to
to important details and toward ignoring the perceive. In some cases, a whole pattern of mul-
irrelevant aspects. This is probably especially so tijoint coordination is presented (e.g., by means
if the learner is a novice, who has less knowledge of analog or digital video), showing important
than an experienced performer about what details information about the movement of a particular
in the video are important. joint in relation to another (e.g., Hatze, 1976). It is
possible that the subject could gain this informa-
Kinematic Feedback tion on his own, but it is unlikely that a learner
would focus on the particular aspects that the
Recall that kinematics refers to measures of "pure instructor considers to be critical. Other kinds
motion" without regard to the forces that pro- of information cannot be sensed at all, however,
duced them (chapter 2). Feedback about kine- such as relative timing differences in two joints
matics involves various measures derived from or subtle changes in velocity; and kinematic feed-
movement such as position, time, velocity, and back can allow the learner to become aware of
patterns of coordination. When coaches or teach- these features. Also, feedback information about
ers give information about movement patterning subtle aspects of the movement's goal has been
(e.g., "You bent your elbow that time"), they are shown to be useful; Phillips and Berkhout (1976)
really providing a (loosely measured) form of had subjects learn gearshifting and acceleration in
kinematic information, a form of KP. Expert music a simulation of heavy truck driving, and showed
or dance instructors and sport coaches seem to be that computer-aided feedback about smoothness
able to sense "what went wrong" and to provide of acceleration produced marked gains measured
verbal descriptors that can serve as suggestions later on a no-feedback transfer test.
for change in the movement. Of course, many dif- But how effective is kinematic KP when com-
ferent features of the movement can be described pared with other types of augmented feedback?
and used for feedback, and a major issue has Several studies have been conducted on this
been the discovery of what kinds of kinematic issue, and the findings reveal some interesting
information would be most useful for learning principles. Most of this research suggests quite
and performance (e.g., Swinnen, 1996; Newell, clearly that the effectiveness of kinematic feed-
1991; Newell & Walter, 1981). back depends on the nature of the task goal. For
Early studies of kinematic information example, subjects were asked to draw geometric
feedback were done by Lindahl (1945; see also shapes on a tabletop in two experiments reported
Tiffin & Rogers, 1943), who analyzed patterns by Newell, Carlton, and Antoniou (1990). The task
of foot-pedal actions in skilled industrial work- goal (a circle) was known in the first experiment,
ers operating a cutting machine (see "Lindahl's but in the second experiment the task goal was
Study Using Movement Kinematics" on p. 39). an unknown, irregular shape. The subjects were
Lindahl determined the most effective pattern given one of three types of feedback: (1) KR about
of foot motion from measurements of highly the error between their movement and the goal;
 Augmented Feedback Iii 403

(2) a digital image of the pattern plus the KR; or from the motions that produced it. For example,
(3) a digital image of the feedback of the produced many different movements can produce the same
movement superimposed on a template of the task trajectory of a batted ball.
goal, plus KR. Learning (as measured in a reten- How does kinematic feedback about move-
tion test without any augmented feedback) was ments affect the acquisition of skills in which the
not affected by the nature of the feedback when movements are not isomorphic with the task goal?
the task goal was well known to the subjects (the A computer-controlled analog of a baseball bat-
circle in experiment 1). However, when the task ting task was developed by Schmidt and Young
goal was unknown, there was a clear advantage (1991) to examine these issues. The task required
for the group that received the KR plus the aug- subjects to "strike" a moving-light "object" by
mented feedback superimposed on the task goal. passing a movement lever through a coincidence
The benefit of augmented kinematic feedback may point as the light went by. The goal was to maxi-
be optimized when its content specifies informa- mize distance, as defined by a combination of the
tion that cannot otherwise be generated from velocity and timing accuracy at the coincidence
sources such as inherent feedback or from other, point. On the basis of research suggesting that a
less detailed sources of augmented feedback. particular movement pattern produced the best
The role of task goal information and available outcome scores (Schmidt & Young, 1991), Young
sources of feedback may also be related to the and Schmidt (1992) conducted a study to assess
findings reported by Swinnen, Walter, Lee, and what kinematic feedback variables facilitated
Serrien (1993). Subjects in this study practiced a learning when presented in relation to the optimal
discrete, bimanual-coordination task in which movement pattern. Their findings revealed that
the actions of the two limbs were not the same. each kinematic variable manipulated (mean or
The left limb was to produce a unidirectional variability of the reversal point; mean or variabil-
elbow flexion movement. At the same time, a ity of the time of the reversal) tended to facilitate
flexion-extension-flexion movement of the right the acquisition of that kinematic variable in the
elbow was to be performed. Without practice this production of the movement. However, only
coordination task is very difficult to perform, as the kinematic feedback about the mean reversal
there is a tendency to make the same actions with point was more effective than outcome KR in
each arm (see chapter 8). Swinnen and colleagues maximizing performance outcome. The effects
(1993) found that the capability to perform each of KP appear to be enhanced, however, when
distinct limb goal improved little with practice in an optimal movement pattern is not used as a
the absence of augmented feedback. Surprisingly, reference criterion, again suggesting that the
however, learning was facilitated equally well by kinematic information may be most useful when
KR (a simple outcome measure of coordination it promotes active, problem-solving activities in
performance) and by the precise augmented kine- the learner (Brisson & Alain, 1996a, 1996b).
matic feedback profiles of the two limbs. Accord- Similar findings reveal that kinematic feedback
ing to Swinnen and colleagues, the findings facilitates specific motor learning outcomes in
supported the idea that the limb coordination rehabilitation. For example, Cirstea, Ptito, and
information provided by the KR was sufficient Levin (2006) examined three groups of patients
to enable subjects to explore new strategies to with hemiparesis as they practiced an arm-
learn the task. Thus, it seemed that practice-and pointing task without vision over 10 sessions
strategies brought about by information sources and in a one-month retention test. Compared to
that affected practice-combined to determine a control group, the individuals who received
the value of augmented feedback. KR about movement end point steadily learned
In this research, the effectiveness of kinematic and improved aiming precision, but not speed.
feedback was assessed in tasks in which the feed- However, the subjects who received KP about
back was identical to the goal of the movement. elbow and shoulder velocities mainly improved
For example, augmented feedback about a dive these performance outcomes. Thus, augmented
or an ice skating jump could be related directly information may contribute to learning specificity
to the movement, as the quality of the movement effects (see discussion in chapter 11).
represents the task goal. However, in other skills, These specificity effects may help to resolve
the outcome of an action may be quite distinct the curiosity about the effectiveness of kinematic
404 Motor Control and Learning

feedback relative to certain attentional focus the internal group, the external group performed
manipulations. In previous sections of the book more accurately both early and later in practice.
we discussed findings regarding the impact on The differences were small at the start of the
performance (chapter 4) and learning (chapter retention test, but the external group continued
11) of instructions to focus one's attention. In to improve their performance in the retention
most cases, research has shown that instructions test compared to the internal group. These find-
to attend to an "external" source of information ings have been replicated in a tennis serve task
were beneficial for performance and learning, (Wulf, Mcconnel, Gartner, & Schwarz, 2002) and
compared to instructions to focus on an "internal" in stroke rehabilitation (Cirstea & Levin, 2007),
source of information (Wulf, 2007). In the context although further work is necessary to dissociate
of the present discussion, it might be correct to the specific effects of attention-focusing instruc-
say that KP directs the learner to focus attention tions and augmented feedback. Nevertheless,
on an internal source (e.g., the motions of a limb), these findings are important in that they provide
whereas KR directs the learner's attention to the the beginnings of a better understanding of both
impact of the movement on the environment (i.e., the potentially positive and negative influences of
externally). From this perspective, even though kinematic feedback on performance and learning.
KP might provide more information than KR, one
might anticipate that KR holds an advantage over Biofeedback
KP in terms of directing the learner's attention to
a more appropriate focus. Going a step further, feedback can be given about
As an example, subjects in a study by Shea features of the movement that are not perceived
and Wulf (1999) practiced the stabilometer and directly-a key feature of biofeedback training. If
received concurrent feedback about the position a particular biological process (e.g., blood pres-
of the platform relative to the horizontal. Some sure) is measured electronically and used as
subjects were told that the feedback represented feedback, then subjects can learn to voluntarily
a line on the platform (external group) while control these (normally unconscious) processes
others were told that this feedback represented (see Richter-Heinrich & Miller, 1982, for a review).
their feet (internal feedback). The results over two Years ago, Basmajian (1963) gave subjects visual
days of practice and a no-feedback retention test and auditory feedback of their own electromyo-
(on day 3) are shown in figure 12.4. Compared to grams (EMGs) and showed how such information

Day 1 Day2 Day3

10

U)
Q)
Q)
8
....
Cl
Q)
~
..... 6
e....
Q)
en
~ 4
er:

2 3 4 5 6 7 2 3 4 5 6 7 2 3 4 5 6 7
Trials (90 s)

FIGURE 12.4 Effects of internal- versus external-focus feedback in acquisition (days 1 and 2) and in retention (day 3).
Reprinted from Human Movement Science, Vol. 18, C.H. Shea and G. Wulf, Enhancing motor learning through external-locus instruction and feedback, pages 553-571,
Copyright 1999, with kind permission from Elsevier.
 Augmented Feedback 405

could allow the subject to learn to control a single form of kinetic feedback. Subjects could use this
motor unit, something that is not normally under information to optimize the form of the force-time
voluntary control. This general idea has been curve (i.e., to produce a maximum impulse).
tried (with only moderate success) in teaching Newell and Walter (1981) and Newell, Sparrow,
subjects who are deaf to speak, with the subjects' and Quinn (1985) have provided similar examples
sounds being transformed into visual informa- with other tasks. The effects of this extra informa-
tion presented on a television screen (Nickerson, tion are relatively permanent too, as they persist
Kalikow, & Stevens, 1976). Mulder and Hulstijn in a short-term no-feedback retention test (Newell
(1985) showed that feedback information about et al., 1985) as well as in tests that occur after a
the EMG from the muscles controlling the big long delay interval (Broker, Gregor, & Schmidt,
toe contributed to learning toe movements, and 1993; van Dijk, Mulder, & Hermens, 2007).
that the gains remained even after the feedback There is good reason to remain cautious about
was removed. the benefits of using kinetic feedback for the
Brenner (1974) and Lang (1974) argued that attentional focus reasons mentioned in the previ-
there is a close relationship between these bio- ous sections. If the provision of kinetic feedback
feedback procedures for training unconscious encourages the learner to adopt an internal focus
processes on the one hand and kinematic feedback of attention in performance, then the potential
for motor learning on the other. If such a relation- benefit of this rich form of augmented feedback
ship exists, however, then it is possible that bio- might be overshadowed by the consequences of
feedback would be expected to have some adverse the ineffective attentional focus.
effect in motor learning as well (Yiu, Verdolini,
& Chow, 2005). As we will see later, augmented
feedback that is provided instantaneously with Knowledge of Results
the completion of performance is beneficial for
performance but can degrade learning. Clinical We now turn our attention to the vast amount of
treatment of speech disorders, for example, is one research on KR-augmented information about
research area in which continuous, instantaneous the movement outcome. Experiments in this
feedback may have an adverse effect on rehabilita- research area have frequently used very simple
tion (for a review see Maas et al., 2008). tasks, such as blindfolded limb-positioning tasks
and timing tasks. The reason is that with these
Kinetic Feedback kinds of tasks, very little if any learning at all
Whereas kinematic measures are variables can occur in the absence of KR. In this way, the
describing pure motion, kinetic measures are relative effectiveness of various manipulations of
descriptors of the forces that produce the kinematic KR can be examined in terms of their impact on
variables. We have long recognized that muscular the learning process.
forces and the durations over which they act are
fundamental outputs of the central structures
Precision of KR
thought to organize movements; the impulse- The precision of KR refers to the degree of exact-
timing theory discussed in chapters 6 and 7 is one ness of the information provided to the learner.
statement of that basic view. As a result, research- For example, if the subject's goal was to make a 10
ers have often thought that feedback in terms of cmmovementand theactualmovementwas 10.13
kinetics would be a "natural" kind of information cm, KR could be provided in a variety of ways.
for the motor system to use for learning. At the most general or qualitative level, the sub-
Some early work supports this view. English ject could be told that the movement was either
(1942) utilized force feedback from a trigger "right" or "wrong." However, differing degrees
squeeze to facilitate riflery training. Howell (1956) of precision could be substituted for these general
had subjects learn a runner's sprint start and feedback statements of "right" or "wrong."
recorded forces applied against a strain gauge (a In the case of "wrong," one could give more
force sensor) that was attached to the foot plate precise KR by saying "long" or "short," meaning
in the starting blocks. The forces recorded during that the person moved beyond or short of the
the time of the action provided a force-time curve, target. One could give still more precise KR by
which was shown to subjects after each trial as a saying "wrong by l," meaning 1 mm off target.
406 Motor Control and Learning

Or, one could say "long 0.1," meaning that the these forms of KR have information about both
movement was 0.1 mm too long, or "long 0.13," factors (e.g., "long 13"). Generally, the evidence
meaning that it was 0.13 mm too long. The KR suggests that there is some benefit to providing
could be even more precise than this, measuring information about magnitude of error, but this
movement accuracy to finer quantitative degrees information is far more useful if the direction is
(e.g., in nanometers). also specified. Knowing that an error was made
In the case of" right," the experimenter would in a particular direction gives a strong indication
need to define, exactly, what movement outcome of the ways in which the movement must be
would satisfy the criterion that distinguished modified next time but information only about
"right" from "wrong." In the early work of Trow- magnitude does not.
bridge and Cason (1932), for example, lines were Another key issue is related to the precision
considered to be correct if drawn within 1/8 in. of the KR, and the classic study in this area was
of the 3 in. goal. In such a case, the "correctness" conducted by Trowbridge and Cason (1932). Four
of the movement is defined relative to a "band- groups practiced drawing 3 in. lines for 100 trials.
width," defined as the degree of acceptable error One group never received KR. Another group
tolerance around the goal. Various combinations received nonsense syllables after drawing each
of qualitative and quantitative forms of KR line (a control condition). A third group received
have been examined, and these manipulations qualitative KR in the form of "right" (if the line
have rather large effects on performance and was within ±1/8 in. of the goal) or "wrong" from
learning. the experimenter. The last group was given pre-
cise, directional KR (longer or shorter) in terms
Qualitative Versus Quantitative KR of the exact deviation, in eighths of an inch, from
The most basic question concerning the precision the goal length. During both acquisition and a
of KR is the kind of information that is presented. no-KR transfer test that followed immediately
Information about the direction of the error is after practice, accuracy was greater for the precise
presented in some, but not all, forms of KR. Infor- and the right-wrong KR groups than for either
mation can also be provided about the magnitude the nonsense-KR or the no-KR group (figure 12.5).
of the error, irrespective of direction. Some of Furthermore, precise KR was more accurate than

3.5

3.0

2.5
E'
~
....
e 2.0
Gi
Q)
..-
:l 1.5
0ti)
..c
<(
1.0

0.5

2 3 4 5 6 7 8 9 10 Transfer
Blocks of 10 trials (100 trials)

FIGURE 12.5 Qualitative and quantitative knowledge of results (KR) effects in acquisition and transfer. The No-KR group
did not perform the transfer rest.
Data from Trowbridge and Cason 1932.
 Augmented Feedback l!I 407

the right/wrong type of KR. These effects have the nature of augmented feedback is determined
been replicated often (Bennett & Simmons, 1984; by a bandwidth about the movement goal. In
Magill & Wood, 1986; Reeve, Dornier, & Weeks, most studies using this method, qualitative KR
1990; Salmoni, Ross, Dill, & Zoeller, 1983), sup- in the form of "correct" or "right" is provided to
porting the conclusion that precise, quantitative the subject when the performance outcome lies
KR is generally more effective for learning than within the boundaries of correctness as defined
qualitative KR. by the bandwidth (similar to the KR provided
These techniques do not permit the separation by Trowbridge & Cason [1932]). However, when
of information about precision of KR from infor- performance exceeds the bandwidth, the experi-
mation about the direction of reported error. Stud- menter provides the learner with specific KR that
ies conducted since the Trowbridge and Cason gives both the magnitude and the direction of
experiment have separated these effects and have error. This method is probably what many teach-
generally shown that the more precise the KR, ers and therapists do spontaneously-correcting
the more accurate the performance, up to a point, relatively poor performance and rewarding rela-
beyond which no further increases in accuracy are tively good performance.
found as KR is made more precise (for reviews, Bandwidth KR has rather substantial effects
see Newell, 1981; Salmoni et al., 1984). Subjects on performance and learning. In fact, the
presumably know that they cannot be responsible research suggests that learning is facilitated as
for errors smaller than a certain size (e.g., 1 mm), the bandwidth becomes larger. There is prob-
as the movement control mechanisms themselves ably an optimal bandwidth size, although more
are more variable than this. Therefore, it is likely research needs to be done to establish what this
that subjects "round off" very precise KR to a might be. Sherwood (1988) conducted one of the
more meaningful level of precision. first studies in this area (see also Annett, 1959).
Subjects were to learn to achieve a rapid elbow
Bandwidth KR flexion movement time as close to 200 ms as
An alternative to giving either qualitative or possible. Subjects in a control (no bandwidth)
quantitative KR is provided by the bandwidth group (0% BW in figure 12.6) were told their
KR method (Sherwood, 1988). With this method exact movement time after each trial. In two other

22

20

UI 18
g
...0
......
Q) 16
Q)
2'i
Ill
·;:
Ill
> 14

12

2 3 4 Retention
Blocks of 25 trials

FIGURE 12.6 Bandwidth knowledge-of-results (KR) effects in acquisition and retention.

Reproduced and adapted from Table 1 and 2 with permission of author and publisher from: Sherwood, D.E. Effect of bandwidth knowledge of results on movement consistency.
Perceptual and Motor Skills, 1988, 66, 535-542. ©Perceptual and Motor Skills 1988.
408 Motor Control and Learning

conditions, subjects were given movement-time ing the size of the bandwidth over the course of
KR only if their outcomes exceeded a tolerance practice, keeping the proportions of error and cor-
limit around the MT goal (±5% or ±10%). Perfor- rect KR relatively constant, have been ineffective
mance inside the bandwidth received no explicit (Goodwin & Meeuwsen, 1995; Lai & Shea, 1999).
KR-which subjects had been instructed to inter- As suggested earlier, these effects make consid-
pret as meaning that their MT had been correct. erable sense and have been replicated in experi-
Although these bandwidth conditions had no ments in which a golf chipping task was learned
differential effects on acquisition performance, (Smith, Taylor, & Withers, 1997). The essence is
as can be seen in figure 12.6, the no-KR retention that, when assisting people in learning a new
test performance was positively related to the size skill, you might provide help when they are
of the bandwidth. doing something wrong, but not when they are
Sherwood's experiment uncovered a number correct (in other words, "If it ain't broke, don't fix
of important issues regarding KR and the learn- it"). The key seems to be in deciding when is the
ing process. For example, one consequence of the best time to intervene and provide augmented
bandwidth KR procedure is that as the tolerance feedback. If an optimal bandwidth exists for each
limits are increased (5% to 10%), the proportion of person, its size would likely depend on a number
trials supplied with error KR diminishes. As will of factors that may change with practice and task
be seen later in this chapter, less frequent error demands (Lee & Maraj, 1994).
KR in acquisition also improves learning. So, one
question is whether or not the bandwidth effect is Erroneous KR
more than just a reduced KR frequency effect. To Imagine a situation in which the provider of
examine this question, Lee and Carnahan (1990a) augmented feedback is inaccurate in giving the
used bandwidth groups of 5% and 10% together feedback. For example, in older bowling alleys,
with yoked control groups; the control groups an illuminated indicator at the end of the lane
received KR on the same trials as their yoked provided KR in terms of how many pins were
counterparts in the bandwidth groups. However, left standing after the first ball was bowled. Since
the key difference was that a bandwidth subject there are times when one pin is hidden from the
interpreted no KR to be feedback that the previ- bowler's view by another pin, the augmented
ous trial performance had been "correct." For feedback from the pin indicator can reveal infor-
the yoked controls, the absence of KR revealed mation that the bowler cannot directly see. But,
nothing about the previous trial. Lee and Carna- if one or more of the indicator lights happened
han found that the bandwidth groups performed to malfunction, the bowler could get an incorrect
more effectively in retention than did their respec- impression of the number of pins that were still
tive control groups, suggesting that the provision standing. That is the issue-what is the impact
of the "correct" KR gave an additional boost to of KR when it is erroneous?
learning beyond that normally associated with Buekers and Magill (1995) conducted studies
less frequent KR. Similar results were reported by on the effects of erroneous KR in an anticipation-
Butler, Reeve, and Fischman (1996) and Wright, timing task-a task for which inherent (visual)
Smith-Munyon, and Sidaway (1997). Moreover, feedback is normally sufficient for learning to
bandwidth KR facilitated learning more than occur (Magill et al., 1991). Subjects in these stud-
a yoked relative-frequency control group in an ies were sometimes provided with incorrect aug-
observational learning paradigm, supporting a mented feedback, indicating that the accuracy of
high cognitive function to the provision of "cor- the previous anticipation had been 100 ms later
rect'' feedback (Badets & Blandin, 2005). than actually had been the case (e.g., someone
Another potentially strong learning effect that who had been 65 ms early in anticipating the
could have been going on in the Sherwood (1988) arrival of the stimulus would be told that she had
study is that the distributions of error KR and been 35 ms late). The consequence of this erroneous
correct KR change as skill improves-the propor- feedback is a motor behavior whereby the subjects
tion of trials followed by error KR is reduced and perform the task with a constant error (CE) of up
the proportion of "correct KR" trials is increased. to -100 ms. These effects are relatively long last-
This seems to be an important component of ing, with large negative CEs occurring after one
bandwidth KR effectiveness, as methods of reduc- week in a no-KR retention test (Buekers, Magill,
 Augmented Feedback I!! 409

& Hall, 1992; Vanvenckenray, Buekers, Mendes, which KR is provided. It is the number of times
& Helsen, 1999) and in transfer tests to novel KR is given divided by the total number of trials,
stimulus speeds (McNevin, Magill, & Buekers, multiplied by 100 for conversion to a percentage.
1994). The findings have been replicated in other In this example, the relative frequency of KR is
types of timing tasks (Ryan & Fritz, 2007; Ryan (40 I 80) x 100 = 50%.
& Robey, 2002), as well as in a soccer ball kicking Which of these two KR-scheduling variables
task in which visual feedback provided by a video is the more critical for learning? Bilodeau and
was erroneous (Ford, Hodges, & Williams, 2007). Bilodeau (1958) were the first to investigate this
These erroneous-KR effects indicate that the question, using a task in which subjects turned a
accuracy of augmented feedback has very power- knob to a target position in the absence of vision.
ful effects on performance and learning, whereby For the four different groups, KR was provided
subjects negate or discount the accuracy of their after (a) every trial, (b) every third trial, (c) every
own error-detection capabilities in favor of trust- fourth trial, or (d) every 10th trial, producing
ing the validity of the (erroneous) augmented relative frequencies of KR of 100%, 33%, 25%,
feedback. The impact of erroneous KR appears and 10%, respectively. The number of trials per-
to be the strongest when it is presented on every formed by these groups, however, was adjusted
practice trial during acquisition. Studies in which so that all groups were presented KR after 10
trials with erroneous KR are alternated with trials trials; therefore, the group with 100% relative
providing correct KR (Buekers, Magill, & Sneyers, frequency received 10 trials, the group with 33%
1994), and those in which trials with erroneous KR relative frequency received 30 trials, and so on.
follow a practice period with correct KR (Buekers Thus, the experiment involved groups that had
& Magill, 1995), show diminished performance different relative frequencies, but constant abso-
effects and no learning effect of erroneous KR. lute frequencies (10) of KR.
Thus, periodic KR that is counterintuitive to inher- The results for each of the four groups are pre-
ent feedback may not be as disruptive to learning sented in figure 12.7. Only the trials immediately
as the situation in which the learner is consistently
following the presentation of KR are plotted. This
is, of course, every trial for the group with 100%
faced with conflicting augmented information.
relative frequency of KR, only one-third of the
Schedules of KR trials for the group with 33% relative frequency,
and so on. The amount of error on each trial, as
We saw in the previous section about bandwidth well as the pattern of change of the errors as trials
KR that determining when to give KR and what progressed, was nearly the same for the four
type of KR to give can have a large impact on groups. Even though the groups differed greatly
performance and learning. These effects relate in terms of the relative frequency of KR, when
closely to a class of KR-scheduling variables over the absolute frequency was equated, no differ-
which the experimenter has specific control. As ence in performance was found between groups.
we will see, these variables also have profound For performance, the critical feature of KR in this
learning and performance effects. experiment was the number of times that KR was
given; the relative proportion of trials followed
Relative- and Absolute-Frequency of KR by KR appeared not to be an important variable.
If error information is required for learning, we Another way to think of this is that the no-KR
might reasonably expect that more KR will result trials were meaningless, neither contributing
in stronger learning. We can distinguish between to nor detracting from performance of the task.
two measures of the "amount" of KR that is Motor learning researchers initially took the equal
provided: absolute frequency and relative frequency performances of the various groups in figure 12.7
of KR. Absolute frequency of KR refers to the to mean that absolute frequency is important for
number of KR presentations received over the learning and that relative frequency is irrelevant.
course of practice. If 80 practice trials are given, But notice that the Bilodeau and Bilodeau study
and the person receives KR after every other trial did not use a transfer design to separate the per-
for a total of 40 presentations, then the absolute formance effects of relative frequency from the
frequency of KR is 40. On the other hand, relative learning effects. Hence, we actually have no way
frequency of KR refers to the percentage of trials on of knowing whether varying relative frequency
410 11 Motor Control and Learning

12

'ii) 9
CD
...Ol
CD

CD
'.!:!.
...0
......
Q)
6
.!
:I
0
UI
..Q
<I:
3

2 3 4 5 6 7 8 9 10
Post-KR trial

FIGURE 12.7 Absolute errors in positioning for trials immediately following knowledge of results (KR). (Group numbers
indicate the percentage relative frequency of KR.)
Reprinted from E.A. Bilodeau and l.M. Bilodeau, 1958, "Variable frequency knowledge of results and the learning of simple skill;' Journal of Experimental Psychology 55: 379.

affected learning. More recently, experimenters being irrelevant for learning, reduced relative-
have included these transfer tests, and the effects frequency effects may be beneficial to learning!
on learning have been mixed. Some studies This general result has surprised many because
showed that reduced relative frequencies of KR it says that the no-KR trials, instead of being
produced learning effects that were as large as those meaningless for learning as they appeared to
in 100% KR conditions (e.g., Lee, White, & Car- be in the Bilodeau and Bilodeau (1958) study,
nahan, 1990, experiment 2; Sparrow & Summers, contributed to the learning in some way. This
1992, experiment 1; Winstein & Schmidt, 1990, contradicted a long-held suspicion that practice
experiment 1). Yet, using similar tasks and slightly without feedback was useless for learning. Fur-
modified methods, other experiments showed that ther, this contribution was not manifested during
reduced relative-frequency conditions produced practice when the KR was present, but was seen
more learning than 100% KR conditions (e.g., Lee in a delayed retention test. Decreasing relative
et al., 1990; Sidaway et al., 2008; Sparrow & Sum- frequency certainly does not diminish learning
mers, 1992; Sullivan, Kantak, & Burtner, 2008, and may actually facilitate it.
adults; Vander Linden, Cauraugh, & Greene, 1993; But there is one additional concern with these
Weeks & Kordus, 1998; Weeks, Zelaznik, & Beyak, studies. When the relative proportion of trials
1993). Similar effects have also been found when that are followed by KR is reduced, a confound-
the provision of KR is reduced in an observational ing variable arises. Compared to a 100% KR con-
learning paradigm (Badets & Blandin, 2004; dition, if the total number of trials during practice
Badets, Blandin, Wright, & Shea, 2006). is held constant, then reduced relative frequency
An example is provided in figure 12.8 (from of KR also results in reduced absolute frequency of
Winstein & Schmidt, 1990, experiment 2). Notice KR. If the researcher decides to make the absolute
that there are no differences between the 100% frequency the same as in the 100% condition,
and 50% relative-frequency groups in acquisi- then the total number of trials must be increased
tion, as Bilodeau and Bilodeau (1958) had found. for the reduced relative-frequency group. In all
However, in 5 min and 24 h no-KR retention tests, the studies cited here, the total number of trials
a clear learning effect was shown that favored was kept constant. Thus, the effects of reduced
the 50% group. Thus, it seems that instead of relative frequency must be considered in light of
 Augmented Feedback 1111 411

15

14

13

.-. 12
iCi 11
Cll
::!:!.
m10
==
a: 9

Imm
Acquisition (12-trial blocks) Retention (12-trial blocks)

FIGURE 12.8 Effects of 100% versus 50% relative frequency of knowledge of results (KR) in acquisition and retention.
Reprinted, by permission, from C.J. Winstein and R.A. Schmidt, 1990, "Reduced frequency of knowledge of results enhances motor skill learning," Journal of Experimental
Psychology: Learning, Memory, and Cognition 16; 910. Copyright© 1990 by the American Psychological Association.

the fact that fewer KR presentations were given. will lie within the bandwidth so that the provision
When we recall that learning increases with the of error KR will be withheld.
number of KR presentations, perhaps it is not However, one further complication arises
surprising that the effects of relative frequency when we consider the effects of reduced relative
are rather mixed. It may very well be that the frequency as a function of the task that is learned.
positive effect of reducing the relative frequency Experiments have shown that when subjects
has been offset by the negative effect of fewer practice several versions of a generalized motor
KR presentations. This certainly contradicts the program, reduced relative frequency of KR facili-
earlier conclusions that providing more feedback tates the learning of invariances common to the
is all-critical for motor learning. And, note that movement pattern, but not the parameterization
delayed no-KR transfer tests were required in characteristics (Wulf, Lee, & Schmidt, 1994; Wulf
order to show these effects-further supporting & Schmidt, 1989; Wulf, Schmidt, & Deubel, 1993).
the use of such transfer designs in motor learn- A possible explanation for the relative-
ing research. frequency effect in motor learning was sug-
The effects of relative frequency appear to be gested by Salmoni and colleagues (1984; see also
clearer if the method used for reducing the pre- Schmidt, 1991a; Schmidt & Bjork, 1992; Schmidt &
sentations of KR is a "fading" procedure. Here, Shapiro, 1986; Winstein & Schmidt, 1990). When
giving fewer KR presentations (trials constant) KR is given on every trial (relative frequency of
seems to greatly improve learning (Sullivan et al., 100%), this condition is very effective for perfor-
2008; Winstein & Schmidt, 1990; Wulf & Schmidt, mance when KR is present, because of a number
1989). The method usually involves providing of temporary factors already discussed (e.g.,
KR relatively often during the initial stages of guidance, motivational, and energizing prop-
practice and then gradually withholding the pre- erties). However, the subject comes to rely too
sentation of KR more and more toward the end of heavily on this information and fails to process
practice. This method actually has an effect very information necessary for learning the task in a
similar to what naturally happens when using the relatively permanent way; subjects use KR as a
bandwidth KR procedure, because skill improve- "crutch." Subjects in conditions of lower relative
ments increase the likelihood that performance frequency, however, do not have such a strong
412 Motor Control and Learning

performance enhancement from KR and so are in initial practice, when guidance and motivation
"forced" to engage in other processes during the are critical; but then the instructor should system-
acquisition phase. These processes result in the atically decrease relative frequency of KR as the
subjects' learning something fundamentally dif- performer becomes more proficient.
ferent, such as the capability to detect one's own
Trials Delay and Summary KR
errors or to be consistent. Perhaps reducing the
relative frequency also encourages one to make The literature discussed so far has involved situ-
between-task comparisons, which might facilitate ations in which KR for a given trial is presented
the abstraction of common movement attributes before the next trial (i.e., KRn occurs before trialM 1
(Shea & Zimny, 1983; Wulf et al., 1994). This learn- in figure 12.1). However, what happens if the KR
ing is not revealed during the acquisition phase from a given trial occurs after the performance
because every-trial KR dominates performance, of the next few trials? Such a procedure, at first
but it does contribute to performance on delayed glance, would appear to be extremely disrupt-
no-KR transfer tests. According to this hypoth- ing for performance; it would be difficult for the
esis, "too much" KR in acquisition is detrimental learner to know which KR to associate with which
if the goal is to produce the movement without movement, particularly when KR is increasingly
KR later, as it usually is. As we will see, this separated from the trial to which it refers. We can
hypothesis can explain a number of seemingly probably think of practical situations in which
contradictory findings in the KR literature and this effect might occur-for example, when a
has been supported by some recent experiments learner performs a number of trials in a series,
to be discussed in sections that follow (e.g., Gua- after which the instructor or therapist gives infor-
dagnoli & Kohl, 2001). mation about each trial or maybe about just one
A few practical implications are possible. First, of the trials in the series. In such situations, the
KR is certainly important for learning, as the first trial in the sequence is separated from its KR
results generally say that increasing the amount by the intervening trials.
of feedback, other things being equal, is ben- This method of giving KR was given the term
eficial to performance and learning. But KR can trials delay by Bilodeau (1956, 1966, 1969). In con-
be given too often; in these cases learners come trast to what occurs in the usual KR paradigm,
to rely too heavily on its motivating or guiding we see in figure 12.9 that one or more trials is
properties. This enhances performance during interpolated between a given movement and its
practice in which KR is present, but it is probably KR. In figure l2.9a, M 1 and KR1 are separated by
detrimental to learning as measured on a delayed M 2-there is a one-trial delay between a given
test in which the learner must perform without movement and its KR. In figure l2.9b there is a
KR. Also, relative frequency of KR should be large two-trial delay, with two trials separating a given

FIGURE 12.9 The trials-delay technique, showing a trials delay of (a) one and (b) two. (A given movement and its knowl-
edge of results [KR] are separated by other trials of the same task.) M 1 refers to movement trial 1. KR 1 refers to the aug-
mented feedback provided about results of movement trial 1.
 Augmented Feedback 1111 413

movement and its KR. You can probably think approach to the final performance level and in
of many different variations of this type of KR the level of final performance. These findings dif-
paradigm. fered somewhat from earlier ones by Lorge and
Bilodeau (1956) investigated the effects of trials Thorndike (1935), who had found that improve-
delay using a lever-positioning task with blind- ment in performance did not occur at all under
folded subjects. In two experiments, she varied the trials-delay method. But there can be little
the number of trials by which KR was delayed. argument that trials delay is a variable that has
In experiment 1, Bilodeau used zero-, one-, two-, drastic negative effects on performance. In the
and three-trials delay; in experiment 2, she used earlier literature (e.g., Bilodeau, 1966), the inter-
zero-, two-, and five-trials delay. Subjects were pretation of these trials-delay effects was in terms
fully informed about this technique and were of learning, but these experiments did not use
questioned to make certain that they understood transfer designs to separate the temporary and
how KR was being administered. relatively permanent effects. However, Lavery
The data from the two experiments are shown (1962; Lavery & Suddon, 1962) and others (e.g.,
in figure 12.10, where absolute error in position- Anderson, Magill, & Sekiya, 1994, 2001; Ander-
ing (for trials following KR) is plotted against son, Magill, Sekiya, & Ryan, 2005) have used
trials for the various trials-delay conditions. transfer designs in the study of this variable (and
For both experiments, performance accuracy modifications of it), and their surprising results
systematically decreased as the trials delay was have had important influences on our thinking
increased. This can be seen both in the "rate" of about how KR operates.

10

Ui' 8
..
G)
G)
Cl

....
G)
:!:!. 7
0
G)
6
....G):::J
0
Ill
.c 5
Ill
c
Ill
G)
:ii: 4

2 4 6 8 2 4 6 8 10 12 14
1~ Exp. I -----1 1- - - - - - - - - Exp. II _ _ _ _ _ _ ___,

Blocks of two trials

FIGURE 12.10 Absolute error in positioning as a function of the amount of trials delay in two experiments. (The group label indicates
the number of trials separating a movement and its knowledge of results [KR].)
Reprinted, by permission of the University of Illinois Press, from J.M. Bilodeau, 1956, 'i>.ccuracy of a sample positioning response with variation in the number of trials by which knowledge of
results is delayed;' American Journal of Psychology, 69, 436. Copyright 1956 by the Board of Trustees of the University of Illinois.
414 Motor Control and Learning

Lavery (1962) used several tasks in which a ball of correct trials was far smaller for the Summary
was propelled up a track to a target. Three methods group than for the two groups with KR after each
were used to give KR. One was the usual condi- trial (i.e., Immediate and Both). The addition of the
tion in which KR is given after every trial, called summary information to Immediate to create Both
"Immediate." A second method was "Summary," did not improve performance very much relative
in which the performance on every trial in a 20-trial to providing the usual postmovement KR (Imme-
sequence was shown, but only after the 20th trial diate), so it is clear that the major determinant of
had been completed; no KR was given after each performance was the immediate KR. But we knew
trial as in Immediate. This summary technique was this before, as this pattern of results is similar to
more or less the same as the trials-delay technique, the pattern in the study by Bilodeau (1956) in that
as the KR for trial 1 was separated from its trial performance in acquisition (while KR was pres-
by the other 19 movements in the block, trial 2 by ent) was hindered by the trials-delay technique.
the next 18, and so on. Finally, the third condition Now consider the measure of relative amount
involved both the immediate postmovement KR learned in this experiment-the performance on
and the summary, labeled "Both." After an initial the transfer trials on days 7, 8, 9, 10, 37, and 93
no-KR practice day, five days of practice were for which no KR was provided at any time. The
given under these conditions. group that was formerly least accurate (i.e., Sum-
Performance on all the tasks averaged together mary) was now the most accurate, and the other
is shown in figure 12.11. In acquisition, the number two groups, which had been the most accurate

80

70

60

-
( .)

~ 50
0
(.)

'EGI
~ 40
GI
Q.

30

20

10

No KR - - - - - - - KR-------- - - - - - - - - No KR - - - - - -

- - - - - - Acquisition --------Transfer _ _ _ ___,,_

Days (20 trials/day)

FIGURE 12.11 Percentage correct trials for various summary knowledge-of-results (KR) conditions. (Immediate had KR
after every trial; Summary had KR about every trial presented after each block of 20 trials; and Both had both forms of KR.)
Reprinted. by permission. from J.J. Lavery, 1962, "Retention of simple motor skills as a function of type of knowledge of results;' Canadian Journal of Psychology 16: 305.
Copyright© 1962. Canadian Psychological Association.
 Augmented Feedback 111 415

(i.e., Immediate and Both), were now the least Optimizing Summary Length
accurate. Furthermore, the latter two groups It would seem that summary KR could easily be
appeared to have lost accuracy with each suc- overdone, with summaries of so many trials that
cessive no-KR day, while the Summary group the guidance properties of KR would be minimal.
did not. The effects persisted to day 37 but were Such thinking leads to the idea that there could
essentially gone by day 93. be an optimal number of trials to be summarized,
Which group learned the most? Using the and that this optimum might also vary with task
performance on the transfer I retention test as the complexity in some way. In an experiment by
measure of relative amount learned, as described Schmidt and colleagues (1989), summary KR
earlier, we are forced to conclude that the Sum- was provided as a graph of performance against
mary (trials delay) condition was more effective trials and was given either after each trial (an
for learning than either the Immediate or the Both immediate-KR procedure) or after 5, 10, or 15
condition. Notice that this is yet another example trials. In a relatively simple movement-timing
showing that the most effective condition for task, increased summary length systematically
performance in acquisition was the least effective degraded performance in the acquisition phase
for learning! The basic experiment was repeated when KR was present, as Lavery had found ear-
by Lavery and Suddon (1962), but with the same lier. But surprisingly, in a delayed no-KR transfer
trials-delay methods as used by Bilodeau (1956), test, the most accurate performance was achieved
and the results were nearly the same as the find- by the group that had (in acquisition) received the
ings shown in figure 12.11. 15-trial summaries, with systematically increas-
At first glance, we might be drawn to the ing error as the acquisition summary length
interpretation that the summary KR per se was decreased. The effect appeared to be related to
in some way effective for learning, providing long-term retention, with systematically poorer
a benefit over and above the normally useful retention as the summary length decreased. The
immediate-KR condition. But look again. If sum- longest summaries produce the most learning;
mary KR was "good" for learning, then we should no clear optimal summary length was evident.
expect the Both group (which also had summary Similar findings were also reported by Gable,
KR) to have benefited in a similar way. To the Shea, and Wright (1991), with subjects in a 16-trial
contrary, though, we see that the Both group condition performing most effectively and no
performed almost identically to the Immediate evidence for an optimal summary size.
group, both in the acquisition phase and in the In another investigation using a more complex,
no-KR transfer phase. One view is that, when KR anticipation-timing task with KP provided rather
was added to the normally effective summary- than KR, summaries given after either 1, 5, 10, or
KR procedure to form the Both group, it lowered 15 trials (Schmidt, Lange, & Young, 1990) were
the level of learning to that of the Immediate used as in the study just described. Figure 12.12
group. In our interpretation (see Salmoni et al., shows the performance in acquisition and on 10
1984; Schmidt, 1991a; Schmidt, Young, Swinnen, min and two-day delayed no-KP transfer tests.
& Shapiro, 1989), it was not that summary KR Again, increasing the summary length degraded
was necessarily responsible for the beneficial the performance in the acquisition phase, with
effect seen in learning, but that immediate KR systematically lower scores as the summary length
was detrimental to learning! This interpretation increased. But in the no-KP transfer tests, the most
is in keeping with the guidance hypothesis that effective summary length for learning was five
immediate KR provides "too much" informa- trials; shorter (one trial) and longer (10 and 15
tion for learners, causing them to rely on it too trials) summaries showed less effective learning.
heavily; thus the subject is not forced to learn the A similar set of results was also reported by Yao,
information-processing activities critical for per- Fischman, and Wang (1994); acquisition perfor-
formance when KR is removed in the transfer test. mance was least effective for conditions with the
Summary KR provides much less guidance, and longest summary lengths (using summaries of 1, 5,
presumably forces the subject to learn the task in and 15 trials). In the no-KR retention test, however,
a somewhat different way, perhaps by prompting the five-trial summary condition was superior to
the learner to gather information through alter- both the every-trial and 15-trial summary condi-
native feedback sources (Anderson et al., 2005). tions (see figure 12.13 on p. 417).
416 Motor Control and Learning

300

200

~
0
(.)
(J)

100

2 3 4 5 6 Immediate Delayed
Blocks of 15 trials Blocks of 30 trials

FIGURE 12.12 Performance score for various numbers of trials included in a summary-feedback presentation for acquisi-
tion (left) and immediate and delayed retention (right).
Reprinted from Human Movement Science, Vol. 9, R.A. Schmidt, C. Lange, and D.E. Young, "Optimizing summary knowledge of results for skill learning," pg. 334, Copyright
1990, with kind permission of Elsevier.

It seems clear from these studies that if optimal because the numeric information overloads the
summary lengths do exist, these are likely to be processing capabilities of the learner. However,
task specific, perhaps in relation to the task's there is an interesting variant of the summary
complexity. Such conclusions are supported in a procedure that has been called average KR. Here,
clever experiment by Guadagnoli, Dornier, and instead of providing KR about a block of trials in
Tandy (1996). In this study, Guadagnoli and col- the summary, the average of the block of trials is
leagues had subjects learn simple and complex determined and this mean score is provided as
versions of a force production task. For the simple KR. In this way, the average represents a statis-
task, the largest (15 trials) summary condition tical summary of the block of trials rather than
produced the most learning; however, for the a graphical summary. In the study by Yao and
complex task, the smallest (one trial) summary colleagues (1994) discussed in the previous sec-
group was optimal. These findings provide sup- tion, two additional groups of subjects received
port for Schmidt and colleagues' (1990) sugges- summary KR that was provided as a statistical
tion that optimal summary-KR sizes are depen- average of either 5 or 15 trials. The results for a
dent on the amount of information provided in temporal measure of performance are presented
the summary, which is determined largely by the in figure 12.13 (the findings for a spatial measure
complexity of the task. were similar).
As described in the previous section, acquisition
Statistical Summaries of KR performance was related inversely to the summary
In many summary-KR experiments, performance size, and no-KR retention performance was most
on a series of trials is presented to the subject in accurate for the five-trial summary group and least
the form of a graph that organizes the augmented effective for the every-trial group. Of particular
feedback about all of the trials in summary fash- interest, however, was that the groups receiving
ion. When multiple KR presentations need to be average summaries performed similarly to the
given, the information is more readily understood groups that received graphical summaries. This
when given graphically than when given numeri- was consistent for both acquisition and retention
cally (Cauraugh, Chen, & Singer, 1993), perhaps and for both the five-trial and 15-trial summary
 Augmented Feedback 417

Now, suppose that only one source of feedback

were used. On what basis is this one source to be
100 chosen? Is it the one that has the most important
90 impact on performance, the one that is most
U) 80
important for a safety concern, or the one that
E meets some other criterion? Moreover, if aug-
';:" 70
e~ 60 mented feedback is provided relatively often,
Cl
can it be about the same information source or
·=E 50 different sources? These ideas have not been
:;;::; 40 addressed frequently in research, although some
s
:I interesting findings about scheduling have been
0 30
reported using KR as augmented feedback (Lee
~"' 20
& Carnahan, 1990b; Swanson & Lee, 1992).
10 Subjects performed a three-segment timing
task in the Lee and Carnahan (1990b) study,
10 min 2 days
Retention delay
with a specific timing goal for each segment. All
subjects were given KR about one segment after
each trial. The question was whether KR should
FIGURE 12.13 Absolute timing errors for various sum-
be presented repeatedly on the same segment for a
mary- and average-KR (knowledge of results) conditions.
series of consecutive trials (blocked-KR schedule)
or whether KR should be given about a different
segment on each successive trial (random-KR
conditions. These data suggest that the learning schedule). 2 The results were rather surprising:
and performance effects of summary KR may The random-KR schedule was more effective
be similar regardless of whether the summary is for both performance and learning of the task.
presented as a graph or as a statistical average (see In acquisition, KR was beneficial when it was
also Weeks & Sherwood, 1994; Young & Schmidt, provided for a given segment, but performance
1992). The similarity of effects of graphical and sta- deteriorated once KR was withdrawn from that
tistical forms of summary KR is also explained well segment (see also Swanson & Lee, 1992). Blocked
by the guidance properties of KR, as the two meth- KR focused learners only on the segment about
ods work in similar ways to reduce the guiding which they were currently receiving KR, whereas
properties of augmented feedback. But attempts to random KR encouraged subjects to process infor-
further tease apart the specific impact of KR sum- mation about all three segments on each trial.
maries have had mixed success (Guay, Salmoni, These results suggest another way in which KR
& Lajoie, 1999; Guay, Salmoni, & Mcllwain, 1992; can have an overly directive or guiding function.
Sidaway, Moore, & Schoenfelder-Zohdi, 1991; In terms of the guidance hypothesis, blocked KR
Wright, Snowden, & Willoughby, 1990). may have been directing the subject's attention to
the one segment on which KR was being deliv-
Blocked Versus Random KR
ered, and treating that segment as just one part of
Up to this point, most of the research we have the whole task. When KR was shifted elsewhere,
reviewed has involved augmented feedback about it guided the subject to a different part of the task,
one information source, such as KR for movement again decomposing the task into parts. These
time or end-point accuracy. But consider the sched- conclusions should be considered with caution,
uling implications if there were many sources for however, until more research has been conducted
which feedback could be provided. Suppose, for using different tasks and feedback sources (e.g.,
example, one were to provide KP about the gait Wulf, Horger, & Shea, 1999).
of a stroke patient along with several forms of
KP. Many potential sources of feedback could be Self-Regulated KR
used, but the amount of feedback would likely To this point in the discussion of KR variables,
be overwhelming if all the feedback sources of we have directed our focus toward variables that
information were provided at once. So, therapists are under the direct manipulation of the experi-
intuitively withhold much of this feedback. menter. We now turn our attention to a different
418 Motor Control and Learning

experimental approach to deciding when to pres- The results of Janelle and colleagues' (1997)
ent KR (Huet, Camachon, Fernandez, Jacobs, & study are shown in figure 12.14. The augmented
Montagne, 2009; Janelle, Barba, Frehlich, Tennant, feedback was useful for learning, as all groups
& Cauraugh, 1997; Janelle, Kim, & Singer, 1995; that received KP scored higher than the control
Wulf & Toole, 1999). In this paradigm, subjects group. Of more importance, though, the self-
perform a movement task and are presented with regulated KP schedule produced more effective
the option of receiving augmented feedback or performance in acquisition and retention than did
not. For example, subjects in a study by Janelle the yoked controls and the summary-KP group.
and colleagues (1997) practiced throwing a ball Why might self-regulation have an effect that
at a target with their nondominant limb. Transi- was stronger than a schedule identical in every
tional KP was provided in a manner similar to aspect except for the fact that it was experimenter
that in the Kernodle and Carlton (1992) study imposed? One of the leading hypotheses suggests
discussed earlier. A control group received no that self-regulation allows subjects to tailor the
KP, and another group received a summary-KP delivery of augmented feedback to suit their
statement after every fifth trial. The group of immediate performance needs. Interviews of
most importance here followed a self-regulated subjects who had practiced with a self-regulated
schedule, in which KP was provided only when schedule indicated that they tended to request KR
subjects asked to receive the augmented feed- after trials in which performance was believed to
back. The final group was another control group have been relatively effective, rather than after
that was yoked to the self-regulated group-the ineffective performances (Chiviacowsky & Wulf,
KP delivery schedule for each subject in this 2002). In response to this finding, Chiviacowsky
yoked group was matched to a member of the and Wulf (2005) performed a nice experimental
self-regulated group. In this way, the yoked con- test by comparing self-regulated conditions in
trol subjects received the same number of KPs which subjects made the determination to receive
as the self-regulated subjects, and on the very KR either before or after the trial. They found a
same trials in the acquisition schedule sequence. benefit for learning when self-regulation occurred
The key difference was that the subject in the after the trial.
self-regulated group actively chose which trials These findings, however, represent a puzzle in
would receive KR; the yoked controls did not. the literature. Self-regulation appears to facilitate

9.5

9.0

8.5

gi 8.0
~ 7.5
E
....
~ 7.0

6.5

6.0

5.5

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Acquisition 1 Acquisition 2 Retention
Trial blocks

FIGURE 12.14 Effects of a learner-determined feedback schedule in acquisition and retention relative to a frequency-yoked
control group, a no-feedback control group, and a summary-feedback group.
Reprinted with permission from Research Quarterly tor Exercise and Sport, vol. 68, pgs. 269-279. Copyright 1997 by the American Alliance for Health, Physical Education,
Recreation and Dance, 1900 Association Drive, Reston, VA 20191.
 Augmented Feedback wi 419

the learning process when practice conditions must also increase. The effects of the KR delay
allow subjects to decide when to receive aug- and the intertrial interval are then confounded,
mented feedback (this chapter) or when subjects so that any resulting change in learning cannot
are given control over the scheduling of trials logically be attributed exclusively to either one of
for multiple tasks (chapter 11). But, recall also them. This fact sometimes makes it difficult to be
from chapter 11 that subjects often have very certain about the particular roles these intervals
weak judgments about their own learning. For have in the learning process, as we see in the fol-
example, Simon and Bjork (2001) found that sub- lowing sections.
jects in a blocked practice schedule severely over-
KR-Delay Interval
estimated their retention performance, and that
random practice subjects underestimated their The KR-delay interval is the amount of time KR
performance. Earlier, Baddeley and Longman is delayed after a movement. Many experiment-
(1978) found that subjects would have preferred ers have examined feedback delays and motor
to undergo a massed practice schedule, although learning, beginning with Lorge and Thorndike
it was the least effective schedule for learning. If (1935). For a variety of reasons, scientists have
the basis for self-regulation is to facilitate acquisi- always expected to find that increasing the KR
tion performance, then why is learning not detri- delay degrades learning. One reason is that
mentally affected, as we have seen in situations analogous effects in instrumental learning in
such as the contextual-interference effect? How animals are particularly strong (Lieberman,
can these findings be reconciled? Vogel, & Nisbet, 2008). Delaying the reward
One possibility suggests that the mere decision (e.g., a pellet of food) slightly in time from the
to receive augmented feedback or not engages the animal's bar-press movement has large effects
learner in the process of self-assessments during on animal learning, and delaying the reward too
practice (see also Cleary, Zimmerman, & Keating, much eliminates learning completely (Fantino &
2006). As we will discuss later on page 426 in Logan, 1979; Tarpy & Sawabini, 1974). Scientists
"How Augmented Feedback Can Degrade Learn- expected something like this for KR delay in
ing," factors that encourage the processing of human motor learning as well. A second reason
inherent feedback are usually considered strong is that because movement information is lost
learning variables. Another suggestion, offered rapidly from memory (e.g., Adams & Dijkstra,
by Chiviacowsky and Wulf (2005), is that the 1966), learning should be less effective as the
strong tendency to request feedback after "good" feedback delay from the associated movement
trials could mean that the KR is serving a strong is increased. This would seem to weaken the
motivational role-confirming the learner's possibility for the learner to associate commands
hunch that the trial's performance was indeed for the movement with its actual outcome-a
"good" (similar to the rationale underlying the concept critical to many early theoretical ideas
bandwidth KR effect). Support for or rejection of about learning.
these ideas awaits further research. However, as reviewed by Salmoni and col-
leagues (1984), the experiments in human motor
Temporal Locus of KR learning examining the delay of KR have almost
uniformly failed to show that increasing the KR
The next two sections deal with the question of delay has any effect at all. For example, Lorge and
when KR is presented in the events prior to and Thorndike (1935) used delays of either 1, 2, 4, or
following a practice trial. The question really con- 6 s and found no effect in an acquisition phase;
cerns the three intervals defined in figure 12.1- but no transfer design was used here to evaluate
the KR delay, the post-KR delay, and the intertrial effects on learning. Perhaps the delay was not
interval-and the ways in which experimentally sufficiently long. Other studies have used much
altering these intervals affects learning and per- longer delays ranging from a few seconds to a
formance. The problem is complicated by the few minutes; one study even used a delay of one
fact that when one of the intervals is lengthened week! Whereas a few studies have shown small,
experimentally (e.g., KR delay) and another is somewhat inconsistent effects on performance,
held constant (e.g., post-KR delay), then thethird the majority of research has shown no effect (e.g.,
interval (in this example, the intertrial interval) Schmidt & Shea, 1976). Recent work has used
420 Motor Control and Learning

various transfer designs to assess the temporary The degrading effects of instantaneous KR are
versus relatively permanent effects of KR delay. strikingly similar to the effects of concurrent KR,
There are numerous studies showing no effects, discussed in the previous chapter, in the study by
or at best very small effects, and we must doubt Armstrong (1970b). Take another look at figure
that delaying KR has a detrimental effect on motor 11.19 (p. 387). In chapter 11 we discussed how
learning. guided practice degraded learning of a spatial-
In contrast, there is some evidence to suggest temporal pattern, relative to a terminal-feedback
that detriments to learning can occur if the KR condition. Armstrong also included a condition
delay is too short. Swinnen, Schmidt, Nicholson, in which augmented feedback was presented
and Shapiro (1990) compared groups of subjects concurrently, as the subject performed the task.
who received KR after each trial-either at a Although this concurrent feedback had a positive
short delay after performance was completed influence during practice, it severely degraded
(3.2 s) or instantaneously upon completion of the learning as seen in the transfer phase, suggest-
trial. As illustrated in figure 12.15, acquisition ing that the concurrent feedback provided only
performance was not affected on the first day a temporary boost to performance. These det-
of practice by the KR conditions. Performance rimental learning effects have been replicated
improvements increased steadily for the delayed- often (Maslovat, Brunke, Chua, & Franks, 2009;
KR group on a second day of practice, but not Ranganathan & Newell, 2009; Schmidt & Wulf,
for the instantaneous-KR group. Learning, as 1997; Vander Linden et al., 1993); but they can
measured in no-KR retention tests after various be lessened by reducing the relative frequency
time intervals, was also facilitated by having KR of trials accompanied by concurrent feedback
delayed for a short time. It seems that the instan- (Camachon, Jacobs, Huet, Buekers, & Montagne,
taneous KR enhanced performance to a point, 2007; Park, Shea, & Wright, 2000). This evidence
but retarded both continued improvement and supports the interpretation that frequent, concur-
retention after that. rent feedback results in a learning effect that is

340

300

260

Cl>
0I.)
220
en
180

140

2 3 4 5 6 2 3 4 5 6 2 2 2
1-ol(--- Day 1 ------I 1- - - Day 2 ----1 10 min 2 days 4 months

Blocks of 15 trials
Acquisition ~ Retention ___.._

FIGURE 12.15 Performance scores of instantaneous- and delayed-KR (knowledge of results) conditions in acquisition and
retention.
Reprinted, by permission, from S.P. Swinnen et al., 1990, "Information feedback for skill acquisition: instantaneous knowledge of results degrades learning;' Journal of
Experimental Psychology: Learning, Memory, and Cognition 16: 712. Copyright© 1990 by the American Psychological Association.
 Augmented Feedback 1m 421

highly dependent on maintaining the provision of as they could occur quite rapidly for these very
concurrent information to support performance. simple motor tasks, and varying the length of
the interval might not severely limit processing.
Post-KR-Delay Interval
Next, consider the other portion of the intertrial lntertrial Interval
interval::...._the post-KR-delay interval, or the time The intertrial interval, or the sum of KR delay and
between the presentation of KR and the produc- post-KR delay (figure 12.1), has been the object
tion of the next movement. In contrast to the of considerable indirect study-mainly because
hypothesis that the subject is trying to remember it covaries when either one of the intervals com-
the aspects of the movement during the KR-delay posing it varies, and not because of much interest
interval, during the post-KR-delay interval it in the intertrial interval per se. According to a
appears that other processes are occurring. In review by Salmoni and colleagues (1984), there
particular, KR has now been delivered, likely indi- are many conflicting results on intertrial-interval
cating that the movement was incorrect in some effects for performance during the acquisition
way. Now the learner must generate a movement phase, obtained from a variety of experimental
that is different from the previous one, hopefully procedures; little generalization seems pos-
one that is more correct. So, in contrast to the sible. McGuigan (1959) and Dees and Grindley
hypothesis that during the KR-delay interval the (1951) have shown, however, that increasing
learner is storing movement information, in the the intertrial-interval length increases learning
post-KR-delay interval the learner is thought to be as measured on no-KR transfer tests, similar to
an active and creative movement problem solver. distributed-practice effects discussed in chapter
If the subject is actively processing KR to 11. Perhaps longer intertrial intervals result in
change the movement during post-KR delay, increased forgetting of the solution to the motor
then shortening the post-KR-delay interval past problem generated on the previous trial and thus
a certain point should decrease learning in the require an active generation of the motor program
task, as the person would not have sufficient time again on the next trial. These forced generations
to develop an effective new movement. Some could be very important for the learning process, as
support for this view exists in the verbal learn- has been inferred from the contextual-interference
ing literature using concept1ormation tasks (e.g., literature discussed in the previous chapter.
Bourne & Bunderson, 1963; Bourne, Guy, Dodd,
& Justesen, 1965; Croll, 1970; White & Schmidt, Interpolated Activities
1972). The literature on motor learning and per- During KR Intervals
formance, however, does not show close parallels What is the effect of requiring the learner to per-
to these findings for concept formation. In the form various activities during otherwise "empty"
acquisition phase, decreasing the post-KR-delay KR intervals? This question is motivated by an
interval does have slight detrimental effects on information-processing viewpoint about KR
performance accuracy in both adults (Weinberg, according to which certain other activities could
Guy, & Tupper, 1964) and children (Gallagher & interfere with various processes that occur during
Thomas, 1980), but no transfer designs were used these KR intervals and thus the effects should
in these studies to assess learning effects. When be seen in learning of the task. As we will see,
transfer designs are used, however, decreasing however, various interpolated activities either
post-KR delay also degrades learning, but only have no influence, a positive effect, or a negative
when KR delay is held constant, not when the impact on learning, depending on the nature of
intertrial interval is held constant. Salmoni and the interpolated activity and the delay interval
colleagues (1984) argued, therefore, that it was the during which it is interpolated.
intertrial interval that seemed to be the important
one for learning. But there is still some evidence Interference During
that learning might be reduced when the post-KR the KR-Delay Interval
delay is very short. Taken together, the evidence The influence of various activities during the
does not suggest that the length of this interval, KR-delay interval may be referred to as "interfer-
per se, is very important for learning. But this is ing" if they distract the learner from processing
not to deny the role of processes that occur here, the inherent feedback from the performance just
422 Motor Control and Learning

completed. For example, Shea and Upton (1976) session with KR present, there were essentially
had subjects perform linear-positioning move- no differences between these subjects and
ments, but two positions were to be practiced another group of subjects who did not estimate
and learned on each trial rather than one. On a their errors. But in a transfer test without KR,
given trial, the subject would produce movement the subjects who were estimating maintained
1, then movement 2, then would engage in the performance nearly perfectly, whereas those
performance of other movements (or would rest if subjects who did not estimate regressed sys-
in the other condition); then after 30 s the subject tematically over trials. One interpretation is that
would receive KR about movement 1 and move- the estimation conditions in acquisition forced
ment 2, then engage in the next trial, and so on. the subjects to attend to their own movement-
Filling the KR-delay interval increased absolute produced (inherent) feedback to a greater extent
error on the acquisition trials, indicating that the than the no-estimation conditions did, thus
extraneous movements had a negative effect on enabling them to acquire an error-detection
performance. And, in the no-KR transfer trials it capability. This capability was not particularly
seemed clear that the decrements in performance useful in acquisition because of the powerful
caused by the extraneous movements did, in fact, guiding properties of KR. But in no-KR transfer,
interfere with the learning of the tasks. Marteniuk subjects who had gained this error-detection
(1986), Swinnen (1990), and Lieberman and col- capability through estimation in acquisition
leagues (2008) provided similar results using were able to maintain performance, whereas the
more complex motor tasks. no-estimation subjects were relatively unaware
What is happening here? One interpretation of their own errors and drifted off target. Swin-
of these findings is that the subjects usually nen (1990; Swinnen et al., 1990) extended and
engaged in various information-processing refined the Hogan-Yanowitz paradigm in various
activities during the KR-delay interval and that ways, using different tasks and transfer tests,
the requirement of the extraneous movements in an attempt to understand these phenomena
in some way interfered with this processing, more completely. Overall, there continues to be
degrading learning as it did. What kind of pro- support for the notion that asking for error esti-
cessing might this be? Marteniuk (1986) argued mation in acquisition is effective for learning as
that the interference is from relatively high-level measured on no-KR transfer tests.
planning processes. But it is also possible that But some additional experiments suggest
the subject must retain in short-term memory that these effects might be more complex than
the sensory consequences of the movement originally conceptualized. Two recent studies
until the KR is presented so that the two can have revealed this to be the case. In one study,
be compared. The retention of information is Liu and Wrisberg (1997) investigated the effects
important in order to develop an error-detection of subjective estimations of movement form
capability (capability to detect errors based on error in a throwing task by the nondominant
inherent feedback sources). If other movements limb (Kernodle & Carlton, 1992). Subjects in
are required, then there will be either a blocked two groups saw the outcome of their throw
capacity to hold the information in short-term either immediately or after a 13 s delay. In two
memory or a reduced precision of the inherent other groups, the subjects provided subjective
feedback, resulting in less effective use of KR estimates of their throwing/arm either just after
when it is presented. seeing the outcome of the throw or during the
delay interval. As shown in figure 12.16, the
Subjective Estimations During performance of these two subjective estimation
the KR-Delay Interval groups was more accurate in retention than
Support for the interpretation just outlined is that of the two groups who did not estimate
provided in situations in which subjects are their movement form. From the perspective
encouraged to undertake error estimation during suggested earlier, this result is rather surprising
the KR-delay interval. Hogan and Yanowitz because in the immediate+ estimation group,
(1978) asked some subjects to estimate their the subjective estimation occurred after the KR,
own errors in a ballistic-timing task prior to not during the KR-delay interval, which is typi-
receiving KR on each trial. In an acquisition cal of most studies of this type. One view of the
 Augmented Feedback 423

:§'
"§ 7
i::'
Jg
:c...
~6
i)'
Cll
:;
(.)

~ 5
Q)
(.)
c:
Cll
E
0 4
't:
Q)
Q.

Aquisition blocks Retention blocks

FIGURE 12.16 Combined effects of feedback delay and estimation in the acquisition and retention of a throwing task using
the nondominant limb.
From J. Liu and C.A. Wrisberg. Adapted with permission from Research Quarterly for Exercise and Sport, vol. 68, pgs. 145·151. Copyright 1997 by the American Alliance for
Health, Physical Education, Recreation and Dance, 1900 Association Drive, Reston, VA 20191.

results is that merely estimating something that These two experiments suggest that error
will be confirmed or corrected by the augmented estimation is an important factor in the use of
feedback is not enough-perhaps one needs to augmented feedback in motor learning. The con-
estimate something about the performance itself, tribution of error estimation to learning appears
which is then supplemented by other augmented to be diminished if it is not accompanied by
information and used in the problem-solving augmented feedback (Guadagnoli & Kohl, 2001).
process. Yet it also appears that estimating something
Another study (Guadagnoli & Kohl, 2001) about performance that encourages the learner
offers a related idea regarding the combined to interpret the augmented feedback provides
effects of subjective estimation and reduced a boost to learning as well (Liu & Wrisberg,
relative frequency of KR. Subjects performed 1997).
150 trials in a force-estimation task followed by The issues about error detection are important
a no-KR retention test one day later. Four groups for theoretical reasons, but there is a strong prac-
were formed based on the factorial combination tical application also. We can think of the self-
of relative frequency of KR (100% vs. 20%) and detected error as a kind of substitute for KR, as it
error estimation (every trial vs. no estimation). informs the subject about the size and direction
The 100% relative-frequency condition produced of the error that was just made. It is unfortunate
the most accurate retention, but only if accom- that nearly all the focus in learning environments
panied by error estimation during practice (see is on performance and that there is almost no con-
figure 12.17). If KR was provided on every trial cern for the development of the learner's error-
in the absence of any error estimation, then this detection capacity (but see Schmidt & White,
condition produced the most error in retention. 1972). If procedures could be developed for
The performance of the other two groups showed increasing the strength of error detection,
that error estimation on every trial was only then learners could develop hypotheses about
moderately effective when KR was presented their performance that could then be checked
on only 20% of the trials, but that reduced rela- against the objective information provided
tive frequency was moderately effective even in later in the form of augmented feedback from
the absence of error estimation (perhaps due to the teacher or coach. Effective teachers and
spontaneous estimation in this group). coaches attempt to establish such error-detection
424 Motor Control and Learning

360

340

I... 320
...
0
Q)
300
.l!!
:I
280
0I/)
.c 260
<(

240

220

With No With No
estimation estimation estimation estimation
100% KR 20% KR

FIGURE 12.17 Combined effects of feedback frequency and estimation in the retention of a force production task.
Data from Guadagnoli and Kohl 2001.

capabilities that can be effectively used for self- effects, it would appear that instructors should
evaluation when the teacher or coach is not be more concerned about extraneous activities in
present. the KR-delay interval than in the post-KR-delay
interval.
Interference During the Post-KR-Delay
Interval
The focus of processing activities during the Theoretical Issues:
KR-delay interval is on movement-produced
inherent feedback. During the post-KR delay,
How Does Augmented
however, the processing activities are likely Feedback "Work"?
focused on using augmented feedback to alter
movement behavior on the next attempt. A The previous sections have presented various
number of early experimenters who interpo- separate facts in connection with the function-
lated activities in the post-KR interval showed ing of augmented feedback in motor learning
that performance was degraded (e.g., Boucher, situations. Some of these have obvious relevance
1974; Rogers, 1974; but see Magill, 1973), but for practical situations, whereas others have dis-
these studies did not use transfer procedures to tinct implications for how we believe feedback
assess learning (see Schendel & Newell, 1976, operates in humans to facilitate learning. In
for a discussion). Later experimenters who used this section, we consider some of these implica-
transfer tests produced mixed results. Swinnen tions.
(1990) and Benedetti and McCullagh (1987)
found that interference during the post-KR delay How Augmented Feedback Can
was detrimental as measured in a no-KR reten-
tion test; Lee and Magill (1983a, 1987) found no
Enhance Learning
detrimental effects of interpolated activities mea- The research presented in this chapter sug-
sured in a transfer test; and Magill (1988) found gests three possible ways in which KR and KP
that such activities were actually beneficial. The operate to affect learning in a positive way, and
rather equivocal nature of these findings makes theories of learning have generally adopted one
it difficult to infer practical applications. How- or more of these positions. Both KR and KP are
ever, given the comparative strength of these considered to have informational, motivational,
 Augmented Feedback 425

and associational functions. These concepts are performance (Newell, 1991)-than through
considered next. the rewarding of correct movements and the
"punishment" of incorrect ones. This interpre-
Informational Functions
tation would seem to contradict the findings
In previous sections we have drawn attention to from manipulations of bandwidth KR; in these
a number of features that are common to KR in experiments, information conveying to the
human motor learning situations and reward in subject that performance was correct gave an
animal learning situations. Both KR and reward additional boost to learning in comparison to
are presented contingent on the nature of the the learning in no-KR trials (Lee & Carnahan,
movement, and both are given after the move- 1990a). We suggest that the boost to learning
ment. What is the evidence that KR and reward came from the information content provided by
are really different? this "no error" type of feedback. That subjects
That KR and reward might be similar is not a resist making changes to performance on the
new idea at all, and it is the foundation of the Law basis of what could be "noise" might be a way of
of Effect, from Thorndike (1927; see Adams, 1978, avoiding the negative influences of too-frequent
1987). This law states that the organism tends to augmented feedback (see "Inducing Maladap-
repeat rewarded movements and to extinguish tive Corrections").
(or avoid) movements followed either by no One further suggestion is that KR has optimal
reward or by punishment. For motor learning, informational value when the learner is uncertain
according to this concept, KR indicating small about the reliability of his inherent sources of
errors or no error was thought to be a type of information. A dictionary provides a useful anal-
"reward," and KR indicating large errors was ogy here. The dictionary is like KR in that it is an
thought of as "punishment." In this way, the externally available, objective, and reliable source
movements followed by nonreward were elimi- of knowledge, providing augmented information
nated, and those followed by reward (i.e., zero such as the spelling or meaning of a word. The
or small error) tended to be repeated, leading to decision to consult a dictionary arises because
decreasing errors with practice. we have questioned the reliability of our inherent
However, numerous lines of evidence suggest knowledge; we do not consult the dictionary oth-
that humans do not use KR as proposed by this erwise. Thus, the dictionary provides the means
interpretation of the Law of Effect. First, when for assessing (and improving) the reliability of
KR is not presented (on no-KR trials), subjects our spelling knowledge. One hypothesis arising
tend to repeat the given movements rather than from this analogy is that augmented feedback
to eliminate them. Only when KR is presented do can be optimally useful when the subject asks
subjects change their movement behaviors, and for it-a concept that has received support from
then quite clearly in the direction of the target. recent experiments (see the earlier section on
It would seem that subjects are not using the self-regulated KR).
KR as a reward, but rather as information about
what to do next. In addition, even a short delay Motivational Functions
of reward in animal learning severely retards As mentioned earlier, receiving information
acquisition, and delaying reward by 30 s or so like KR and KP can play a strong motivating, or
can eliminate learning. Of course, we do not find "energizing," role. Augmented feedback may
these effects at all in humans, as the delay of KR make the task seem more interesting, keep the
seems to have no effect on motor learning. Thus, learner alert, cause the learner to set higher per-
reward in laboratory rats and KR in humans seem formance goals, and generally make boring tasks
to involve fundamentally different principles of more enjoyable. Some of the effects of motivation
operation (see "Elwell and Grindley on Knowl- are probably performance phenomena, which
edge of Results"). can be expected to subside when the feedback is
For these major reasons, the current belief withdrawn after training. But there is an indirect
about augmented feedback is that it produces learning effect that should not be ignored. When
learning more by the provision of informa- learners are highly motivated, they are inclined
tion about what was wrong with the previous to practice more often, longer, and with more
trials-and by prescriptive means to improve intensity and seriousness. Of course, deliberate
426 Motor Control and Learning

ELWELL AND GRINDLEY ON KNOWLEDGE OF RESULTS

Elwell and Grindley (1938) provided the first major challenge to Thorndike's ideas regarding
the role of KR in motor learning; this is developed more completely in three subsequent papers
(Dees & Grindley, 1951; MacPherson, Dees, & Grindley, 1948, 1949). They suggested that KR
provided more than just a rewarding function, and their arguments formed the basis for what
was later called the informational role of KR. The authors stated:
"In the acquisition of a muscular skill, such as that described in the present paper, the learning
cannot be regarded merely as the strengthening of the tendency to repeat movements which
have been 'rewarded' (by a high score). If a subject missed the bull's-eye he tried, next time, to
correct for his error by altering his response in the appropriate direction .... Knowledge of results,
when the movement was not completely successful (i.e., when it did not result in a bull's-eye)
introduces also a tendency towards a specific kind of variation of the response which has just
been made. We may call this the 'directive effect' of knowledge of results:' (p. 51)

practice per se is a critical variable for learn- ance function toward the target; it also provides
ing, and any factor that increases it will almost a rule about the relationship between internal
surely enhance learning (Ericsson, 1996; Ericsson, commands and the outcomes that were produced
Krampe, & Tesch-Romer, 1993). Recent evidence in the environment.
suggests that there may be a more direct effect
on learning as well (Chiviacowsky & Wulf, 2007; How Augmented Feedback Can
Lewthwaite & Wulf, 2010), which may require Degrade Learning
revised views of the motivational role of feedback
in the future. Another view of how KR works is that it guides
the learner to making the correct movement.
Associational Functions Thus, when the learner makes a movement, KR
A different view is that KR is associational- informs the person about how the movement
providing associations between stimuli and was inadequate, and the learner then changes
movements. One version of this concept is pro- the movement to one that (hopefully) will be
vided within schema theory (Schmidt, 1975b), more adequate. Augmented feedback thus car-
according to which KR is thought to operate ries inherent "instructions" about which aspects
associationally as well as in the ways that Adams of the movement should be changed, as well as
(1971) has suggested (both theories are discussed about the directions those changes should take.
in more detail in chapter 13). In schema theory, According to this position, KR does not provide
with respect to rapid movements that are presum- any direct strengthening of the movement but
ably controlled by motor programs, the person creates it indirectly by guiding the person to the
associates the KR received on a trial (a measure proper action. Once the proper actions are being
of what happened in the environment) with the produced, other processes take over to help the
parameters of the generalized motor program person learn the task.
(GMP) that were issued to produce that outcome This view is fundamental to Adams' (1971)
in the environment. With practice, the learner learning theory, which says that KR presented
comes to develop a rule (or schema) about the after each trial of a slow positioning movement
relationship between what the limbs were "told guides the person toward the correct location.
to do" and "what they did when told to do it." Then, as the learner achieves positions close to
On this basis, knowing what kinds of internal the target, she also receives kinesthetic feedback
commands tend to produce certain kinds of associated with the proper position, and this
movements, the learner has a way of selecting feedback forms an internal representation of
the parameters of the movement on future trials. being at the target (a reference of correctness).
Thus, in this view, KR serves more than a guid- This internal representation becomes stronger
 Augmented Feedback 111 427

with each successive trial near the target and make the error larger on the next trial. The idea is
thus provides an increasingly effective means for that KR induces movement variability, not all of
detecting errors. Thus, according to Adams, KR which is adaptive in producing improved learn-
has a guidance role in driving the subject closer ing. Sometimes augmented feedback can have
and closer to the target so that a reference of cor- maladaptive corrective properties (R.A. Bjork,
rectness can be formed. personal communication), in which case with-
Considered in this way (as envisaged by holding feedback (and stabilizing performance)
Adams, 1971), the guiding influences of aug- seems to be beneficial for learning. Presenting
mented feedback on learning should always be information that encourages a subject to correct
positive. As we have seen, however, in some an action that was essentially accurate may have
experiments the KR effects showed that increased a detrimental impact on learning (Schmidt, 199la;
guidance degraded learning (leading to doubts Schmidt & Bjork, 1992).
about Adams' theory; see chapter 13). We con'- Bandwidth KR effects illustrate how mal-
sider reasons why feedback can degrade learning adaptive corrections may be avoided. Under
in the next sections (see also Salmoni et al., 1984; bandwidth KR conditions, there exists a zone
Schmidt, 1991a; Schmidt & Bjork, 1992). of acceptable error within which movement is
considered correct. Defining the actual width of
Blocking Other Processing Activities the band of correctness, as well as what would
When augmented feedback is provided fre- be considered maladaptively corrective and what
quently, immediately, or otherwise in such a would be considered too imprecise, is a challenge
way that various processing activities are not for future research. However, we suspect that an
undertaken, then there will likely be a decrement optimal KR bandwidth may be closely related to the
in learning. One of the negative influences of aug- precision of an individual's motor control capa-
mented feedback may be to block the processing bilities, although even within an individual this
of inherent sources of feedback, which then leads is likely to change (e.g., with learning and aging).
to the failure to learn error-detection capabilities
for this task. Augmented feedback is often a very
salient source of information, and one that will be Summary
attended to even when doing so may not be in the
learner's best interest (Buekers et al., 1992). The Feedback is that class of sensory information that
presentation of instantaneous KP (Swinnen et al., is movement related, and it can be classified into
1990), which was discussed earlier, is an example two basic categories-inherent (intrinsic to the
of a case in which the saliency of the augmented task) and augmented (supplementary to the task).
feedback is maximized. We interpret results Two major classes of augmented feedback include
of this type as suggesting that the augmented KP, which is information about the form of the
feedback blocked the processing of alternative movement, and KR, which is verbal postmove-
sources of information and reduced the learning ment information about performance outcome.
effectiveness of the practice session as measured Much research suggests that the provision of aug-
in retention. mented information is the single most important
variable for motor learning (except for practice
Inducing Maladaptive Corrections itself, of course).
One of the fundamental views about the directive Information about the learner's movements
function of augmented feedback is that it tells the (KP) can be given through video feedback,
learner what went wrong and how to fix it. As recordings of the force-time characteristics of the
we found in our discussion of precision of KR, movement (kinetics), or representations of the
more precise KR can be beneficial, but only up to movement trajectories (kinematics); and all these
a point. The idea is similar here. When each trial appear to have positive effects on performance
is followed by information about errors, there is and perhaps on learning. The impact of KP on
a tendency for the subject to make a change for learning appears to be strongest when it precisely
the next trial based on that error. The problem is specifies information that is critical for movement
that motor performance is variable, and a change efficiency and that cannot be obtained from other
meant to correct a very small error might actually sources of feedback.
428 Motor Control and Learning

Research on KR precision shows that perfor- a. Choose any skilled trade that would
mance improves with increases in precision up to require motor learning (e.g., carpen-
a point, with no further increases in performance ter). Provide examples for each of the
thereafter. Presenting combinations of qualitative different kinds of inherent and aug-
and quantitative KR, based upon a goal-related mented feedback that could be useful
bandwidth of correctness, has strong implications in learning this trade.
for both application and theory. b. Using one of the augmented-feedback
Early research indicated that the relative fre- examples from the answer in la,
quency of KR (the percentage of trials on which describe how the temporal locus of
KR was given) was irrelevant for learning, presenting this information to the
whereas the absolute frequency (the number of learner would affect learning.
KR presentations given) was the critical deter-
c. Models, physical guidance devices,
minant. More recent data using transfer designs
and augmented feedback are meth-
contradict this position, indicating that both are
ods of providing external sources of
clearly important. Trials on which no KR is given
information to the learner. Compare
appear to contribute to learning in the task, but
and contrast these methods in terms
not as much as the KR trials do. The trials-delay
of their potential effect in a learning
and summary-KR procedures, in which the KR
environment.
for a given movement is separated from the
movement by other trials, were shown to produce 2. Find a research article (published in the past
detrimental effects on motor performance but five years) that examines the influence of
positive effects on learning. knowledge of performance on the perform-
The effect of delaying KR, that is, the effect of the ing or learning of a motor skill.
interval from the movement until KR is presented,
has been found to be negligible for learning most
motor tasks, as long as KR is not presented too Web Resources
soon after performance. Filling this interval with
activities not related to the task degrades learn- This Web site describes augmented feedback
ing. However, filling this interval with activities teaching devices to facilitate motor learning:
related to the task, such as subjective estimation www.thespeedstik.com/
processes, enhances learning. If the post-KR-delay
interval-the interval from the KR until the next This Web site offers golf swing training aids:
movement-is too short, subjects appear to have www.dwquailgolf.com/training/your_pro_
difficulty generating a new and different move- swing_trainer. html
ment on the next trial. However, filling this interval
has uncertain effects on learning.
Augmented feedback appears to have several Notes
possible mechanisms for enhancing learning.
It acts as information. It acts to form associations 1 Although we have distinguished between various

between movement parameters and resulting types of augmented feedback, of which KR is one, we will
action. And it acts in a motivational role. Aug- generally refer to many aspects of this work in relation
mented feedback also has a guidance property to the term KR. However, exceptions will be made when
that can enhance performance but degrade a clear distinction is necessary.
2 Note the differences between the use of these terms
learning.
and the use of random and blocked practice in chapter 11.
In that work, the same task is repeated in blocked practice,
Student Assignments and switching between tasks occurs in random practice.
In the Lee and Carnahan experiment, the same task is
performed on each trial, but KR is given either about the
1. Prepare to answer the following questions same segment over a series of trials (blocked KR) or about
during class discussion: a different segment on successive trials (random KR).
 13

o far in the discussion of motor learning, our requirements, the development of error-detection
major concern has been the most important processes, and the like.
empirical findings about the acquisition of skills. The chapter is divided into three major sec-
It is time now to consider the underlying reasons tions. The first section presents some fundamen-
for these findings and to ask about the nature tal ideas about the learning process. With this
of the motor learning processes that cause the information in mind, together with information
motor system to behave in the ways identified from the previous two chapters, we then present
in previous chapters. A part of this process is various theoretical views about motor learning.
theoretical, in that we search for a fundamental Two of these, which are considered major theo-
understanding-stated as theories-of how the retical advances in the history of motor learning
system "works" when it learns. But part of it is research, are presented in the second section of
practical, in that a solid understanding of the sys- the chapter. The third section presents different
tem's function provides suggestions for practical perspectives on the learning process-perspec-
application to situations that have not actually tives that in one form or another can be consid-
been studied; there is nothing as practical as a ered major hypotheses about learning.
good theory (Kerlinger, 1973).
In this chapter, we consider the many ways
in which various people have conceptualized Characteristics of the
the motor learning process. All these theo-
retical perspectives have as their basic goal an
Learning Process
understanding of the changes in skill that occur
with practice. However, we will see that a phe- Without a doubt, the most notable thing that
nomenon as broad and common as this can be happens when people practice is that they dem-
explained in various ways and at a number of onstrate increased proficiency in the task. Some-
levels of analysis (biomechanical, cognitive, and times this is so obvious that it hardly needs to be
so on). At the same time, we will see that the mentioned, while in other cases the changes are
concepts basic to these various theoretical ideas more subtle and require special methods, observa-
are already familiar from previous chapters, tion, and rationale in order to be examined. In this
having to do with such notions as the building section we describe a number of ways in which
of new motor programs, changes in attentional the learning process has been characterized,

Ill 429
430 Motor Control and Learning

in terms of the various descriptions of the ways involved improving the efficiency of the pattern.
individuals change in their capability to perform Other two-stage views were later suggested by
a motor skill with practice. Adams (1971) and Gentile (1972). A three-stage
view of learning was suggested by Fitts (1964;
Stages of Motor Learning Fitts & Posner, 1967) and later Anderson (1982,
1995). These three stages, referred to as the cogni-
Many have noticed that learners appear to pass tive, fixation, and autonomous stages, are discussed
through relatively distinct stages (or phases) as in more detail in this section. As you read these
they practice a skill. Bryan and Harter (1897, 1899) explanations, however, remind yourself that these
were among the first to study the acquisition of stages are not discrete and fixed stages, but have
skill in considerable detail (see "Bryan and Har- "fuzzy" borders (see Anson, Elliott, & Davids,
ter' s Hierarchy of Habits"). Learning was defined 2005, for an excellent discussion of Fitts' stages
as a two-stage process by Snoddy (1926). Subjects of learning).
in Snoddy' s research learned to make hand move-
ments but saw their hand only in a mirror-a Stage 1: Cognitive
task that requires abilities needed by a dentist, When the learner is new to a task, the primary
for example. According to Snoddy (1926), the problem to be solved concerns what is to be
adaptation stage involved acquisition of the neu- done-that is, what actions need to be taken in
romuscular pattern required to perform the task. order to achieve the goal of the task? Naturally,
Once the pattern was learned, the facilitation stage considerable cognitive activity is required so that

Bryan and Harter's Hierarchy of Habits

A fascinating early set of studies regarding the perceptual and motor changes that occur with learning
was conductedbyWilliam Lowe Bryan (a psychologist) and Noble Harter (a telegrapher and student
of Bryan's). The result of their shared interests was two landmark papers regarding the acquisition
of telegraphic skills (Bryan & Harter, 1897, 1899). In these papers, Bryan and Harter presented the
results of experiments that cornpared novice and expert telegraphers, as well as data they obtained
by charting the acquisition of telegraphy skill over many months of practice. These papers present
many interesting findings(Lee & Swinnen, 1993), but we will focus on one in particular.
Skill, in Bryan and Harter's view, was a process of achieving a hierarchy of habits. At the most
basicJevel, telegraphy involves the ability to discriminate (perceptually and motorically) between
unitsoftime,A dot is one "unit" of continuous auditory signal. A dash is three "units" of continuous
time. One unit of .no signal occurs between dots and dashes within a letter (e.g., the letter G is a
dash-dash~dot).Three continuous units of no signal denotes that a new letter is beginning, and six
units rnarksa new word. This "language" of telegraphy lent itself well to Bryan and Harter's view of
l(;J~rni!")ga~ a hierarchy of habits.The discrimination of time was learned quickly.
· The alphabet became the next challenge. This, too, is usually learned quickly, and performance
in sending and receiving code improves rapidly. However, Bryan and Harter then noticed something
peculiaraboutthe practice curves of some of their subjects: Periods of time would go by during
which little or no improvement occurred atall, which were followed later by rapid improvements. They
called these periods plateaus in performance that occur prior to the formation of a new, advanced
capability. They proposed that rather than hearing dots and dashes, with learning, the telegraphers
"hear" letters. With further practice they then "hear" words, and, for the most skilled, even larger units
of a sentence. Presumably, the plateaus in performance occur because the maximum performance
capability of one habit places a limitdn performance, which is then lifted when a higher-order habit
is formed.Although some ofBryan and Harter's views have been challenged at times (e.g., Keller,
1958), many of the basic concepts of progression through stages and to higher orders of skill have
peen retained in a number of conceptualizations of skill acquisition that remain popular today.
 The Learning Process 431

the learner can determine appropriate strategies 1972; Shaffer, 1971, 1980). Automaticity is usually
to try to get the movement in the "ballpark." evidenced with respect to particular kinds of
Effective strategies are retained, and inappro- simultaneous tasks, primarily those that we could
priate ones are discarded. As a result, the per- class as verbal-cognitive; some other motor task
formance gains during this stage are dramatic could in fact interfere with a performance in the
and generally larger than at any other stage in autonomous stage, as discussed in chapter 4 in
the learning process. Performance is usually detail. Even so, the performer gives the impres-
very inconsistent, perhaps because the learner is sion that she is performing without having to
trying many different ways of solving the prob- "pay attention" to the actions. This stage has the
lem. The use of instructions, models, augmented benefit of allowing the person to process infor-
feedback, and various other training techniques mation from other aspects of the task, such as the
(discussed in chapters 11 and 12) is most effective strategy in a game of tennis or the form or style
during this stage because they assist the learner of movement in ice skating or dance.
in this problem-solving process. Probably most A major problem for motor behavior research
of the improvements in the cognitive stage can is that this stage, which is of immense importance
be thought of as verbal-cognitive in nature, the for understanding high-level skills, is only rarely
major gains being in terms of what to do rather studied in experiments on motor learning. The
than in the motor patterns themselves. Adams reasons are probably obvious. In paradigms in
(1971) termed this stage the verbal-motor stage. which subjects practice on laboratory tasks, such
practice should continue for months before even
Stage 2: Fixation
approaching the levels of skill shown by high-
The second stage of motor learning begins when level musicians, athletes, and industrial workers.
the individual has determined the most effective It is very difficult to convince subjects to devote
way of doing the task and starts to make more this kind of effort in experiments. Alternatively,
subtle adjustments in how the skill is performed. we could use other, more natural tasks that learn-
Performance improvements are more graduat ers are practicing anyway; but it is difficult to
and movements become more consistent in the manipulate and control the many variables that
fixation stage. This stage can persist for quite a would need to be used for a scientific understand-
long time, with the performer gradually produc- ing of the learning processes.
ing small changes in the motor patterns that will Some efforts at understanding the principles
allow more effective performance. Many writers of automaticity have been made in this direction
(e.g., Adams, 1971; Fitts, 1964) think that the by Schneider and colleagues (Schneider & Fisk,
verbal aspects of the task have largely dropped 1983; Schneider & Shiffrin, 1977) in reaction-
out by this stage, with the performer concentrat- time (RT) tasks, and by Logan (1985, 1988) using
ing on how to do the action rather than on which speeded-decision tasks. Unfortunately, research
(of many) movement patterns should be pro- involving more complex motor tasks is rarely
duced. This stage and the next (autonomous) are taken to this stage of learning (but see Jabusch,
equivalent to what Adams called the motor stage. Alpers, Kopiez, Vauth, & Altenmiiller, 2009, for
Stage 3: Autonomous a recent exception).
After many months, perhaps years, of practice, Individual Differences and Motor
the learner enters the autonomous stage, so
named because the skill has become largely Learning
automatic in the sense discussed in chapter 4. Some important hypotheses for motor learn-
That is, the task can now be performed with less ing are framed in the language and methods of
interference from other ongoing activities. It is individual-differences research (see chapter 9).
easy to find examples of high-level performers Beginning with the concept that a given motor
engaging in secondary tasks without interfer- performance is based on some small set of under-
ence-for example, the concert pianist who can lying motor abilities, one hypothesis simply states
shadow digits or do mental arithmetic without that this set of abilities changes in its makeup as
interference while sight-reading and playing practice continues. The abilities themselves do
piano music (e.g., Allport, Antonis, & Reynolds, not change; this would violate the assumption
432 1111 Motor Control and Learning

(discussed in chapter 9) that abilities are to a large sensitive the person was to applied tensions. A
extent genetically defined and unmodifiable by second test called spatial orientation was a paper-
practice. But what does change, according to this and-pencil test related to a subject's perception
view, is the particular collection of abilities that of orientation in space.
underlie the skill being learned. First, Fleishman and Rich divided their group
of people into two, on the basis of their perfor-
Studies Using Individual-Difference
mance on the kinesthetic sensitivity test; then they
Variables
plotted the groups' performances separately for
Fleishman and Hempel (1955) and Fleishman and the two-hand coordination test. As seen in figure
Rich (1963) contributed important investigations 13.la, the subjects classified as high and low on
in this area. In the Fleishman-Rich study, subjects the kinesthetic sensitivity measure were not dif-
practiced the two-hand coordination task, in ferent on the two-hand coordination test in early
which two crank handles had to be manipulated trials; but later in practice, the subjects high in
to cause a pointer to follow a moving target on kinesthetic sensitivity began to outperform those
a target board (figure 2.5). Separate from the low in kinesthetic sensitivity. The interpretation
practice on this test, the subjects were given two of these results is that kinesthetic sensitivity
additional tests. In one, they were asked to lift is an ability that increases in importance with
small weights and to judge the weight relative practice, at least for this task. Next, consider
to a standard weight. This test was called kines- the spatial orientation measure (figure 13.lb).
thetic sensitivity because it seemed related to how The subjects classified as high on this test were

.
a;en 30

-co
c: 25
0

-UI
CD

c: 20
:I
.E

..
CD
en 15
co
CD
>
<( 10

a 2 3 4 5 6 7 8 9 10

..
a;
en 30
sc:
0 25

-UI
CD

c: 20
:I
.E

..
CD
en 15
co
g!
<( 10

2 3 4 5 6 7 8 9 10
b Blocks of four 1-min trials

FIGURE 13.1 Performance on the two-hand coordination test as a function of practice trials. (a) Groups classed as high
and low on a kinesthetic sensitivity test are plotted separately; (b) groups classed as high and low on a spatial orientation
test are plotted separately.
Reprinted, by permission, from E.A. Fleishman and S. Rich, 1963, "Role of kinesthetic and spatial-visual abilities in perceptual motor learning," Journal of Experimental
Psychology 66: 9. Copyright© 1963 by the American Psychological Association.
 The Learning Process 433

stronger performers on the two-hand coordina- some extent) from the spatial relations test; that
tion test than subjects classed as low, but only for is, the subjects high on this measure outperformed
initial performance. This result is interpreted as those classified as low. However, late in practice
evidence that spatial orientation is an ability that (but not early in practice), the two-hand coordi-
is important for early proficiency in this task but nation test performance could be predicted from
appears to have little to do with performance in kinesthetic sensitivity. This notion of prediction
later practice. implies that the correlation between the two-hand
Another way to view these results is that, for coordination test and, for example, the kinesthetic
the two-hand coordination test, there is some sensitivity test would be zero in initial practice
collection of abilities that underlies it on trial 1. and larger in later practice. More generally, the
This collection includes some abilities related to hypothesis that the collection of abilities underly-
the spatial orientation measure, and it does not ing some skill will change with practice says that
include abilities related to kinesthetic sensitivity. the correlations between measures of various
As practice continues, the collection of abilities abilities and the criterion task performance will
(but not the abilities themselves) changes, so that change with practice.
at the end of practice the task is made up of a
lntertrial Correlation Analyses
somewhat different set of abilities. This collection
could include some of the same abilities as in early Take any motor task and measure a large number
practice, but it now has abilities related to kines- of subjects on each of a series of trials. Then cor-
thetic sensitivity and does not have any abilities relate the performances obtained on every trial
related to spatial orientation. Practice results in with those on every other trial and arrange these
a shift in the abilities underlying a task (see also correlation values in an intertrial correlation matrix.
Bartram, Banerji, Rothwell, & Smith, 1985). Such a matrix is shown in table 13.1, reproduced
from Jones' (1962, 1966) work on the two-hand
Practice and the Predictability of Skilled coordination test. The bottom half of the matrix
Performance is omitted for simplicity (it is the mirror image of
If the collection of abilities underlying a particular the top half). There are a number of interesting
performance becomes rearranged systematically features of tables like this, as has been pointed
with practice, then the prediction of success in the out by Jones (1966).
skill will be based on different ability measures Remoteness Effects First, notice that across any
in early versus late practice. Fleishman and Rich row of the table, the correlations become system-
(1963, figure 13.1) showed that, early in learning atically smaller; they drop from 0.79 to 0.70 in the
(but not later in learning), performance on the first row, from 0.87 to 0.82 in the second row, and
two-hand coordination test could be predicted (to so on. The top row represents the correlations of

Note. The boxed-in section forms the diagonal of the matrix, and the shaded portion is the "superdiagonal:'
Adapted from Jones 1966.
434 Iii Motor Control and Learning

trial 1 with trial 2, trial 1 with trial 3, trial 1 with practice. In this sense, the task becomes "simpler"
trial 4, and so on up to trial 1 with trial 8. Thus, (in the sense of having fewer underlying abilities)
the number of trials between the two trials being with increased practice.
correlated increases as we move to the right
Individual Differences and Stages of
along any row. As a general rule, as the number
Learning
of intervening trials increases, the correlation
decreases. This effect is often called the remote- A variation of Fitts' stages of learning view by
ness effect, because the correlations between the Ackerman (1988, 1989, 1990, 1992; Ackerman &
trials depend on how "remote" (how far apart) Cianciolo, 2000) suggests that, early in practice,
the trials are from each other. performing the task should be based on abilities
What is the meaning of the remoteness effect? having to do with thinking, reasoning, mechani-
First, remember that the correlation between two cal knowledge, and so on. General intellectual
tests (in this case, two trials of the "same" test) is abilities (information-processing skills) are the
related to the number of common abilities they most important determiners of individual dif-
share. As two tests become more separated in the ferences in performance during the cognitive
practice sequence, and become less correlated, the stage of skill acquisition. Later in practice, these
argument is that these performances are becom- abilities should not be involved as much; and
ing dependent on fewer and fewer of the same perhaps other abilities such as movement speed,
abilities. In this sense, the remoteness effect is RT, strength, and steadiness become the most
just another way to say that the motor abilities important. Once the idea of the task has been
change with practice, as did Fleishman's research, acquired, the role of general intelligence as a
discussed in the previous section. determiner of individual differences drops off,
replaced by more "motor" abilities during the
Adjacent-Trial Effects Examine the data in table
fixation stage in performance.
13. l again, this time concentrating on the correla-
tions between adjacent trials-that is, between Predicting Individual Differences During Dif-
trials 1and2, between trials 2 and 3, and so on. ferent Stages According to Ackerman's theory,
These correlations can be found on what is called the correlation between tests of intellectual,
the superdiagonal (the shaded area), or the line of perceptual-speed, and psychomotor abilities will
correlations that lies just above the diagonal of differ during different stages in learning. These
the matrix (unfilled squares in table 13.1). Notice predictions are illustrated in the three graphs in
that, as the adjacent trials are chosen later and figure 13.2. Figure 13.2a suggests that the correla-
later in the sequence, the correlations steadily tion between general intellectual ability and task
increase. The correlation between trials 1 and 2 performance will be highest during the cognitive
is 0.79, whereas the correlation between trials 7 stage (stage 1) and will drop off quickly thereafter.
and 8 is 0.94, the highest in the matrix. This is The correlation between perceptual-speed tests
another well-known finding. The change in these and task performance should be low during the
adjacent-trial correlations along the superdiago- cognitive and autonomous stages (stages 1 and
nal suggests that performances become system- 3) but much higher during the fixation stage
atically more stable, in terms of their underlying (stage 2), as presented in figure 13.2b. Figure 13.2c
ability structure, as practice continues. depicts very little contribution of psychomotor
Practice as a Process of Simplification An even abilities until the autonomous stage (stage 3), at
more restrictive descriptor of the intertrial corre- which point increasingly higher correlations are
lation matrix is what is called the superdiagonal predicted.
form, for which any four arbitrarily chosen cor- Evidence for Ackerman's Integrated Model
relations within the matrix must possess a par- Ackerman asked subjects to perform a simple-
ticular mathematical relationship with each other. RT task for six sessions, during which subjects
(A discussion of the nature of this relationship is pressed a key on a numeric keypad in response
beyond the scope of this text, but see Jones, 1966.) to the number shown on a screen (e.g., press
The important point is that this superdiagonal the "l" key in response to the number "1"). As
form is derived from the hypothesis that the would be expected, subjects had little difficulty
number of abilities systematically decreases with in figuring out what to do in this task and there-
 The Learning Process 435

c
0
-~

~
0
0

c
0
~
~
0
0

c
0
~
~
0
0

Stage 1 Stage 2 Stage 3

c

FIGURE 13.2 Predicted correlation of individual differences in general intellectual, speed, and psychomotor abilities at dif-
ferent stages of learning.
Reprinted, by permission, from P.L. Ackerman, 1990, ''A correlational analysis of skill specific city: Learning, abilities, and individual differences," Journal of Experimental
Psychology: Leaming, Memory and Cognition 16: 887. Copyright© 1990 by the American Psychological Association.

fore would be expected to perform as if starting Then, in session 7, Ackerman transferred his
in stage 2 of practice rather than in stage 1. The subjects from the simple, compatible stimulus-
correlations of task performance with general response (S-R) mappings to less compatible
and perceptual-speed abilities are presented in mappings in which subjects pressed the key
figure 13.3. As can be seen, the correlations are designated by a two-letter abbreviation system;
higher for perceptual speed than for general the first letter indicated the numeric keypad
ability in the training phase, as is predicted for row (e.g., L =lower row), and the second letter
stage 2. Moreover, the general trend taken by indicated the numeric keypad column (e.g., M =
these correlations over the six sessions is similar middle column). Thus, the stimulus "LM" (lower
to the predicted correlations illustrated in figure middle) indicates that the "2" key should be
13.2 for stage 2 of practice. pressed; a "UR" denotes the "9" key. The rationale
436 111 Motor Control and Learning

was that this incompatible mapping would Error-Detection Capabilities

require considerable cognitive activity initially
in practice, forcing the learner back into stage It is well known that a major outcome of practice
1. On the right side of figure 13.3, the correla- is the capability to produce more effective move-
tions between RT performance and the abilities ment behaviors, but an additional outcome of
tests were much higher early in transfer for practice is that learners become more capable of
the general ability, but were similar to those evaluating their own movement behaviors. That
for perceptual-speed ability later in practice. is, it seems that a learner develops a kind of error-
Again, the shape of these functions over practice detection capability with practice that, in many
in the transfer phase appears rather similar to instances, can be used as feedback to inform the
the predictions seen in figure 13.2 for phase 1 individual about his own errors.
of practice. For example, one of the distinguishing features
These findings provide support for Acker- of expert jugglers seems to be their ability to detect
man's theory. In addition, Ackerman (1988) errors in the way that the hand releases the object.
found support for the psychomotor abilities If the force is too large or too small (affecting the
predictions when he reanalyzed data sets from amplitude of the throw), or if the angle of release
Fleishman (1956; Fleishman & Hempel, 1955, is slightly off line (affecting the location where it
1956). These data showed that the correlations will be caught by the other hand), then spatial and
between the rate of arm movement (a psycho- timing adjustments will have to be made in order
motor ability) and three different performance to maintain the juggling pattern. Beek and Lewbel
tasks increased with practice, as predicted by (1995) argued that novices and experts differ in the
the theory (see also Adams, 1957). More recent type of information they use and how quickly they
studies (Ackerman & Cianciolo, 2000), using a detect an error; novices tend to use visual feed-
flight-simulator task, have provided additional back of the object in flight, while experts can detect
support for the predictions shown in figure 13.2, an error by monitoring the sensory feedback as
a and c. However, the prediction that perceptual- (or even before) the object leaves the hand. Thus,
speed abilities become most important during some of the skill differences between novices and
phase 2 of skill acquisition (figure 13.2b) was experts can be accounted for by the "advance"
not supported. information an expert processes indicating that

0.5

s
~ 0.4
:cas
"C
c:
as 0.3
Ii:
c:

I
..::: 0.2
c:
0
~
'! 0.1
8
7 8 9 10 11 12 13 14 15
Training Transfer
(sessions of practice) (sessions of practice)

FIGURE 13.3 Test of theoretical predictions illustrated in figure 13.2.

Reprinted, by permission, from P.L. Ackerman, 1988, "Determinants of individual differences during skill acquisition: Cognitive abilities and information processing," Journal of
Experimental Psychology: General 117: 299. Copyright© 1988 by the American Psychological Association.
 The Learning Process 437

an error has occurred (or will occur). Another between the subject's subjective and objective
interpretation is that experts possess the ability scores should increase.
to monitor the motor commands sent to the body The statistic used to estimate this agreement
parts that execute the movements, by comparing was based on correlations (see section in chapter
using these commands to an internal model of 2 for review). For a block of 10 trials, each subject
the expected sensory consequences (Wolpert & would have 10 objective scores and 10 subjective
Ghahramani, 2000). scores. These pairs of scores were correlated for
Studying the acquisition of error-detection each subject separately for each of 17 blocks of
mechanisms in skills is complicated, however, trials in the experiment. The idea was that, if the
because different types of skills produce different error-detection capability was weak, almost no
results. A major distinction is between rapid and agreement would exist between the objective
slow movements, examined next. and subjective errors, and the correlation should
be near zero. But if error detection increased in
Objective-Subjective Correlations in accuracy with practice, then the objective and
Rapid Movements subjective scores should agree to a greater extent,
Schmidt and White (1972) used a ballistic-timing and the magnitude of the correlation statistic (r)
task in which the subjects moved a slide 23 cm, should increase over blocks of trials. 1
with a follow-through, so that the movement The average within-subject correlations are
time (MT) was as close to 150 ms as possible. presented in figure 13.4. On the first block, the
The subject made a movement, then guessed the average correlation was about 0.28, indicating
MT outcome score in milliseconds, and then was a relatively weak association between objective
given knowledge of results (KR, or the actual and subjective errors. But as practice continued,
score) in milliseconds. The subject's guess was the average correlation increased to the point
subtracted from the subject's actual score, and that on day 2 the values approached 1.0. This
was termed subjective error. The actual score was evidence suggests that the learners became more
subtracted from the goal score and termed objec- and more sensitive to their own errors through
tive error. If an increased capability to detect errors the development of error-detection processes (see
is acquired with practice, then the agreement also Rubin, 1978). 2

1.0

0.9

0.8
Cl)~
> IJ)
tl :g 0.7
Cl) ::I
:g .!:!.c:
IJ)
0.6
.'.. 0
g :"§ 0.5
Cl) Cl)
Cl) :: 0.4
>
;:;
0
CJ
al 0.3
:c
0 0.2

0.1

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

- - - - - - - - Day 1 [ [~oay2~
Blocks of ten trials

FIGURE 13.4 Average within-subject correlation between objective and subjective error as a function of practice trials.
(Increased correlation is interpreted as gains in capability to detect errors; correlations are transformed to Z' units.)
Reprinted, by permission, from R.A. Schmidt and J.L. White, 1972, "Evidence for an error detection mechanism in motor skills: A test of Adams' closed-loop theory," Journal of
Motor Behavior 4: 143-153, reprinted with permission of the publisher (Taylor & Francis Ltd, http://www.informaworld.com).
438 mi Motor Control and Learning

Objective-Subjective Correlations in performance and learning that applied to these

Slow Movements movements were the same as for any other kind
This is not so with slow movements. It appears of movement, and that using a well-established
that, for some slow movements at least, the set of empirical laws from positioning move-
error-detection processes may be responsible ments would produce a solid basis for theoriz-
for actually producing the action. Because there ing.
is ample time to use feedback, it is thought that
the subject in a positioning movement evaluates
A Feedback Emphasis
intrinsic feedback against the learned refer- Adams believed that all movements are made by
ence of correctness and moves to the position comparing the ongoing feedback from the limbs
recognized as correct (Adams, 1971). If so, the to a perceptual trace-the reference of correct-
error-detection capacity, being used to position ness, stored in memory, which is learned during
the limb at the target, cannot then be used again practice. When the person makes a positioning
after the movement as a basis for telling the movement, inherent feedback is produced that
experimenter about the error in positioning; if it represents the particular locations of the limb in
were used, the errors would all be close to zero, space. These stimuli "leave a trace" in the cen-
as the action is based on that position that feels
/1
tral nervous system (hence the name perceptual
like" it has minimum error. Then, if the subject trace). With repeated practice, the person comes
is asked to report the error in positioning, she closer and closer to the target over trials; and
will have no idea whether or not the movement on each trial another trace is laid down, so that
was on target. eventually a kind of collection" of traces devel-
/1

Schmidt and Russell (1974) performed an exper- ops. With practice (and KR), the learner's move-
iment analogous to that of Schmidt and White but ments become increasingly closer to the target
using a slow, linear-positioning task. In contrast and with increasing consistency. Therefore, each
to Schmidt and White (figure 13.4), Schmidt and trial provides feedback that tends to represent
Russell found consistently low within-subject the correct movement with increasing frequency.
correlations between objective and subjective In turn, the collection of perceptual traces comes
errors, with most of the correlations being only to represent the feedback qualities of the correct
about 0.20 even after 100 trials of practice. These movement. Then, on subsequent trials, the learner
findings suggest that the error-detection processes moves to that position in space for which the dif-
were used to position the limb in the slow task and ference between the ongoing (inherent) feedback
that further estimates of error after completion of produced and the perceptual trace is minimized.
movement were based largely on guesswork by
Because the perceptual trace associated with the
the subjects (see also Nicholson & Schmidt, 1991,
correct movement becomes stronger with each
who used timing tasks). These ideas figure heavily
KR trial, the errors in performance decrease with
in the concepts of schema theory, presented in the
practice.
next section of this chapter.
Adams' theory assured a guidance role for
In the 1970s, two important papers were pub-
KR (although his writings do not use this term).
lished on the motor learning process-dosed-loop
Learners, according to Adams, are not passive
theory (Adams, 1971) and schema theory (Schmidt,
recipients of reward, but rather are actively
1975b). Because of the impact that these theories
engaged in verbalization and hypothesis forma-
had on subsequent motor learning research, we
describe them in some detail next. tion about the task to be learned. To Adams, KR
provides information to solve the motor prob-
lem. After a trial, KR is given that provides infor-
Closed-Loop Theory mation about how the next movement should
be made to more closely achieve the task goal.
Adams (1971) developed his closed-loop theory In early learning, the learner uses KR in relation
of motor learning using a well-established set to the perceptual trace to make the movement
of empirical laws of motor learning, most of more precise, so that KR guides the movement
which were based on slow, linear-positioning to the target on successive trials. In such a view,
movements. He believed that the principles of KR does not produce learning directly. Rather, .
 The Learning Process 439

it creates the appropriate situation (i.e., being

on target) so that the actual learning processes
can operate. The movement's feedback produces
an increment in "strength" for the perceptual
trace.
We created the graphs in figure 13.5 to illustrate
the learning process in Adams' theory. In the early
stage of learning (figure 13.Sa, top graph), the
subject produces some correct movements but
produces many incorrect movements, too. Thus,
the movement feedback provides an increment to
a Correct trace
learning of the correct perceptual trace, but this
trace is based on other, incorrect traces as well.
In this stage of learning, performance is likely to
be inaccurate and variable because of the spread
..c:
of trace strengths among correct and incorrect ca:I c,
c:
perceptual traces. With the guidance of KR, the c.. ...
- Q)
Q) -Ul
learner produces more and more correct move- (J
... Q)
Q) (J
ments, which has the effect of strengthening the c. !ti
!:;
correct perceptual trace and reducing the relative
strength of incorrect perceptual traces, as illus-
trated in the middle and bottom graphs (figure
13.5, b and c). The reduction in relative strength b Correct trace
of these incorrect perceptual traces improves the
likelihood that the correct perceptual trace will
guide the limb to the goal position with increasing
frequency (i.e., less variability). 3
One of the interesting implications of Adams'
theory is that errors produced during the course ..c:
m c,
of training are harmful to learning. This is the :I c:
c.. ...
- Q)

case because when an error is made, the result- Q) ....

(J Ul
... Q)
ing feedback is necessarily different from the c.
Q) (J
!ti
!:;
feedback associated with a correct movement,
and thus will increment the strength of an incor-
rect perceptual trace. The relative strength of the
correct perceptual trace will be degraded a little
bit as well. One prediction, then, is that guid-
ance should be particularly useful as a training c Correct trace
method, as it prevents errors.
Adams also sought to explain how learners FIGURE 13.5 Adams' theory, represented as a growth
develop error-detection capabilities. He argued in perceptual trace strength. As skill develops (from top to
that after the movement was completed, the bottom panel}, the correct perceptual trace accumulates
individual could compare the feedback received proportionally more repetitions, and the shape of the distri-
bution becomes more peaked at the mode.
against the perceptual trace, the difference rep-
resenting the movement error that the person
could report to the experimenter or use as self- add to the perceptual trace, again without KR in
evaluation in the form of subjective reinforce- later learning.
ment. Presumably, this subjective reinforcement Contrary to earlier closed-loop theorists, Adams
could be used to keep the movement on target realized that in order for the system to have the
without KR; and, according to the theory, keep- capacity to detect its own errors, two memory
ing the movement on target can provide gains states must be present-one to produce the action
in learning because the feedback continues to and one to evaluate the outcome. What if the same
440 111 Motor Control and Learning

state that produced the movement also evaluated relation to a perceptual trace, then these animals
it? If the movement were chosen incorrectly, the should not have been able to produce the actions
feedback from the movement and the reference they did. Adams (1976b) has countered this
of correctness would always match, producing argument by saying that the animals may have
a report of no error on every attempt. In Adams' shifted to some other source of feedback, such as
theory, the movement is selected and initiated vision, to substitute for the lost sensations from
by another memory state that Adams called the the responding limbs. This may be the case for
memory trace-a "modest motor program" respon- some of these studies, but it does not apply to
sible for choosing the direction of the action, initi- all of them (e.g., Polit & Bizzi, 1978, 1979; Taub
ating it, and giving it a "shove" toward the target. & Berman, 1968). Also, Adams' theory does not
The perceptual trace then takes over the control of account for the data from various species showing
the movement to stop it at the final target location. the existence of central (spinal) pattern genera-
tors-structures apparently capable of generat-
Limitations and Contradictory ing complex actions without feedback from the
Evidence responding limbs (see chapter 6). The failure
to recognize the role of open-loop processes in
One characteristic of" good" theories is that there
movement control is a serious drawback for
should be no contradictions among the logically
Adams' theory.
derived predictions. Contradiction does appear
A second line of evidence against Adams'
to exist in Adams' theory, though, regarding
theory was provided by the literature on vari-
subjective reinforcement for slow positioning
ability of practice. Because the perceptual trace
movements. Adams viewed the perceptual trace
is the feedback representation of the correct
as providing (a) the basis for placing the limb at
action, making movements different from the
the correct target location and (b) a basis for the
correct action (in variable practice) should not
performer to detect how far that movement was
increase perceptual trace strength. Thus, Adams'
away from the target location after the movement
theory predicts that variability of practice should
has been completed. Schmidt (1975b) argued that,
be less effective for learning the criterion target
if the perceptual trace is used to position the limb,
than is practice at the target itself. In chapter 11
then no additional information can be available
we reviewed this literature and found no clear
about the amount of actual error produced. As
evidence that variable practice was less effective
discussed previously, Schmidt and Russell (1974;
than practice at the transfer target; and often the
Nicholson & Schmidt, 1991) provided evidence
evidence said that variability in practice was
that no error-detection mechanism exists after the
superior to practicing the transfer target itself
completion of slow positioning movements, even
(e.g., Shea &Kohl, 1991). Because Adams' theory
after 100 trials of practice, contrary to Adams'
explicitly claims that experience at the target
predictions. However, Schmidt and White (1972;
location is critical for the development of the
Nicholson & Schmidt, 1991) found strong error-
perceptual trace, this evidence is quite damaging
detection mechanisms after rapid movements, for
to his position.
which the perceptual trace presumably cannot
Lastly, the role of KR in Adams' theory was
be used to guide the limb during the movement.
to guide the learner to making the correct move-
Adams did not make a distinction between these
ment. However, as we reviewed in chapter 12,
fast and slow movements, yet the evidence shows
there is clear evidence that, when KR serves a
that they develop and use error-detection mecha-
guidance role, it has a degrading influence on
nisms very differently (e.g., Newell, 1976b).
learning, not the enhanced effect as would be
Certainly one of the most damaging lines of
predicted by Adams' theory.
evidence with respect to Adams' theory is the
work on deafferentation in animals (Taub, 1976)
and humans (Lashley, 1917), reviewed in chapter
Summary
6. Organisms deprived of all sensory feedback At the time Adams' theory was proposed, it rep-
from the limbs can move reasonably skillfully, resented a major step forward for motor learning,
and they can even learn new actions (e.g., Taub as it presented a plausible, empirically based
& Berman, 1968). If the only mechanism for theory for researchers to evaluate. We believe
controlling skilled actions involved feedback in that such evaluations have shown the theory to
 The Learning Process 441

have a number of limitations, as outlined here, ness, also one of the strengths of Adams' theory.
and that it no longer accounts for much of the According to schema theory, recall memory
currently available evidence on motor learning. is not thought to have an important role in slow
But the theory served its intended purpose. It positioning movements. The major problem for
generated substantial research and thinking, and the learner is the comparison between movement-
it paved the way for newer theories that account prod uced feedback and the reference of correct-
for the older data together with newer data. Thus, ness. In these movements, the recall state merely
it remains as a key legacy in the growth of motor pushes the limb along in small bursts, with the
learning research. person stopping when the movement-produced
feedback and the reference of correctness match.
Here, in these slow movements, the agent that
Schema Theory produces the action is the same as the agent that
evaluates it; hence no postmovement subjective
Largely because of dissatisfaction with Adams' reinforcement can exist as is the case for rapid
theory, Schmidt (1975b) formulated a theory movements. We have already presented evidence
that was considered a "rival" to Adams'. The that rapid movements do, and slow movements
primary concern with Adams' position was the do not, provide postmovement subjective rein-
lack of emphasis on open-loop control processes, forcement (Nicholson & Schmidt, 1991; Schmidt,
and the schema theory has a strong open-loop Christenson, & Rogers, 1975).
component. Yet, at the same time, many aspects
of Adams' theory are very appealing, such as Schema learning
the emphasis on subjective reinforcement, the
The schema concept is an old one in psychology,
concern for slow movements, and the need to
having been introduced by Head (1926) and
have one memory state that is responsible for
later popularized by Bartlett (1932). For these
producing the movement and another state that
researchers, the schema was an abstract memory
is responsible for evaluating it. Thus, schema
representation thought of as a rule, concept, or
theory borrowed heavily from Adams and others
generalization. Schmidt (1975b) attempted to use
by retaining the most effective parts and replac-
the basic idea of the schema (or rule) to form a
ing, changing, or eliminating defective ones. Also,
theory of how motor skills are learned.
the new theory was based heavily on knowledge At the heart of Schmidt's view of schema learn-
about motor control and used these concepts in ing is the idea that movements are made by first
conjunction with ideas about learning processes selecting a generalized motor program (GMP),
to attempt to explain the learning of both rapid structured with invariant features (such as rela-
and slower movements (see also Schmidt, 1980, tive timing), then adding parameters as required
2003).
in order to specify the particular way that the
program is to be executed for any one particular
Recall and Recognition Memory instance (see chapter 6 for details). After a GMP is
Schema theory holds that there are two states of selected and a movement is made by adding the
memory, a recall memory responsible for the pro- parameters, four types of information are avail-
duction of movement and a recognition memory able for brief storage in short-term memory: (1)
responsible for movement evaluation. For rapid, information about the initial conditions (bodily
ballistic movements, recall memory is involved positions, weight of thrown objects, and so on) that
with the motor programs and parameters, struc- existed before the movement was made; (2) the
tured in advance to carry out the movement with parameters assigned to the GMP, (3) augmented
minimal involvement from peripheral feedback. feedback about the outcome of the movement; and
Recognition memory, on the other hand, is respon- (4) the sensory consequences of the movement-
sible for evaluating the inherent feedback after the how the movement felt, looked, sounded, and so
movement is completed, thereby informing the on. These four sources of information are stored
subject about the amount and direction of errors. only long enough that the performer can abstract
Such structures satisfy the goal of having the two schemas. These abstract rules of how the
memory state that produces the action be different sources of information are interrelated are called
from the memory state that evaluates its correct- the recall and recognition schemas.
442 Motor Control and Learning

Recall Schema ditions of the movement, shown in figure 13.7.

The first of these relationships is termed the recall Here, the relationship between the parameters
schema because it is concerned with movement used and the outcome produced will depend on
production. Figure 13.6 represents the kind of pro- the nature of the initial conditions, such as dif-
cess that occurs, according to the recall-schema ferent objects to be thrown. These different initial
idea. On the horizontal axis are the outcomes in conditions are represented as different regression
the environment, such as the distance a ball trav- lines in figure 13.7.
eled after being thrown. On the vertical axis are How does the individual use the recall schema?
the parameters that an individual assigned to the On a future trial using this GMP, the person sets as
GMP. The co-occurrence of the parameter and the a goal the desired environmental outcome, labeled
movement outcome produces a data point on the point A on figure 13.7. Also, the particular initial
graph. With repeated movements using different conditions are noted (e.g., the weight of the object
parameters and producing different outcomes, to be thrown), which might fit into the category
other data points are established. As the number represented by line 2 in figure 13.7. Then, with
of throws accumulates, a relationship between the use of the relationship established by past experi-
size of the parameter and the nature of the move- ence, the rule is employed to select the parameter
ment outcome is established; this relationship is (labeled point B) that will come closest to accom-
represented by the regression line drawn through plishing that goal. The value of this parameter is
the points. 4 With each successive movement using then applied to the GMP to produce the action.
the program, a new data point is produced and Recognition Schema
the relationship is refined slightly. After each new The recognition schema, for movement evaluation,
movement, the various sources of information are is thought to be formed and used in a way similar
lost from working memory, so all that remains to the recall schema. Here the schema is composed
of the movement is the updated rule, termed the of the relationship between the initial conditions,
recall schema in LTM. the environmental outcomes, and the sensory con-
But this is not the entire story. The relationship
sequences. This relationship is represented as the
also includes information about the initial con-

....
Cl)
a;
E
f!
Cll
Cl..

A
Movement outcome
Movement outcome

FIGURE 13.7 The hypothetical relationship between

FIGURE 13.6 The hypothetical relationship between movement outcomes in the environment and the param-
movement outcomes in the environment and the param- eters that were used to produce them for various initial
eters used to produce them. conditions: the recall schema.
Human motor behavior: an introduction by R.A. Schmidt. Copyright 1982 by Human motor behavior: an introduction by R.A. Schmidt. Copyright 1982 by
TAYLOR & FRANCIS GROUP LLC - BOOKS. Reproduced with permission TAYLOR & FRANCIS GROUP LLC - BOOKS. Reproduced with permission
ofTAYLOR & FRANCIS GROUP LLC - BOOKS in the formatTextbook via of TAYLOR & FRANCIS GROUP LLC - BOOKS in the format Textbook via
Copyright Clearance Center. Copyright Clearance Center.
 The Learning Process 443

schema can occur because the location on the

horizontal axis will not be known. Similarly, if
sensory consequences are missing (e.g., as in
temporary deafferentation), then no recogni-
tion schema development can occur. In passive
~c: movements, no parameters are issued to the GMP
Q)
:l
g (indeed, no GMP is selected to be run off), so no
Ill
c: recall schema updating can occur.
0
(,) Also, note that, according to schema theory,
~c
Ill
there are positive benefits from the production of
c:
Q)
movements whether they are correct or not. This
en is so because the schema is the rule based on the
relationship among all stored elements, and this
relationship is present just as much for incorrect
movements as for correct ones. Adams' theory,
you may remember, views errors as disruptive,
A
as they degrade the relative strength of the correct
Movement outcome
perceptual trace (see figure 13.5).
FIGURE 13.8 The hypothetical relationship between Variability of Practice
movement outcomes in the environment and the sensory The theory predicts that practicing a variety of
consequences produced for various initial conditions: the
recognition schema. movement outcomes with the same program (i.e.,
Human motor behavior: an introduction by R.A. Schmidt. Copyright 1982 by by using a variety of parameters) will provide
TAYLOR & FRANCIS GROUP LLC - BOOKS. Reproduced with permission a widely based set of experiences upon which
of TAYLOR & FRANCIS GROUP LLC - BOOKS in the format Textbook via
Copyright Clearance Center. a rule or schema can form. When the range of
movement outcomes and parameters is small, all
the data points such as those shown in figure 13.6
three lines shown in figure 13.8. Before the move- are clustered in one place, and less certainty exists
ment, the individual selects a movement outcome about the placement of the line. 5 When a new
and determines the nature of the initial conditions. movement is required, greater error will occur
Then, with the recognition schema, the person can in estimating the proper parameters, expected
estimate the sensory consequences that will occur sensory consequences, or both. Shapiro and
if that movement outcome is produced. These, Schmidt (1982) found considerable evidence that
called the expected sensory consequences (labeled practice variability is a positive factor in motor
point C), serve as the basis for movement evalu- learning, especially for children (see "Variability
ation. The expected sensory consequences are of Practice" in chapter 11).
analogous to Adams' perceptual trace. Novel Movements
Some Predictions About Schema Schema theory also predicts that a particular
movement outcome (specified by a particular
Learning value of the parameter) need not have been pro-
The theory says that we acquire skills, at least in duced previously in order to be produced in the
part, by learning rules about the functioning of future. This is so because the basis for producing
our bodies-forming relationships between how a new movement is a rule about parameter selec-
our muscles are activated, what they actually do, tion based on the performance of earlier similar
and how those actions feel. Thus, movements for movements. Research has shown that, after varied
which any of the four stored sources of informa- practice, novel movements can be produced about
tion are missing will result in degraded learning as accurately as they can be if the novel movement
of the rules. One of the most critical of the sources had been practiced repeatedly, and sometimes
is movement outcome information (augmented more accurately (see the section on variability of
feedback, such as KR); if the person does not practice in chapter 11). This evidence suggests
receive augmented information about the move- that motor learning may be primarily rule learn-
ment outcome, then even if the other sources of ing and not the learning of specific movements.
information are present, no strengthening of the Such ideas have been used for a long time in
 g."t~~\~t~f"~b6~~ai6eo~. The. set shot is character-
.· ···· · ·ci~wer{eave the ground. The set shot
..•..• •· ... ·. ·..•.. ;~IY' ~asy:to
block •. However, it is the type
c:~~e~·;~~y,many.tirt1esinpractice, almost exclu-
estlo~·;~ .· ·. n, is• tfjis:.Do yea.rs of free throw practice
•p~rrafllete(. :·:opl'i~tefor.thefree throw distance must
~~·~ed,ictedbyscflematheory)? Or does this
. orl~~·freethiow?Or do b,oth processes .occur?
.~t~is;~~estion using skilled players from college
·ertorw~ct{'l~nythou5ands of free throws in practice
set S.tn:>ts at distances of 9 to 21 ft (the official free
a.ti on~ On the floc:)r 2 ft a.part. Several key predictions
•. lotions of sct:Tema theory and the principles of force
i~t, ~eli:tznfk,, Hawkins, Frank, &Quinn, 1979). First, a
;}he success of. the shot to the distance from the basket
~pa.rarnet~fized With increased levels of force as the
· •f6rce var:iablutyJIJ !he movement output and increasing
't'the accuracy between 9 and 21 ft would fall on or very
'Sio~ •1iiie•• because of the principles reviewed earlier in this
'oh

• 1inthe left side of figure 13.9, with the mean data points
l.:t~s ~~~port all butone of the predictions: The regression
·· •:; i.lfne nicely fits four.of t~e five data points. However, the
• • key finding, which: does not support schema theory,
· •occurred at the. distance of 15 ft-the foul line. From
. tnJs distance, the .players' success was considerably
:• • tii9her than predicted based on the schema regression.
{ ·~.similar effec:t of di§tance has been found in skilled
·1•.• l:>\i).$eballp~tchers---whowere much more accurate (than
r.pre~icted)attne regulation pitching distance compared
:·.:: \e>. gther clistanc0s; including distances just one foot
,·< Rlgseror f;:rrtherJro.rn. i:tome plate than the normal pitch-
''clistan~e (E)0.5 ft; Simons, Wilson, Wilson, & Theall,
} ..Bul,Wheilthese basketball skills were performed
.. \(ic.es<(~re~tiqeta.L, .2010)or by experts using jump
(wh~ret~e.feetdo leave the floor), all of the data
fitriicely on the regression line, as predicted by
· 1~~~e!llf1 theory (s~ethe rig.ht panel in figure 13.9). There
....·.~~$,·(lo ad\{an1f!Qe f()ra jump shot taken from the foul
·•· ''tilia: prE)~~lll~blyQecausejump shptsare.not practiced
freqO.E(rlt1x:a.ttt"te foul line than at any other loca-
.•. ~et shg~~·w~- .· . . .
.• ~hatdo these re.su.lts suggest? For the set shots,
: ·~grnethi~g~p~cificanq unique to the free throw distance
. ·;~~~ been.i~~rQ~d Jhroughrnany years of practice at this
· qi~tanqe: T;h.i~.speqific; capabiHty facilitated performance
·:~e t~ken ~tan ahgle straight cm to the basket, making
i~ting Witllinaclass of basketball set-shot skills. Note

444111
 The Learning Process 445

that this particular finding is not consistent with the schema view, as the theory gives no specific
preference for any particular parameterization, regardless of the number of specific practice instances
that have been executed. This study illustrates many questions of practical interest that have been
discussed at a theoretical level in this and previous chapters, such as force-variability principles,
variable practice, schema theory, and specificity of learning. ·

movement-education situations with children, that the evidence strongly supports the GMP
where the pupils are presumably developing a set view (see chapter 6), the theory is mute in terms
of rules or schemas about their motor behaviors of how the program is formed in the first place,
and consequently being helped to be more profi- and this deficiency is readily acknowledged as a
cient performers in novel situations in the future major problem with the theory (Schmidt, 2003).
(Nicholson & Schmidt, 1991; Schmidt, 1976b, 1977). The following sections highlight other limitations.
Error Detection Knowledge-of-Results Frequency
Schema theory predicts that there should be no Strengthening of the schema depends on the
capability for error detection after a slow move- subject's knowledge of the movement outcome,
ment, whereas such capability should exist after a so higher levels of KR relative frequency would
rapid movement. This is the case because the error- be expected to enhance schema learning as
detection capability is actually used to produce the compared to lower levels. When relative KR fre-
slow movement, leaving behind no capability with quency effects on overall learning were evaluated
which to detect errors. Based on the information in chapter 12, the results appeared to contradict
about closed-loop processes presented in chapter this schema theory prediction because reduced
5, if the movement was rapid (as was the case in the frequencies either had no effect on learning or in
Schmidt & White, 1972, study), the subject would some cases enhanced it, rather than degrading
compare the feedback from the movement to the it, especially so if the KR was "faded" over trials
reference of correctness to define an error after the (Sullivan, Kantak, & Burtner, 2008; Winstein &
movement was completed. The error-detection Schmidt, 1990). These findings are further com-
process is not responsible for producing the action, plicated by KR variables that appear to influence
and it evaluates the correctness of the action only the learning of parameters and invariances in
after the movement has been completed. For rea- different ways, which is also contrary to schema
sons discussed before, there is insufficient time theory (Shea & Wulf, 2005).
for the performer to take in the feedback, evalu- Contextual Interference
ate it, and make corrections before the movement and Cognitive Operations
is completed. According to the theory, the recall
One key prediction of schema theory is that vari-
schema is thought to produce the movements, and
able practice would result in stronger rule learn-
the recognition schema is responsible for compar-
ing than nonvariable (or constant) practice, and
ing the movement-produced feedback with the
evidence supports that general prediction (but
learned reference of correctness for evaluating the
note the discussion in "Especial Skills"). How-
movement afterward. As mentioned during the
ever, schema theory makes no prediction about
discussion of Adams' theory, empirical evidence how the variable practice should be scheduled.
supports this prediction (Schmidt & Russell, 1974; Recall from the discussion of random versus
Schmidt & White, 1972). blocked practice effects in chapter 11 that the
amount of variable practice in these studies was
limitations and logical Problems equal. Because they share a common breadth of
The emphasis of schema theory on the GMP practice variability, schema theory fails to pre-
concept represents both a major strength and dict the learning differences that occur between
major limitation of the theory. While we believe random and blocked practice (Lee, Magill, &
446 !!I Motor Control and Learning

Weeks, 1985). The explanations for contextual- that it makes testable predictions that can be
interference effects stress the importance of cogni- falsified by experimental testing. Nevertheless,
tive operations during practice, which highlights the theoretical perspectives that we consider
a more general limitation of schema theory, as the next represent important advances in furthering
theory provides no rationale for learning effects our understanding of the complex interaction of
due to cognitive operations such as imagery, processes involved in motor learning.
mental practice, and observational learning (Shea
& Wulf, 2005; Sherwood & Lee, 2003). Cognitive Perspectives
Summary Although both closed-loop theory (Adams, 1971)
and schema theory (Schmidt, 1975) emphasized
Schema theory has provided an alternative to the role of memory structures in skill, the learn-
Adams' closed-loop theory of motor learning. ing process depended on movement repetition
Compared to Adams' theory, it has the advantage and feedback. The development of the perceptual
that it accounts for more kinds of movements; it trace in Adams' theory and the recall and recog-
seems to account for error-detection capabilities nition schema in Schmidt's theory were mecha-
more effectively and seems to explain the produc- nistic processes. Research conducted since these
tion of novel movements in open-skills situations. theories were published, on learning variables
Some logical problems need to be solved, and it is such as the contextual-interference effect (chap-
not clear that this can be done without discarding ter 11) and various augmented-feedback effects
the entire theoretical structure (Petrynski, 2003; (chapter 12), suggests that the role of cognitive
Schmidt, 2003; Shea & Wulf, 2005). There are some processes in learning might be more complex
apparent failures of the evidence to agree with the than originally conceptualized in Adams' and
theoretical predictions as well (e.g., Klein, Levy, Schmidt's theories.
& McCabe, 1984; Keetch et al., 2005). While the Both of the major hypotheses regarding the
theory was a step forward, it should be clear that contextual-interference effect suggest that cogni-
it does not provide a complete understanding of tive processes play a key role. In the elaboration
the data on motor learning. Even so, the theory hypothesis, explicit contrasts and comparisons
provides a useful framework for thinking about of the tasks to be learned were thought to
skill learning because it is consistent with the benefit learning. In the reconstruction view, it
literature on the GMP. was the process of planning a different action
to be performed that boosted learning. In both
Differing Theoretical hypotheses, learning was more effective if the
elaboration or reconstruction was made more
Perspectives difficult (i.e., in random practice), suggesting
of Motor Learning that the effort with which the cognitive processes
were undertaken had a critical impact on learn-
The ideas presented next are probably best ing (Lee, Swinnen, & Serrien, 1994; Sherwood &
described as hypotheses about the learning of Lee, 2003; Vickers, Livingston, Umeris-Bohnert,
motor skills. They really do not satisfy the basic & Holden, 1999).
criteria for consideration as theories for a number This cognitive emphasis suggested that some-
of reasons. First, many of them are directed at thing more was occurring during learning than
only certain kinds of tasks, such as continu- executing movements and receiving inherent
ous tasks, positioning tasks, and tracking; and feedback, which could not explain the differences
more generality is usually required for a theory. between random and blocked practice. In both
As well, some of these theoretical perspectives practice conditions the subjects received the same
concern only a few experimental variables, and amount of practice on the same tasks. The effects
theories (such as closed-loop theory and schema on learning of augmented-feedback variables
theory) are usually thought to have more com- such as concurrent feedback (chapter 12) also
plete structures that are capable of explaining fit a cognitive perspective well. Movements that
the effects of a variety of independent variables. are produced with the assistance of concurrent
Further, an important ingredient of a theory is feedback experience the same efferent commands
 The Learning Process 447

and intrinsic feedback as nonguided move- what cognitive processes are being challenged
ments. However, as argued in the KR-guidance and how these processes change over the course
hypothesis (Salmoni, Schmidt, & Walter, 1984; of learning. Nevertheless, it is clear from the many
Schmidt, 1991), such variables tend to minimize studies conducted in the past several decades that
the learner's need for evaluation of subjective the effects on learning of practice and augmented-
information and other cognitive operations that feedback variables are much more complex than
are ordinarily undertaken in the preparation for at first believed. The challenge-point framework
the next trial. represents an attempt to characterize the com-
The effects of these practice variables on the plexity of these relationships within a cognitive
learning process are complex, however-they perspective. Some empirical explorations of the
appear to have a strong dependence on the nature framework support its explanatory power, espe-
of the task and the experience level of the subject cially with individuals who have a compromised
(Wulf & Shea, 2002). Some of these complexities motor system (Lin, Sullivan, Wu, Kantak, & Win-
were conceptualized in a theoretical framework stein, 2007; Onla-or & Winstein, 2008; Sullivan et
by Guadagnoli and Lee (2004), who suggested al., 2008).
that cognitive processing during practice is
affected by the degree to which the subject is Hierarchical Control Perspectives
challenged during the practice period. The nature
of the task, the conditions of practice, and the As people learn, at least with some tasks, a
experience level of the learner interact to deter- change occurs such that motor control is shifted
mine the amount of challenge present during to progressively "lower" levels in the nervous
acquisition trials. For example, random practice system. The idea that motor behavior is hier-
is considered more challenging than blocked archical means that some "higher" level in the
practice and therefore should benefit learning. system is responsible for decision making and
But, driving on a busy highway would be more some "lower" level is responsible for carrying
challenging than driving in a deserted parking out the decisions. With respect to the information-
lot-and certainly much more challenging for processing analysis, the decision-making pro-
the novice learner than the semiskilled driver. cesses of the system are considered to be at a
The framework suggests that variables such as "higher" level in the hierarchy than the motor
random practice and concurrent feedback are programming level. The hierarchical control
effective to the degree that they challenge the perspective suggests that with practice, control
cognitive processes of the learner. The frame- is shifted from the "higher" to the "lower" levels
work suggests, however, that there exists a point in the system.
at which these cognitively challenging practice A good example that demonstrates research
conditions may not be needed. Indeed, they may in this perspective was a study by Pew (1966),
be detrimental to learning if used for tasks that are who used a tracking task in which the subject
already inherently challenging. Similarly, learn- controlled the movement of a dot on a monitor by
ing may be sufficiently challenged in individuals pressing one or the other of two buttons. Press-
whose performance capabilities are put to the test ing the right button caused the dot to accelerate
merely by the demands of the task. In such cases, to the right, and the acceleration could be halted
nonchallenging practice conditions (e.g., blocked and reversed by pressing the left button, which
practice or concurrent KR) would be expected to caused the dot to accelerate to the left. If no button
facilitate, rather than be detrimental to learning. was pressed, the dot accelerated off the screen
Although the Guadagnoli and Lee frame- in one direction or another. The subject's task
work has provided an explanation for some of was to keep the dot in the center (this is called a
the complex relationships for learner, task, and compensatory tracking task).
practice variables that exist in the literature, it has A record from one of the subjects, with the
limitations. Certainly, the concept of task difficulty, velocity and position of the dot shown for early
though frequently discussed as an important and late practice, is presented in figure 13.10.
factor in the motor learning literature, remains In early practice (figure 13.lOa), the subject was
a construct with an elusive definition. As well, making about three button presses per second,
the framework stops short of identifying exactly and the dot was never positioned near the center
448 mi Motor Control and Learning

of the screen for very long. In this mode of control presses. With some subjects, the durations of the
the subject pressed the button, waited for the right and left buttons were adjusted, perhaps
visual feedback from the screen, decided that the as a kind of "parameter" of the programmed
dot was accelerating off the screen, then planned a activities.
movement to reverse it, pressed the other button, It is easy to see the advantages of shifting the
and so on. Here, the subject is using the executive control from the decision-making level to the
(e.g., the information-processing stages) level motor program level. Foremost is the freeing of
predominantly, so that the "highest" level in the the attentional mechanisms for use on higher-
system is consistently involved in the production order aspects of the task (e.g., strategy), for doing
of every movement. other simultaneous tasks, or for simply resting
Compare figure 13.lOa to figure 13.lOb, which so that the organism does not become fatigued.
is from the same subject but later in practice. Here This freeing of attention is one of the major events
the motor behavior is quite different. First, the rate that occur when people learn, and it is discussed
of responding is much faster, about eight move- further in later sections of this chapter.
ments per second. Next, the dot is much closer
to the target because the button was pressed to Progression-Regression Hypothesis
reverse the direction of the dot before the dot Of particular relevance to tracking tasks is a
was very far away from the target. Although we hypothesis presented by Fitts, Bahrick, Noble,
cannot be absolutely certain, the mode of control and Briggs (1959) about how changes in motor
appears to have changed. It appears that now behavior occur with practice. In many tracking
a long string of movements is prestructured as tasks, both in the laboratory and in the outside
a unit, perhaps governed by a motor program. world, the movements of the track to be followed
Thus, a separate decision from the executive level are made up of a number of components that can
is no longer required for the control of each button be described according to the physical principles
press. Pew viewed this finding as evidence for the of motion. At the simplest level is the position of
hypothesis that with practice, the subject shifted the track at any moment. The next most complex
the control from an executive-based level to the aspect of the track is its velocity at any moment.
lower-level control of the motor program, free- A third and yet more complex aspect of the track
ing the decision mechanism for other activities is its acceleration at any moment. In design-
and making the movement more effective. Now, ing servo systems to regulate some mechanical
instead of controlling every button press, the system, engineers can devise a simple system
executive level was controlling groups of button that responds to (a) only the position of the track,

Early in practice Late in practice

~ -38 ~ -38
~ ~
~ ~
~ ~
"g "g
~ +38 ~ +38
Time Time

+5 +5
e
~
e
~
c: c:
0 0
:;::: :;:::
"iii "iii
0 0
a. a.
-5 -5
a Time b Time

FIGURE 13.10 Performance records from a button-press tracking task in (a) early and (b) late practice. (Top records show
instantaneous velocity, and bottom records show position with target represented as zero; responding is more rapid and more
accurate in later practice.)
Reprinted, by permission, from R.W. Pew, 1966, 'i'\cquisition of hierarchical control over the temporal organization of a skill;' Journal of Experimental Psychology 71: 768.
Copyright© 1966 by the American Psychological Association.
 The Learning Process 449

(b) the position and velocity, or (c) the position, understanding of the hierarchical nature of the
velocity, and acceleration. With each increase in underlying changes in motor control when skill
the number of components being tracked, pro- is achieved with practice or reduced under stress
gressive increases are required in the complexity or with forgetting.
(and expense) of the mechanical or electronic
devices that are to track them. Making Movements Automatic
The progression-regression hypothesis for humans For more than a century (e.g., James, 1890), the
presented by Fitts and colleagues (1959) holds idea of automaticity has been that, as a by-product
that, when the learner practices a tracking task, of learning, skilled performers become able to
a progression develops in the learner's behavior perform with minimal attention cost and minimal
in the direction of acting more and more like a interference from other cognitive information-
complex tracking system. Early in practice, the processing activities. In chapter 4, we qualified this
person responds only to the simplest elements of basic idea considerably, saying that "automatic"
the display (position). With increased practice, the responding appears to involve a lack of interfer-
learner becomes able to use velocity information, ence with respect to particular secondary tasks;
and even later comes to use information about the notion that a given task is interference free
acceleration as well. The regression portion of the for all secondary tasks is not supported by the
hypothesis refers to what happens to the learner evidence (Neumann, 1987). However, it does seem
under stressful conditions or when forgetting of likely that the interference from many simultane-
the movement has occurred (perhaps as a result ous cognitive information-processing activities is
of a long layoff). According to the hypothesis, decreased, or even eliminated, with practice, thus
the person regresses to a simpler level of control freeing the individual to engage in other higher-
(from acceleration to velocity, or from velocity to order aspects of the task-such as planning strate-
position), with systematically reduced tracking gies in tennis or race car driving, or projecting an
performance as a consequence. Thus, the hier- affective emotional style in acting, music, or dance.
archical nature of learning involves progressing Automaticity can be considered in essentially
to levels of more complex information, although two ways. First, and most common, is the idea
performance effects may show reversals in the that specialized information-processing struc-
shift between levels in the hierarchy (regressing tures are learned with practice and that they
to less complex information). handle portions of the processing requirements
A number of experimenters have studied learn- of the overall task, such as feature detection and
ing in tracking tasks with respect to the progression- movement selection (e.g., Logan, 1988; Neumann,
regression hypothesis. Fuchs (1962) found that 1987; Schneider & Fisk, 1983; Schneider & Shiffrin,
the role of position cues in tracking decreased 1977; Shiffrin & Schneider, 1977). Each process can
with practice, while the role of acceleration cues occur when the appropriate stimulus conditions
increased; and these effects were reversed when are presented and essentially triggered into action
a secondary task was added to induce stress (see without awareness; indeed, sometimes the pro-
also Garvey, 1960). More recently, researchers cess cannot even be prevented (Schneider, 1985).
have improved on the methods used in the earlier By handling information processing in this way,
work and have provided additional evidence for the performer decreases the interference with
a shift in movement control consistent with the other cognitive activities that compete for the
hypothesis (Hah & Jagacinski, 1994; Jagacinski & same common resources. If, in a given task, all
Hah, 1988; Marteniuk & Romanow, 1983). these processes can be so acquired, then the task
At least for tracking tasks, learners appear to can be thought of as" automatic," in the sense that
respond to systematically different aspects of the the entire movement can occur without interfer-
track with practice and to reverse these trends ence from particular groups of secondary tasks.
with stress. We should be careful not to go too In this view, the organism does not decrease the
far with these conclusions, because we have amount of environmental information processing
no independent way of knowing exactly which that must be accomplished; rather, it processes
stimuli were being used here. But the evidence this information differently-via specialized
is certainly consistent with the progression- structures-and more quickly and with less inter-
regression hypothesis, and it contributes to an ference from other simultaneous tasks.
450 ill Motor Control and Learning

However, another view is possible (see outcomes tend to become more consistent, pre-
Schmidt, 1987). In at least some kinds of tasks dictable, and certain with experience. Recall that
(e.g., predictable and stereotyped), a major variable error and other measures of variability
process of learning appears to be a shift from (chapter 2) were devised to capture this aspect of
high-level conscious control to a lower-level motor behavior. In the study of these phenomena,
programming control, as discussed in a previous the patterns of movement are measured by vari-
section. With predictable tasks, the regularities ous kinematic procedures (e.g., video analysis,
of the environmental information can be learned position-time records, computer simulation).
and therefore can be anticipated during per- Changes in the trial-to-trial consistency of these
formance. If so, then the person does not have measures have been noted in an impressive
to process this information directly, but rather variety of tasks, such as driving (Lewis, 1956),
preprograms long sequences of action based on throwing (Stimpel, 1933), handwheel crank-
the prediction of the environmental information. ing (Glencross, 1973), table tennis (Tyldesley &
Musicians "memorize" sheet music so that they Whiting, 1975), tracking (Darling & Cooke, 1987;
are not dependent on it, and experienced driv- Franks & Wilberg, 1984), keyboarding (Salthouse,
ers no longer have to watch their feet as they 1986), bimanual coordination (Lee, Swinnen,
move from accelerator to brake in the car. Thus, & Verschueren, 1995), and many others. Such
being able to avoid processing environmental generalizations perhaps seem to be particularly
information frees those (conscious) information- appropriate with respect to the acquisition of
processing activities for other tasks and makes closed skills, which have as a major goal the
the task appear" automatic," at least with respect production of a consistent action in a stable (or
to particular kinds of activities. In this view, the predictable) environment-the kinds of skills for
person does not necessarily process information which stable motor programs are most highly
any more effectively or faster, but rather learns to suited. These changes in movement pattern
avoid having to process information by shifting consistency probably represent some of the most
to motor programming modes of control. persistent phenomena in the motor learning
Of course, it could be that both of these view- area.
points are correct but that each is relevant for a The Acquisition of Sequencing: The Gearshift
different class of movement tasks (e.g., those that Analogy Another hierarchical change in move-
have predictable and those that have unpredict- ment control with practice involves the ways
able environmental information). Even within in which movements are sequenced. MacKay
a particular skill, one can imagine specialized (1976, personal communication) suggested that
structures for detecting environmental informa- motor programs might be generated by stringing
tion; then, sequences of preprogrammed output together smaller programmed units of behavior
could be generated that reduce the reliance so that eventually this string of behavior is con-
on such information for the next few hundred trollable as a single unit-such as in learning to
milliseconds. Both viewpoints provide ways of shift gears in a car. 6 As you may remember, the
conceptualizing the acquisition of automaticity act of shifting gears when you were first learning
in high-level motor learning, and they present was a slow, jerky, step-by-step process; you lifted
interesting issues for research in motor skills. your foot from the accelerator, then depressed the
Creating Motor Programs clutch, and then moved the shift lever (probably
in three distinct movements as well), until the
Of course, we know that many changes occur in entire act was completed (or until the car rolled
our movements when they are subjected to prac- to a stop going up a hill). Contrast this behavior
tice, with actions tending to become more consis- to that of a race car driver, who appears to shift
tent, smoother, less effortful, and more routine or gears in a single, rapid action. Not only does the
automatic with experience. These are all powerful movement occur much more quickly, but also
changes, and in the next sections we consider some the elements of the action are performed with
of the experimental evidence for them. precise timing, and the actions of the hands and
The Acquisition of Movement Pattern Con- both feet are coordinated in relatively complex
sistency One important change in movement ways. In relation to the behavior of the early
behavior with practice is that the movement learner, the action seems to be controlled in a very
 The Learning Process 451

different way, perhaps as a single programmed trial to trial would be greater than the variability
unit. from the end of element 3 to the beginning of ele-
MacKay suggested that the various elements ment 4. This is the case because the first two ele-
are learned in a progressive way to form the ments (2 and 3) are in different units (controlled by
entire action. Figure 13.11 is a diagram of how different programs), while the latter two (3 and 4)
this might work. Assume there are seven ele- are supposedly controlled by the same program.
ments in the entire sequence and that these are at Turning this logic around, if we found intervals
first controlled one at a time, each by a separate in the sequence in which temporal variability was
motor program. With some practice, the first two very high, this could be taken as evidence that the
elements might come to be controlled as a single behaviors occurring at the opposite ends of this
unit; the next three elements could compose interval of time are members of different motor
another unit, and the last two could compose a programs. This is similar to the method used by
third. Finally, with considerable experience, the Young and Schmidt (1990, 1991) and Schmidt,
entire sequence might be controlled as a single Heuer, Ghodsian, & Young (1998) to investigate
unit. This view is of a type of hierarchical control the acquisition of new bimanual coordination
in that it specifies how the program is structured programs (see also chapter 8).
from the beginning, progressively growing in Combinations of Reflexes Another way that
length by adding parts. Other possibilities exist motor programs are thought to be formed in prac-
as well (Marteniuk & Romanow, 1983). tice is through the combination of fundamental
We should be able to see evidence of the reflexes (Easton, 1972, 1978). According to this
changes in these structures by using a fundamen- viewpoint, higher levels in the motor system are
tal principle of variability: The variability (incon- capable of tuning or adjusting lower spinal levels
sistency) of the elements within a unit should be so that the existing reflexes (e.g., the stretch reflex)
considerably smaller than the variability between can be controlled in ways that result in skilled
units. In figure 13.11 (middle practice), if we were actions. Thus, rather than hypothesizing that
to measure the interval from the end of element the motor system builds a set of commands that
2 to the beginning of element 3, the relative vari- come to exist as a stored motor program, Easton
ability of this interval (expressed as the SD of the held that the" commands" are really ways of con-
interval divided by the mean interval length) from trolling the preexisting reflexes. Such emphases

>·•' ~};.. .. :;;· C· . so1k1~veltc>:r1g~r, :<: . ·. . : .<P ·• ,·:<·.

>;,. ·.. · ·:sii1tt1~vi;.fo~~ra :,: ... o:,'. ~·~;
~:~,, ,; o; ~; >

FIGURE 13.11 The gearshift analogy. (Initially, seven elements are each controlled by separate programs; later, they
become grouped so that they are organized into a few units or even into a single unit.)
Adapted from MacKay. 1976. personal communication.
452 Motor Control and Learning

on reflexes are also seen in the views of Fukuda with more independence or with a different
(1961) and Hellebrandt, Houtz, Partridge, and dependency. In the earlier example of throwing
Walters (1956), as discussed in chapter 5; but with the nondominant arm, one possible strat-
here, while the reflexes are thought to be of assis- egy for the performer would be to fix or limit
tance to the overall programmed action when the degrees of freedom involved in the act. This
increased force or speed is required, they are not is done in order to reduce the contribution of
the fundamental basis of it. But Easton's view- their independent variability, and hence reduce
point also has a great deal in common with the the complexity of the action. In this way, the
ideas of Greene (1972), Turvey (1977), and others, performer can achieve a relatively crude level
all of whom argue that rapid movements consist of success at the task by reducing the number
of controlling structures that are constrained to of ways in which things can go wrong. In a nice
act as a single unit, perhaps by tuning of spinal demonstration of this stage of learning, Southard
systems or by utilization of reflexes. These ideas and Higgins (1987) showed that people who
share much in common with the theoretical learned a racquetball shot initially restricted the
perspectives of the Russian physicist Bernstein motion of the elbow and wrist and performed
(seep. 8), whose ideas on learning are presented the shot with a more whole-arm action. A similar
next. set of findings was reported by Hodges, Hayes,
Horn, and Williams (2005) over 10 days of prac-
The Bernstein Perspective tice in a soccer-chip task. Participants gradually
Suppose you were asked to throw a ball over- achieved more initial successes at the task as the
hand with your nondominant arm. In all like- range of motion at the hip was reduced with
lihood, your performance would look quite practice.
clumsy-the movements of the body would lack Stage 2: Releasing and Reorganizing
the fluid motion that characterizes a throwing Degrees of Freedom
motion with the dominant arm. The motions
of the nondominant arm would probably be According to Bernstein's arguments, in stage
described as much more fixed and restricted in 2 of learning, the constraints on the degrees of
the range of motion. This example characterizes freedom are loosened, allowing both for greater
an important concept of learning initiated by independent motion and for a higher level of
the work of Bernstein (1967). The concept is that success. Bernstein's introspections about the
learning involves a process of solving the degrees learning process have received support in a
of freedom problem-discovering ways in which study by Vereijken, van Emmerik, Whiting, and
the independent parts of the moving body can Newell (1992), for example. Subjects practiced
be organized in order to achieve a task goal. In a ski simulator task that required whole-body
Bernstein's perspective, learning occurred in movements to move a platform from side to
three stages, although the stages that he pro- side over curved metal rails (see figure 2.9, p.
posed were different than the stages discussed 36). The sides of the platform were anchored to a
earlier in the chapter (Adams, 1971; Fitts, 1964). frame by rubber springs; when the springs were
According to Bernstein (1967), the stages oflearn- subjected to force from the "skier," the platform
ing involve (a) freezing degrees of freedom, (b) would oscillate from one side of the frame to
releasing and reorganizing degrees of freedom, the other. Subjects practiced this task over seven
and (c) exploiting the mechanical and inertial days, attempting to produce large-amplitude
properties of the body. displacements of the platform.
Changes in some of the kinematic measures
Stage 1: Freezing Degrees of Freedom of performance are presented in figure 13.12. In
Bernstein suggested that early in practice the this figure we can see that in the pretest, the plat-
learner attempts to "freeze" as many of the form movements were made with quite restricted
degrees of freedom as possible, allowing as angular movements of the hip, knee, and ankle.
few as possible of the body parts to move At this early stage of learning, subjects seemed
independently. With practice, more and more to "freeze" the range of motion of the lower limb
of the degrees of freedom are "thawed out"- and trunk, perhaps just to get any movement of
individual body parts appear to move either the platform at all. By the end of the first day, the
 The Learning Process 453

60 Second, in learning to do a handstand on the still

rings, the learner comes to freeze the degrees of
50
freedom in the knees, hips, and trunk, so that
(j) balance control is ultimately achieved primar-
~
Cl
ily by movements of the wrists. Unfortunately,
~
CIJ 40 very few studies have addressed this stage of
CIJ Bernstein's perspective on learning, revealing a
Ci
r::
111 30
gap in empirical evidence.
'E Overall, we think that these ideas (about freez-
:§. ing and freeing degrees of freedom) form a useful
15CIJ 20 description of learning's effect on movement con-
Cl
r:: trol in some tasks. However, the counter-examples
111
a: described here indicate that this does not provide
10
a universal account of all motor learning.
Another key concept suggested by Bernstein
0
Pretest 1 2 3 4 5 6 7
was that independent degrees of freedom are
Practice day (last trial per day) assembled into functional units that act together.
When two or more independently moving
degrees of freedom combine" to perform as
/1

FIGURE 13.12 Changes in the frequency and amplitude

of movements during practice of the ski simulator task. one functional movement, the independent parts
are said to be coupled-they act as coordinative
structures (or functional synergies) to coordinate
range of motion of each of the joints had been the independent parts to work as if they were a
extended considerably, resulting in much larger single unit. This is perhaps exemplified by the
amplitudes of the platform but with reduced fre- gearshifting concepts presented earlier in this
quency. By the end of the seventh day, oscillation chapter (see figure 13.11).
frequency had increased dramatically, along with Another good example of coupling indepen-
further increases in amplitudes and joint ranges dent degrees of freedom occurs when you try
of motion. Thus, greater success in displace- rubbing your stomach while patting your head.
ment and frequency of platform oscillations was In this task you are asking the two limbs to per-
achieved with greater range in the motion of the form two different actions. How does your motor
lower limbs and trunk-supporting Bernstein's system deal with each limb when performing this
suggestion that practice results in a release of task? Most likely, if you have never practiced the
the degrees of freedom (Vereijken, Whiting, & task before, you will find it difficult because of the
Beek, 1992; see also Arutyunyan, Gurfinkel, & strong tendency to perform similar actions with
Mirskii, 1968, 1969; Newell, van Emmerik, & each limb-either patting both the head and the
Sprague, 1993). stomach or rubbing both (chapter 8). However,
Note, however, that the concept of releasing research suggests that with practice, you can
degrees of freedom has not received universal overcome the tendency to couple these parts as
acceptance among researchers. For example, a a coordinative structure.
study of violinists by Konczak, vander Velden, A series of studies that demonstrated the
and Jaeger (2009) revealed that shoulder motions effects of practice on bimanual coordination
of the bowing arm were actually reduced as a was conducted by Walter and Swinnen (1990,
function of practice, indicating that the degrees 1992, 1994; Swinnen, Walter, Lee, & Serrien, 1993;
of freedom underwent freezing, not freeing, Swinnen, Walter, Pauwels, Meugens, & Beirinckx,
as a function of practice. We can hypothesize 1990). Subjects were asked to initiate rapid, dis-
at least two other exceptions. First, in learning crete actions of the left and right limbs simultane-
to windsurf, the performer gradually learns to ously. The left arm moved a lever toward the body
freeze the degrees of freedom in the knees and with a single, rapid elbow flexion movement. The
hips, so that most of the controlling actions are right arm also moved a lever toward the body.
in the shoulders and arms-actions used to However, midway through the movement the
manipulate the sail's orientation to the wind. right arm was required to reverse the direction of
454 111 Motor Control and Learning

its movement twice. Thus, the subject's task was Stage 3: Exploiting the Mechanical-
to produce a unidirectional, flexion movement Inertial Properties of the Limbs
of the left arm and a flexion-extension-flexion The final stage of Bernstein's perspective on
movement of the right arm, the arms starting learning is the alteration of movement control so
and moving simultaneously (figure 8.9, p. 275). that the motor system can take advantage of (or
As with the task of rubbing your stomach exploit) the built-in mechanical-inertial proper-
while patting your head, Walter and Swinnen ties of the limbs. This notion is tied strongly to
found that subjects tended to perform similar
mass-spring control, discussed in chapter 7, in
actions with the two limbs: There was a less pro-
which certain spring-like properties of the limb
nounced reversal for the limb that the subject
system can be used to the performer's advantage
intended to reverse, and there was evidence of
to reduce the need for complex computations and
a reversal in the limb that the subject did not
information processing, to reduce energy costs, or
intend to reverse (Swinnen, Walter, & Shapiro,
to make the movement faster and more forceful.
1988). Thus, neither of the limb movements was
Schneider, Zernicke, Schmidt, and Hart (1989)
performed as intended. Rather the functional
unit was of two limbs performing similar, albeit studied this question using film analysis of rapid
three-joint arm movements together with a bio-
hybrid, actions of the individual goals; this was
seen via high (within-subject) correlations among mechanical analysis that allowed the estimation
the kinematics of the two limbs. Leaming was of torques in each of the participating joints. Near
viewed in terms of the success with which each the middle of this maximum-speed movement,
limb performed its own goal with practice. Leam- the subject was to reverse his hand movement at
ing was enhanced if it was supplemented with a target, at which the arm was briefly extended
augmented feedback (Swinnen et al., 1990, 1993), upward at about 45°. Early in practice, the sub-
as we would expect from our understanding of jects tended to use a shoulder flexion torque at
feedback effects in chapter 12. the target (reversal point), as if they were holding
Learning was also enhanced under adapted their arm up against gravity. But later in practice,
conditions, whereby the actions were performed the shoulder flexion torque tended to drop out,
slowly at first and then were gradually increased to be replaced by an extension torque. Now the
in speed (Walter & Swinnen, 1992). Related limb appeared to be "thrown" at the target, to
findings have been provided in experiments be "caught" by the shoulder extensors in order
involving the acquisition of handwriting skills to reverse its direction and bring it back down
(Newell & van Emmerik, 1989) and dart throwing quickly. Certainly, the structure of the GMP had
(McDonald, van Emmerik, & Newell, 1989), as changed markedly across practice, employing
well as in bimanual aiming tasks involving asym- systematically different muscle groups for essen-
metric amplitudes (Sherwood, 1990; Sherwood & tially the same set of positions early and late
Canabal, 1988). in practice. There were many other changes in
One interpretation of these findings is that, movement trajectories and in the forces produced
by overcoming the existing coupling of degrees as well (see also Spencer & Thelen, 1999).
of freedom, the limbs are somehow uncoupled A different type of analysis was performed
in a way that allows them to move more or less by Gray, Watts, Debicki, and Hore (2006), who
independently. This interpretation has some reported on the differences between the baseball
controversy, however, as others have shown that throwing motions of the dominant (skilled) and
learning a new bimanual pattern actually results nondominant (unskilled) arms. Their analyses
in an increased dependence between the hands: revealed evidence of restricted ranges of motion
that learning results in the development of new for the movements of the unskilled arm, pro-
bimanual GMPs involving tight, complex link- viding additional support for Bernstein's stage
ages between the limbs (Schmidt et al., 1998). This 1 (freezing degrees of freedom). However, one
controversy is complicated by the evidence that of the strengths of Gray and colleagues' (2006)
the duration of the movement task has a large study was the comparison of the unskilled arm
role in the control and learning of coordinated mechanics and the mechanics of the skilled arm,
actions (chapter 8). More research is needed to thereby revealing the effects of practice and learn-
address these issues. ing. Their analyses showed that, compared to the
 The Learning Process 455

unskilled arm, the skilled arm tended to exploit would forever be locked into performing only
interaction torques at various joints to achieve these two patterns. But research suggests that
higher arm velocities. new patterns can be learned, leading to a number
The interpretation of these studies was that the of important issues about the learning process
motor system learned to use various passive iner- as a consequence (Swinnen & Wenderoth, 2004).
tial properties of the system, and that the benefit Recall from chapter 2 that studies of motor
could be realized not only in terms of increased learning often devised experimental tasks that
speed but also in terms of decreased energy were as unique as possible to the learner. The
costs. These studies clearly support Bernstein's reasons for this were straightforward-if sub-
hypothesis and reveal many interesting changes jects came into the laboratory with skills already
that occur in the movement control processes learned (e.g., typing skills), then it would decrease
during the third stage of learning (see also Newell the skill that could be gained during the earliest
& Vaillancourt, 200la). stages of learning. Traditional motor learning
tasks such as the pursuit rotor (figure 2.5, p. 32)
The Haken-Kelso-Bunz Model and the Bachman ladder (figure 2.8, p. 35) satis-
(Again) fied these criteria because it was highly unlikely
that subjects had ever learned skills that were
The HKB model of movement coordination was remotely similar to the skills required to perform
discussed in detail in chapter 8. As mentioned these "novel" tasks. These advantages however,
there, the performance of discrete tasks appears must be weighed against the major disadvantage
to be fundamentally different when compared that learning does not occur against the back-
to performance of continuous tasks. We have ground of a "blank slate"-we can never know
discussed several times in previous chapters that how learning this new task is influenced by the
discrete skills are dependent on a motor program skills that the learner possessed prior to practice.
for their execution. In contrast, continuous motor The self-organization approach, which was
skills are more dependent on interactions with the critical in the development of the HKB model,
environmental stimuli for their regulation. There- overcame some of the problems associated with
fore, it should not be surprising that perspectives "novel tasks" and provided many advantages for
on the learning of continuous motor tasks differ the study of motor learning. The approach both
in many respects from those constructs developed exploited the requirements for a learning task and
to explain the learning of discrete tasks.7 provided a unique window into the influence of
In chapter 8 we presented evidence that there previously acquired skills on new learning (see
are two preferred coordination patterns by which also Kelso & Zanone, 2002; Schaner, Zanone, &
continuous oscillations of two limbs or fingers Kelso, 1992; Zanone & Kelso, 1994, 1997). Because
can be reliably produced-in-phase and anti-phase in-phase and anti-phase coordination are known
coordination (the observed coordination pattern and measurable stable patterns that exist prior
is measured in terms of the relative phase lag to practice, the performance of new bimanual
between the individual cycles; 0° or 180° in these coordination patterns could be evaluated over
two instances; see chapter 2 for review). Much of the course of practice trials and compared to the
the research conducted on coordination dynam- performance of these existing skills.
ics uses the Haken-Kelso-Bunz (1985) model as In one experiment (Zanone & Kelso, 1992;
a basis (see chapter 8). Basically, the HKB model see also Kelso & Zanone, 2002; Zanone & Kelso,
states that intrinsic stabilities of the motor system 1997), subjects attempted to coordinate the rela-
attract moving degrees of freedom to perform in tive phasing of the index fingers on both hands
accordance with one of the system's naturally by rhythmically oscillating them in time to two
stable states. For the oscillating-fingers research blinking lights, which alternated in 90° relative
presented in chapter 8, the strong tendency was phase. Before, during, and after each of five days
to coordinate the fingers in an in-phase or an of practice, Zanone and Kelso had subjects per-
anti-phase pattern. Of course, theoretically, there form a type of transfer test in which the visual
are an infinite number of bimanual coordination metronomes started by blinking simultaneously
patterns that can be produced. However, if the (in 0° relative phase) and then increased in phase
system were not amenable to change, then we offset by 15° after every 20 s until 180° relative
456 !Ill Motor Control and Learning

phase had been reached. The main finding was in the lower-left. Here, the subjects showed a
that the 90° pattern was learned and became short-lived performance bias that reverted to
relatively stable with practice. An unexpected anti-phase. This result is interesting because it
finding, though, was that some subjects showed showed an effect that was very much like that
a reduced stability for the anti-phase (180°) pattern predicted by the progression-regression hypoth-
as practice trials accumulated on the 90° pattern. esis (regression to a previously mastered level of
This finding provided support for Zanone and learning). It also is consistent with a phenomenon
Kelso's argument that learning does not involve to be discussed in chapter 14 termed warm-up
simply adding a new skill to a subject's repertoire. decrement, in which retention loss has a dramatic
Rather, learning occurs against the background impact on performance. But, note here that the
of an individual's existing skills, resulting not decrement was very specific, as performance
only in the acquisition of new patterns but also reverted to a previously stable pattern, not to
in a change in the previously stabilized patterns. an unstable state.
This latter finding is a controversial one, however, The application of the HKB model and new
as it implies that new learning may result in the experimental paradigm to the study of motor
unlearning of previously acquired skills. Other learning by Zanone and Kelso (1992) has gener-
studies suggest that the destabilization of the ated considerable interest. Investigators have
anti-phase pattern is only a performance bias, studied the role of many different variables with
not a permanent destabilization in performance the purpose of discovering how new coordina-
(Fontaine, Lee, & Swinnen, 1997; Lee et al., 1995; tion patterns develop as a function of practice.
Smethurst & Carson, 2001 ), reminiscent of our pre- For example, an existing coordination pattern
vious discussions on the learning-performance may help to either stabilize or destabilize a
distinction. new pattern (Fontaine, Lee, & Swinnen, 1997;
Another finding with use of this approach is Hurley & Lee, 2006; Kostrubiec & Zanone, 2002;
also of interest. In the study by Lee and colleagues Wenderoth & Bock, 2001; Wenderoth, Bock, &
(1995), which used many of the same methods Krohn, 2002) in ways that are different than the
as Zanone and Kelso's (1992), when learning a retention of that pattern (Tallet, Kostrubiec, &
90° pattern, subjects showed strong influences of Zanone, 2008). The acquisition of a stabilized new
the existing stable patterns. As an example, the pattern appears to be greatly accelerated if move-
progress in learning for one subject is illustrated ments are unpaced (Kovacs, Buchanan, & Shea,
in figure 13.13. The panels in this figure represent 2009), perhaps due to the influences of discovery
a plot of the relative motion of the right limb learning processes (Hodges & Franks, 2002a;
together with the relative motion of the left limb. Hodges & Lee, 1999). And other factors, similar
Recall that plots of this type were presented in to those seen with more traditional approaches
chapter 2, for in-phase and anti-phase coordina- to the study of practice (chapters 11and12) and
tion (figure 2.13, p. 42). transfer (chapter 14), have been recast within a
For the 90° pattern investigated by Lee and self-organization framework (Faugloire, Bardy, &
colleagues, the" correct" plot was represented as Stoffregen, 2009; Lay, Sparrow, & O'Dwyer, 2005;
an ellipse (overlaid in the top left panel of figure Ronsse, Miall, & Swinnen, 2009), broadening the
13.13).8 All subjects practiced the ellipse pattern conceptual approaches to viewing these more
during three days of practice. There are two traditional ideas.
observations in particular to note in this figure.
First, for most subjects the initial performance
trials were performed in anti-phase; much of Summary
the initial stages of learning involved "break-
ing away" from the attraction to perform this The empirical laws of motor learning presented
bimanual task as a previously acquired stable in previous chapters are the focus of hypotheses
pattern. This process was revealed over the prac- or theories that are directed at explaining them,
tice trials on day 1, with fewer and fewer anti- and this chapter presents some of the more
phase cycles being produced and coordination important of these formulations. Learners appear
moving toward 90°. A second important finding to pass through various stages (or phases) when
concerned the initial trial on day 2, illustrated they practice a skill: a cognitive stage in which
 The Learning Process 457

Day 1 (pretest) Day 1 (after 20 trials) Day 1 (posttest)

Right limb

Day 2 (pretest) Day 2 (posttest) Day 3 (posttest)

FIGURE 13.13 Effects of practice on the development of a novel bimanual timing skill. The goal is to produce a relative
timing represented by an oval-shaped Lissajous figure.

emphasis is on discovering what to do, a fixation theory holds that the learner acquires a reference
stage in which the concern is with perfecting the of correctness (called the perceptual trace) through
movement patterns, and an autonomous stage in practice and that the improvements in skill result
which the attentional requirements of the move- from the increased capability of the performer to
ment appear to be reduced or even eliminated. A use the reference in closed-loop control. Schema
major direction for understanding skill learning theory is based on the idea that slow movements
has been provided by the individual-differences are feedback based, with rapid movements being
tradition. A significant finding is that the set of program based; with learning, the subject devel-
abilities underlying a skill appears to change ops rules (or schemas) that allow for the genera-
with practice, so that the collection of abilities tion of novel movements. Both theories can claim
underlying a skill is systematically different in a number of lines of experimental support, but
practiced and unpracticed subjects. The change neither is capable of explaining all the evidence
is in the direction of less involvement of cogni- on motor learning.
tive abilities and greater involvement of motor Several other theoretical perspectives have
abilities with practice. These characteristics of spawned a number of hypotheses about the
abilities and learning provide insight into why learning process. The cognitive perspective sug-
accurate prediction of high-level motor behavior gests that processes involved in the planning
is so difficult to achieve. and evaluation of movement affect learning,
Two major theories of motor learning are especially so when performance difficulties chal-
closed-loop theory and schema theory. Closed-loop lenge these cognitive processes. The hierarchical
458 m; Motor Control and Learning

control perspective holds that the control of the their memory requirements and pre-
skill is systematically shifted from higher-level dictions regarding transfer to novel
control processes to lower-level processes involv- variations of a learned skill.
ing motor programming. Motor programs are 2. Find a research article that was designed
assumed to be constructed through practice, but to specifically test a prediction of either
it is not known how such structuring occurs. In closed-loop theory (Adams, 1971) or
the progression-regression hypothesis, learn- schema theory (Schmidt, 1975).
ing to track is viewed as a progression toward
higher-order control, with regression to a previ-
ously achieved level of skill the result of stress Web Resources
or forgetting. The Bernstein perspective addresses
his three stages of learning, in which, at least The following link provides a list of Web sites on
for some tasks, degrees of freedom are initially learning theory (psychology):
"frozen," then later released and reorganized
with learning. In a later stage of learning, the www.nova.edu/-burmeist/learning_theory.
mechanical-inertial properties of movement html
can be exploited for movement control that is This British Web site provides many good links
more precise, faster, or more efficient (or some to information about skills learning:
combination of these).
The self-organization view (HKB model) http://news.bbc.co.uk/sportacademy/hi/
stresses the importance of interactions between sa/learning_centre/newsid_2151000/
a person's sensory-motor system and the world 2151977.stm
in which actions are performed. The examination
of humans who are learning new coordination
patterns of relative phasing has contributed an Notes
important method for studying the evolution of
1 Note that the unit of measurement for the y-axis in
skills against the background of existing skills.
The method emphasizes changes in stabilities figure 13.4 is Z, not r. This is a typical transformation
performed on correlation values prior to statistical analy-
and instabilities, both temporary and long-term.
sis because the r data are not normally distributed (the
scores are truncated at the two extremes, r = +1 and -1),
which violates an assumption for analyses of variance.
Student Assignments The Z-transformation provides a normal distribution to
the data.
1. Prepare to answer the following questions 2 Other statistics have been used to estimate error-

during class discussion: detection capabilities as well, such as the absolute error
a. Using any sport skill, describe the skill between objective and subjective scores (Newell, 1974)
characteristics of an athlete who is in and d' in signal-detection analyses (Rubin, 1978).
the Fitts phases oflearning: the cogni- 3 We have assumed normal distribution here, with the
tive phase, the associative phase, and mode represented as the correct perceptual trace. With
the autonomous phase. learning, the shape of the distribution of incremented per-
ceptual traces moves from platykurtic (flat) to leptokurtic
b. Using the same sport skill as in question (peaked). Of course, other distributions are also possible.
1a, describe the skill characteristics For example, in learning to play golf, many more people
of an athlete who is in the Bernstein have trouble with slicing the ball than with hooking the
stages of learning: the freezing degrees ball. In this case, the shape of the distribution would be
of freedom stage, the releasing and expected to change in both kurtosis and skewness, as a
reorganizing degrees of freedom stage, function of learning.
4 Learning in Adams' theory was represented by
and the exploiting the mechanical-
inertial properties stage. changes in the normal distribution-a more pronounced
mode and reduced variability. In schema theory, learning
c. Compare and contrast closed-loop is represented by a regression-line analogy. Amore power-
theory (Adams, 1971) and schema ful regression equation is developed with learning-one
theory (Schmidt, 1975) in terms of that has reduced residual variability and for which the
 The Learning Process !II 459

regression coefficient (R2 ) approaches 1 (see discussion portions thereof become fused into larger units, which
at end of chapter 2). in turn make a constantly diminishing demand upon
5 In terms of the regression-line analogy, the pre- consciousness" Gastrow, 1906, p. 42).
dicted success of a novel variation of the regression line 7 The HKB model described in this section repre-

is reduced considerably when extended to parameters sents one of several perspectives that consider how an
beyond those actually experienced in practice. Therefore, individual's sensory-motor system interacts on various
greater breadth in the variability of practice should extend levels with the environment to influence the regulation
the predictability of the schema to novel parameters. of movement. An overview of various approaches is
6 The term "gearshift analogy" is probably incorrect provided in Turvey and Fonseca (2009), Beer (2009), and
from a historical point of view. A similar idea was pro- Latash, Scholz, and SchOner (2007).
posed by Jastrow before cars even had gearshifts: "At 8 In this study the left and right limbs were to produce

the outset each step of the performance is separately and different amplitudes at a 90° phase lag. In other studies, in
distinctly the object of attention and effort; and as prac- which subjects learn 90° phase lags with equal amplitudes,
tice proceeds and expertness is gained ... the separate the left limb-right limb figure to be produced is a circle.
 t one point in the process of revising Fundamental Distinctions
this text, the two authors got together to
discuss some ideas over a long bike ride on the and Definitions
beach in Venice, California. Although the second
author had not ridden a bike in many years You may have the impression that motor learning
and, indeed, had never ridden this particular and motor memory are two different aspects of
bike before, he managed to avoid causing any the same problem, one having to do with gains
serious harm to the sunbathers and volleyball in skill, the other with maintenance of skill. This
players gathered on the beach that warm spring is so because psychologists and others tend to
day. Should we be surprised that the skill of bike use the metaphor of memory as a place where
riding is retained and transferred so easily? And information is stored, such as a computer hard
what factors might influence how well we retain drive or a library. Statements like "I have a good
and transfer these and other types of motor memory for names and dates," or "The subject
skills? Such concerns about how well skills are placed the phone number in long-term memory,"
retained over time and how well they transfer are representative of this use of the term. The
to different situations are of both theoretical and implication is that some set of processes has led
practical importance-theoretical because of to the acquisition of the materials, and now some
the need to understand how the motor system other set of processes is responsible for keeping
is structured so that skills can be produced "on them in" memory.
/1

demand," and practical because usually much

time and effort have gone into the learning
Memory
of the skills, and we need to know how such A common meaning of the term motor memory
investments can be protected from loss. This is "the persistence of the acquired capability for
chapter is about the empirical relationships performance." In this sense, habit and memory
and principles concerned with retention and are conceptually similar. Remember, the usual
transfer. test for learning of a task concerns how well the

lfil 461
462 111 Motor Control and Learning

individual can perform the skill on a retention losses in performance, or losses in retention. So,
or transfer test. That is, a skill has been learned if if you understand what measures of behavior
and only if it can be retained "relatively perma- suggest about learning, then you also understand
nently" (see chapter 10). If you can still perform the same about forgetting.
a skill after not having practiced it for a year
(or even for a day or just a few minutes), then Retention and Transfer
you have a memory of the skill. In this sense, Retention refers to the persistence or lack of
memory is the capability for performance, not a persistence of the performance, and is considered
location where that capability is stored. Depend- at the behavioral level rather than at the theo-
ing on one's theoretical orientation about motor retical level (table 14.1). It might or might not
learning, memory could be a motor program, a tell us whether memory has been lost. The test
reference of correctness, a schema, or an intrinsic on which decisions about retention are basei;l is
coordination pattern (Amazeen, 2002). From this called the retention test, performed at a period of
viewpoint, as you can see, learning and memory time after practice trials have ended (the reten-
are just different sides of the same behavioral
11
tion interval). If performance on the retention
coin," as Adams (1976a, p. 223) put it (see also test is as proficient as it was immediately after
Adams, 1967). the end of the practice session (or acquisition
phase), then we might be inclined to say that
Forgetting no memory loss (no forgetting) has occurred. If
Another term used in this context is forgetting. performance on the retention test is poor, then
The term is used to indicate the opposite of we may decide that a memory loss has occurred.
learning, in that learning refers to the acquisi- However, because the test for memory (the reten-
tion of the capability for movement whereas tion test) is a test of performance, it is subject to
forgetting refers to the loss of such capability. It all the variations that cause performances to
is likely that the processes and principles having change in temporary ways-just as in the study
to do with gains and losses in the capability for of learning. Thus, it could be that performance
moving will be different, but the terms refer is poor on the retention test for some temporary
to the different directions of the change in this reason (fatigue, anxiety), and so one could falsely
capability. "Forgetting" is a term that has to do conclude that a memory loss has occurred. (At
with theoretical constructs, just as "learning" this point it might be helpful to review the
does. Memory is a construct, and forgetting is learning-performance distinction presented in
the loss of memory; so forgetting is a concept chapter 10.)
at a theoretical, rather than a behavioral, level For all practical purposes, a retention test
of thinking. and a transfer test are very similar. In both cases,
As shown in table 14.1, the analogy to the the interest is in the persistence of the acquired
study of learning is a close one. At the theoreti- capability for performance (habit). The two types
cal level, learning is a gain in the capability for of tests differ only in that the transfer test has
skilled action, while forgetting is the loss of same. subjects (all or some) switching to different tasks
On the behavioral level, learning is evidenced by or conditions, whereas the retention test usually
relatively permanent gains in performance, while involves retesting subjects on the same task or
forgetting is evidenced by relatively permanent conditions.

Relatively permanent losses in perfor-

mance, or retention losses
 Retention and Transfer 1111 463

Measuring Retention and 14.1 shows the hypothetical scores of a group

of subjects who practiced the pursuit rotor task
Transfer (see figure 2.5, p. 32) for 30 trials and then, after
a retention interval, performed a retention test
Tests of retention and transfer provide indicators involving 30 additional trials. The absolute-
about the persistence of an acquired habit during retention score is 20, because performance in
an absence from practice, or about the way in trial 1 of retention is approximately 20 s of time
which previous practice influences performance on target (20 s TOT). Notice that the absolute-
on a new task. Unfortunately, straightforward retention score is not based in any way on the
conclusions from such tests are not always pos- level of performance attained in the practice trials.
sible. Next, we present the most common and
Relative Retention
important of the various methods and measures
of retention and transfer that have been devised Various measures of relative retention are possi-
by researchers, and we suggest which ones pro- ble, such as those using a difference score and those
vide the most useful information. using percentage scores. These measures express in
various ways the absolute-retention score relative
Retention of Learning to scores obtained during the practice trials.
In motor memory research, a number of different Difference Score Probably the most common
measures of retention have been used, and these relative-retention score is a difference score that
different methods provide somewhat different supposedly represents the "amount" of loss in skill
interpretations about the underlying forgetting over the retention interval. It is computed by taking
processes. The most common of these methods the difference between the performance levels
are absolute retention and various measures of at the end of the practice session and the begin-
relative retention. ning of the retention test. In the example given in
figure 14.1, the difference score is 5 s, as the group
Absolute Retention performed with a TOT of 25 s before the retention
By far the most simple (and scientifically justifi- interval and 20 s afterward. Such measures are aes-
able) measure of retention is absolute retention, thetically pleasing to many investigators because
defined simply as the level of performance on they seem (erroneously, however) to represent the
the initial trial(s) of the retention test. Figure forgetting processes more or less directly.

25

~ 20
a;
El
.!!! 15
c:
0
GI
E
j:: 10

5 10 15 20 25 30 5 10 15 20 25
Original-learning trials Retention interval Retention-test trials

FIGURE 14.1 Hypothetical performance curves on the pursuit rotor for original-learning and retention-test trials.
464 rn Motor Control and Learning

Percentage Score A second kind of relative- be the "rate" of relearning, even if the first trial
retention score is a percentage score, which or so show poor performance (due, for example,
represents the "amount" lost in retention over to warm-up decrement, discussed later).
the retention interval relative to the amount of
Contrasting the Various Retention
improvement that occurred on the task in the
Measures
practice session. That is, the percentage score is
the difference score (as defined earlier) divided by While it may seem that these various methods
the amount of change in performance during the merely provide subtle differences in the measure-
practice session (another difference score), then ment of a single process (forgetting), this is not
multiplied by 100 for conversion into a percent- the case. According to an analysis of the problem
age. In the example in figure 14.1, the percentage some years ago (Schmidt, 1971a, 1972a), the
score is the difference score (5 s) divided by the relative-retention scores are flawed by a variety
amount of performance change during the prac- of factors. The basis of the problem is that all
tice trials (25 - 10 = 15 s) and multiplied by 100, these scores come from performance measures,
or 5 I 15 X 100 = 33.3%. The meaning usually with changes in performance being used to infer
given to the percentage score in this case is that something about the changes in the internal state
one-third of the amount of original improvement (habit or memory) that underlies performance.
during practice was lost over the retention inter- Therefore, all the problems with performance
val. Be careful, though, because such estimates curves that we mentioned with respect to the
are sensitive to temporary factors that alter per- measurement of learning (ceiling and floor
formance during practice (e.g., fatigue, random effects, for example, in chapter 10) also apply to
practice) and thus alter the size of the denomina- the measurement of forgetting. In particular, dif-
tor. However, these scores are sometimes useful ference scores are subject to a variety of influences
when one wishes to compare (usually informally) that cloud interpretations about forgetting, cast-
the retention on two different skills, perhaps with ing doubt on their usefulness. Moreover, the per-
different scoring systems. centage score is based on two difference scores,
Savings Score A third measure of retention, one divided by the other to gain the percentage,
which was introduced long ago by Ebbing- clouding the issue even further. The savings score
haus (1913) and has regained popularity in suffers a similar problem since the assessment of
recent years (e.g., Keisler & Willingham, 2007; "savings in relearning" itself employs a difference
Krakauer & Shadmehr, 2006; Seidler & Noll, score in its computation.
2008), involves the "savings" in relearning. The problem is not just a technical or academic
That is, after a retention interval, one measures one (Schmidt, 1971a). Some of the most funda-
the number of trials required for the subjects to mental variables in forgetting have empirical
reach the level of proficiency achieved in original effects that seem to depend completely on the
practice. In the example in figure 14.1, the sav- ways in which retention is measured. If forget-
ings score would represent the number of trials ting in figure 14.1 is measured by the absolute-
"saved" in the retention test in the process of retention method, then numerous studies show
reaching the 25 s of TOT that had been achieved that absolute retention increases as the amount of
at the end of the practice session. Notice that practice increases, just as we might suspect. But
the number of trials to relearn is generally less if forgetting is measured by the relative-retention
than the total number of practice trials; in this methods, then relative retention (computed from
case (as opposed to 30 trials in acquisition) 12 the same set of data) decreases as the amount of
retention-test trials were required to reach 25 practice increases (see Schmidt, 1972a). Thus,
s TOT. Therefore, in the retention session, the the relationship between forgetting of skills and
subject regained the same level of proficiency as the amount of practice would be completely dif-
had been achieved in the practice session-but ferent depending on how retention is measured.
in this instance it required 18 fewer trials than it Obviously, this has caused, and will continue to
did in the practice session (savings = 30 - 12 = cause, many confusing situations for students
18 trials). The idea of a savings score is that the who are attempting to understand the principles
more complete the retention, the faster should of motor forgetting. The absolute-retention score
 Retention and Transfer oo 465

minimizes these problems, and it is the most Positive and Negative Transfer Consider the
simple and straightforward one to use. possible results of such an experiment as shown
in figure 14.2. Here, the task of interest is task B,
Transfer of Learning so task A performance is not graphed. In figure
Transfer is usually defined as the gain (or loss) 14.2, group I, which had task A prior to task B,
in the capability for performance in one task as performs task B more effectively than does group
a result of practice or experience on some other II, which did not have the experience with task
task. Thus, we might ask whether practicing a A. In this case, we conclude that experience on
task like badminton would produce benefits or task A has provided increased capability for task
losses (or neither) for another task such as tennis. B, equal to 30 units on trial 1 of task B. When
If it turns out that the performance of tennis is the practice on task A enhances subsequent per-
more effective after badminton experience than formance on task B, we say that positive transfer
it would have been with no previous badminton occurred from task A to task B.
experience, then we would say that the skills Now consider what happens with another
acquired in badminton have "transferred to" hypothetical group (group III). As seen in table
the skills involved in tennis. It is as if something 14.2, group III practices task Z (rather than task A
that is learned in the badminton situation can be as group I did) prior to trials on task B. In figure
carried over to (or applied to) the task of playing 14.2 the performance for group III is less skilled
tennis (Schmidt & Young, 1987). in relation to that of group II by 20 units on trial
1 of task B. For the reasons just mentioned, we
Transfer Experiments conclude that experience on task Z has interfered
Experiments on the transfer of learning can use a with group III' s capability for performance on
variety of experimental designs, but we will not task B. In this case, we would say that negative
consider them all here (see Ellis, 1965, for a com- transfer occurred from task Z to task B.
plete description). In the simplest of all designs, Proactive and Retroactive Transfer In the
assume that there are just two groups of subjects examples given so far, the transfer seemed to
(groups I and II). In table 14.2, group I practices
task A for some arbitrary number of practice
trials, after which this group is transferred to
practice on task B. Group II does not practice
task A at all, but merely begins practicing task B. 80
You can think of tasks A and Bas any two activi-
~ 70
ties; they could be different tasks such as bad-
Xl
minton and tennis, or they could be two slightly :S 60
different variations of the same task, such as the Q)

pursuit rotor at different speeds. Thus, when g

IU
50
E
the two groups begin practice of task B, the only
~ 40
systematic difference between them is whether or g_
not they have had previous experience on task A. III 30
.II:
Ill
~ 20

10

2 3 4 5 6 7 8 9 10 11
Trials on task B

FIGURE 14.2 Performances on task B for a group with

no prior experience (11) or with prior practice on task A
(group I) or task Z (group Ill). If group I outperforms group
II, then positive transfer has occurred. If group Ill performs
more poorly than group II, negative transfer has occurred.
466 111 Motor Control and Learning

work "forward" in time from taskAor Z to task B. group II in the experiment. The data from groups
This is termed proactive transfer. However, we can I and II are illustrated again in figure 14.3. On
also consider retroactive transfer, that is, transfer trial 1 the difference between the two groups is
that seems to work "backward" in time. Consider 30 units (labeled as points X and Y). At the end
the more complex experimental design shown in of practice, group H's performance level is 20
table 14.3. Here, two different treatment groups units (point C) and has therefore improved by 40
(groups IV and V) both perform task B. Then, units (60 - 20). The amount of improvement in
group IV performs task Q while group V per- task B by group II can be represented as the total
forms nothing. Later, both groups return to task improvement shown in task B (or X- C). Thus,
B for a retention test. If the retention performance group I's experience with task A has provided
on task B is more effective for group IV than for 30 out of the possible 40 units of improvement,
group V, we say that positive retroactive transfer or 75% transfer. In terms of a more general
occurred from task Q to task B; practicing task formula,
Q seemed to "enhance" the capability already Percent transfer= (X - Y) I (X - C) X 100 (14.1)
shown on task B. Alternately, if the performance
of task B on the retention test is less effective for in which X = 60, Y = 30, and C = 20 score units.
group IV than for group V, we say that negative The formula can also be used for negative transfer
retroactive transfer (or interference) occurred; as shown in figure 14.2. Here, the values X and C
here, practicing task Q seemed to degrade the remain the same, but Y (the initial performance
level on task B by group III) is larger than it was
capability for the previously practiced task B.
for group I (i.e., 80). Being careful to keep the
The retroactive- and proactive-transfer designs
signs of the numbers straight, and noting that the
are similar in that they both consider the perfor-
numerator of the equation is a negative number
mance on the initial trials of task B in the reten-
tion test (or the test phase in table 14.2) to be the
critical data indicating transfer. Some measures
of these different performances are described in
70
the next sections.
Measurement of Transfer :§: 60
gi
The "amount" of transfer from one task to another E
; 50
can be assessed in a number of ways, all of which (,)
c
suffer from the basic problems raised many times ca
earlier about the measurement of performance, ...
E 40
.g
learning, and forgetting; thus none of these ! 30
methods will be very satisfactory in measuring m
.II:
transfer. Rather they are used to describe the rela- gj 20
I-
tionships among curves such as those in figure
14.2 and are occasionally helpful in discussion 10
of the results of different transfer experiments.
Percentage Transfer One method of estimating 2 3 4 5 6 7 8 9 10 11
the amount of transfer is to consider the gain in Trials on task B
performance as a result of experience on task A
as a percentage of the "total amount learned" by FIGURE 14.3 The calculation of percentage transfer.
 Retention and Transfer !Ill 467

(i.e., X - Y, or 60 - 80, or -20), we calculate trans- "spent" in the simulator (task A) is inexpensive
fer as -20 I 40 3100 = -50%. relative to the time "saved" in learning to fly the
Roughly speaking, we can interpret the per- MD-11 (task B), and it is safer as well. In such
centage transfer as the percentage of gain (or loss) situations, the effectiveness of a simulator-based
on task Bas a result of prior practice on task A. training program is often evaluated in terms of
Positive transfer of 100% would imply that the financial savings, such savings being the number
performance on the first trial of task B for group of hours saved on task B (the MD-11) multiplied
I is at the final level of performance (i.e., point C by the number of dollars per hour of practice on
in figure 14.3) demonstrated by group II. Transfer task B. In the case of the MD-11, dollar amounts
of 0% would mean that the two groups are the of savings can be very large.
same in initial performance on task B (i.e., both
at level X).
The reason this measure is inadequate, of
Retention and Motor
course, is that the amount of improvement on Memory
task B (i.e., X - C) will depend on the amount of
practice provided, on the scoring system used One of the most frequently studied theoretical
for task B, on the nature of the subjects, and on issues in psychology-an issue that people often
countless other arbitrary factors that affect the disagree about-concerns memory. Is memory
shapes of performance curves. But using percent- a result of some processing of an event, or does
age transfer measures can serve a useful purpose memory refer to the processing itself? Are there
in describing the relationships among the curves; different types of memory, such as memories for
just be careful not to take the finding of, say, 75% movements, for sensations, for smells, and the
transfer too literally. like, or is there just one memory, whose retention
Savings Score Another, far less frequently used characteristics are a product of the nature and type
method for describing the amount of transfer is a of processing that is conducted? Questions such
savings score, as already discussed. Here, the sav- as these are hotly debated topics. For example,
ings score represents the amount of practice time a scan of the chapters in Byrne (2008) reveals an
"saved" (i.e., reduced) on task B by having first extremely wide diversity of topics, studied at
practiced task A. In figure 14.3, group I (which many different levels of analysis. For the most
had practiced task A previously) begins its perfor- part, these topics are beyond our present pur-
mance of task Bat a level of performance equiva- poses. Rather, we present some of the evidence
lent to that shown by group II after six trials. It is about the retention (this section) and retention
possible to say that group I "saved" six trials in loss (next section) of motor skills.
the learning of task B by having first learned task
Retention of Skill for Continuous
A. But this is not the whole story; the "savings"
on task B are almost certainly compensated for by Tasks
a "loss," because task A had to be practiced, and That many motor skills are nearly never forgotten
the practice time on task A is usually going to be is almost a cliche. Examples such as swimming
longer than the amount of time "saved" on task and riding a bicycle, in which performance after
B. That is, for learning task B, usually nothing is many years of no intervening practice is nearly as
as efficient as practicing task B (see chapter 11 for proficient as it was originally, are frequently cited.
discussion on practice specificity). Ideas about such examples, though, are seldom
Such" savings" begin to have importance when based on acceptable experimental methods;
the financial cost of practice is considered. A fortunately, many laboratory examples of these
common example is in learning to fly an airplane, situations have been studied, and these results
such as the McDonnell Douglas MD-11. To prac- seem to say the same thing.
tice in the actual MD-11 aircraft would be very Although many studies could be cited to
costly, so computer-based simulators that closely illustrate the point, we consider a representative
resemble the airplane cockpit are frequently used study with long retention intervals by Fleishman
for practice (see figure 14.15, and the related and Parker (1962). They used a three-dimensional
discussion, later in the chapter). Here, the time compensatory tracking task (the Mashburn task,
468 111 Motor Control and Learning

figure 2.Sb, p. 32), with movements of the hands using the pursuit rotor and stabilometer tasks,
in forward-backward and left-right dimensions found nearly no retention losses after retention
and movement of the feet in a left-right dimen- intervals of 21 days; later, he found only small
sion. Subjects practiced in sessions for 17 days, losses in performance on the stabilometer task
and then separate groups performed retests after with retention intervals of up to one year (Ryan,
either 9, 14, or 24 months. 1965). There are many other examples, and the
The scores for practice and retention tests are generalization continues to hold. Continuous
shown in figure 14.4, where scores for all three motor tasks are extremely well retained over very
retention groups have been averaged together long retention intervals, just as the cliche about
in the practice session. After the different reten- the bicycle would have us believe.
tion intervals, the various groups were nearly
equivalent, and none had shown any appreciable Retention of Skill for Discrete Tasks
losses in proficiency even after two years of
layoff. Some tendency was seen for the two-year While there is ample evidence of nearly complete
group to have slightly less proficiency than the retention of continuous skills, the picture appears
groups with shorter retention intervals, but the to be quite different for discrete skills. Consider
differences were very small and the losses were an example by Neumann and Ammons (1957).
regained completely in three sessions. These The subject sat in front of a large display with
small differences are not very meaningful when eight pairs of switches arranged in an inside
one compares the retention-test performance to and an outside circle of eight switches each. The
the level of performance at the start of practice. subject was to turn the inner switch "on" and
Certainly, this continuous task was retained then discover which switch in the outer circle
nearly perfectly for two years. was paired with it; a buzzer sounded when the
Other studies, using different continuous correct match was made. Subjects learned the task
tasks, have shown very similar effects. Meyers to a criterion of two consecutive errorless trials,
(1967), using the Bachman ladder climb task, and then retention intervals of 1 min, 20 min, two
demonstrated nearly no loss in performance for days, seven weeks, and one year were imposed
retention intervals of up to 12 weeks. Ryan (1962), for different groups of subjects.

2500

....
0
CD
2000

.!
::J~
- UI
g~ 1500
.c ::J

-:s..
~
m >-
"tJ ..
CD m

Cl ..
CD m
1000
-~
.5
c
m
CD
:!: 500

Six-minute practice periods

Training sessions Retention test

FIGURE 14.4 Mean performance on a three-dimensional tracking task in original learning and after three retention intervals.
Reprinted from E.A. Fleishman and J.F. Parker, 1962, "Factors in the retention and relearning of perceptual motor skill," Journal of Experimental Psychology 64: 218.
 Retention and Transfer 469

8 8

7 7

ti)
Q)
6 ti)
Q)
6
ti) ti)
c: c:
&. 5 &. 5
.
ti)
Q)

....
ti)
~
....
~
0
.
() 4
.
()
~
0
4

()
3 ()
3
c: c:
Ill Ill
Q) Q)
:a 2 :a 2

9 18 27 36 45 54 63 9 18 27 36
a Acquisition trials b Retention and relearning trials

FIGURE 14.5 Mean performance of a discrete task in original learning and after various retention intervals.
Reprinted. by permission, from E. Neumann and R.B. Ammons, 1957, 'Acquisition and long term retention of a simple serial perception motor skill,'' Journal of Experimental
Psychology 53: 160.

The main findings are presented in figure 14.5. have a heavier emphasis on verbal-cognitive ele-
Some losses in performance appeared after only ments (learning which switch in the inner circle
20 min, and the losses became progressively is paired with which switch in the outer circle in
greater as the length of the retention interval the Neumann & Ammons study, for example),
increased. In fact, after one year, the performance there is more loss for the discrete tasks over time.
was actually less correct than the initial perfor- Ideas similar to this have generated considerable
mance in practice had been, suggesting that the interest among neuropsychologists who study
forgetting was essentially complete. However, differences in the retention characteristics of vari-
notice that in all cases the improvements during ous tasks (e.g., see "Retention of Motor Skills in
the retention trials were more rapid than in the Amnesia").
original-practice session (as indicated by compar- However, while it is true that most of the dis-
ing the slopes of the relearning and practice ses- crete tasks that have been studied in retention
sion curves), indicating that some memory for the situations seem highly verbal-cognitive (e.g.,
skill was retained, which facilitated performance Schendel & Hagman, 1982), there is no reason
in these relearning trials. that discrete tasks must be so. Certainly, one can
think of many discrete tasks that have relatively
Continuous Versus Discrete Tasks little reliance on verbal-cognitive abilities (e.g.,
Why is there such a large difference in the reten- throwing, striking, pole-vaulting). What would
tion characteristics of continuous and discrete be the retention characteristics of a discrete task
skills, with continuous tasks having nearly per- that was highly "motor" in nature? Lersten (1969)
fect retention and discrete tasks having such poor used an arm movement task (the rho task) in
retention? A number of hypotheses have been which a circular and a linear movement compo-
proposed to explain these differences, and they nent had to be performed as quickly as possible.
are discussed next. He found approximately 80% loss (of the original
amount of improvement) in the circular phase,
Verbal-Cognitive Components and a 30% loss for the linear component, with
One hypothesis is that verbal-cognitive compo- retention intervals of one year. Similarly, Martin
nents are somehow more quickly forgotten than (1970) used a task in which the subjects moved
motor components; because discrete tasks seem to the hand over two barriers and then returned to
470 Motor Control and Learning

Retention ofMotor Skills in Amnesia

The examiriation of amnesia patients has provided some surprising information about the retention
ofmotorskf'lscompar~d to other types of i.ntormation. Two patients are particularly noteworthy. One
famous patient {H.M.) was studied by neuropsychologists tor many years following an operation on
the temporal lobes in his brain. The result of the surgery was a devastating memory deficit, leav-
ing him unable to reta!.ninformation in memory for more than very brief durations. H.M. was able
to learn motor skills, however, such as mirror tracing and pursuit rotor tracking. Remarkably, H.M.
showed impressive retent.ion of these motor skills after periods of no practice, despite the fact that
he could not remember everhaving practiced these tasks or the experimenters who had conducted
the expl3rirnents {Milner, Corkin, & Teuber, 1968)!
Memory in apatieritwithAlzheimer's disease (M.T.) was documented during the play of two rounds
ofgolfvvith neuropsychologist Daniel Schacter. This patient had beendiagnosed with a progres-
sively deteriorating memory disorder, and at the time of the golf "experiments" showed extremely
poor performance on standard tests of verbal memory. What makes M.T.'s case interesting is that
his golf skill had remained relatively unimpaired. According to Schacter (1983), he remained "able
to execute acomplexsetof acquired perceptual-motor procedures in a relatively fluent manner ...
generally hit the ball straight and frequently hit it forrespectable distances ... frequently sank putts
up to 5or 6Jeetlong,andtwice holed putts from over 20 feet" (p. 239). Nevertheless, M.T.'s memory
deficits caused frequent problems in playing golf. For example, if M.T. was the second person of
the twosome to hit his tee shot and left the teeing area immediately, then he had a good probability
of finding his balf. However, if he teed off first, he usually had no idea where the ball had gone and
occasionally had forgotten that he had already played his tee shot!
Th.e exi$tence of motor retention for newly acquired learning Jn people with amnesia (H.M.) and
tor a previously acquired skill (M.T.) in the presence of severe retention deficits for other types of
information is a type of memory dissociation, Similar dissociations for preserved retention of motor
skill, combined With memory loss for information about the details of the practice session, have
since been documented tor healthy subjects (e.g., Hikosaka et al., 2002; Verdolini-Marston & Balota,
1994).These dissociations have been explained by some theorists as supporting the view that the
retention of (or memory for) motor skills is fundamentally different from the retention of other types of
information, such as verbal knowledge (e.g., Roediger, 1990; Schacter; 1987). Various dichotomies
have been usedtoctescribe this distinction, such as implicit versus explicit memory and declara-
tive versusproceclural memory, representing a continuing source of experimental and theoretical
curiosity in contempor~rry research.

a starting switch as quickly as possible, finding practice increases. In tracking, for example,
approximately 50% retention loss over a four- there are many instances within a trial lasting
month retention interval. The large amount of 30 sin duration in which the pointer and track
loss in retention for discrete skills that can be become separated, with each instance requiring
considered "mostly motor" is similar to the loss a separate adjustment. Thus, a single "trial" may
experienced by Neumann and Ammons' subjects require many separate "discrete" actions. Con-
(figure 14.5), suggesting that there is more to these trast this situation to that for discrete tasks, for
effects in retention than just the "motorness" of which a trial typically consists of a single adjust-
the tasks. ment or action. It stands to reason, therefore,
that with the same number of learning trials, the
Amount of Practice continuous task receives far more practice than
One of the major factors determining absolute the discrete task. The extra amount of practice,
retention is the amount of original practice, with according to this hypothesis, leads to increased
retention increasing as the amount of original retention, since it is well known that absolute
 Retention and Transfer 111 471

retention is directly related to the amount of suffered no loss in relative timing accuracy. These
original practice. findings suggest that at least some information
from learning discrete tasks is retained quite
What is a "Trial"?
well. Moreover, these findings make sense from
Another notion, related to the one just presented, a schema theory view (Schmidt, 1975b). One has
is that the definition of trial is quite arbitrary; a no need to retain parameter information over
trial can refer to both a 200 ms reaction-time (RT) long periods of time, because that information is
performance and a 2 min duration performance used only briefly to update the schema. In con-
on a tracking task. This poses a problem for defin- trast, schema theory suggests that the retention
ing the amount of original practice for the task, characteristics of GMPs are quite strong so that
and it is also a problem in connection with the the invariant features of the action can be recalled
retention test. Remember, the level of absolute and parameterized as needed. Certainly, much
retention is measured in terms of the perfor- more work could be done to explore the ideas
mance on the first few "trials" of retention-test introduced in Swinnen' s experiments.
performance. If a "trial" is a 2 min performance,
there could be a great deal of relearning occur-
ring within a trial for the continuous task, with Retention Loss
no relearning within a trial for a rapid discrete
performance. So the initial movements within In this section we present four different research
the first trial for the continuous task could show methods used to investigate retention loss in
considerable retention loss, but the experimenter motor performance, followed by a discussion of
might not detect it because the error in the initial related theoretical and experimental issues about
performance would be" averaged" with the later the processes through which retention loss occurs.
portions of the trial on which performance was Each method highlights some important features
more proficient. Because this could not occur for about performance loss that are revealed under
the discrete task, it is possible that the amount different task conditions.
of forgetting is typically underestimated for the
continuous task and not for the discrete task, Iconic Memory and Motor
making the two kinds of tasks appear to be dif- Performance
ferent in their retention characteristics when they
might otherwise not be. Fleishman and Parker As we discussed in chapter 5, motor performance
(1962) found a great deal of improvement within benefits considerably from the availability of
a continuous-task trial, as might be suspected. visual information, especially for actions that
require precise end-point accuracy, such as
Retention of Generalized Motor manual aiming (e.g., typing; moving a cursor).
Programs Versus Parameters However, there is considerable evidence to sug-
Another possible difference in the forgetting of gest that continuous visual information is unneces-
continuous and discrete tasks is that researchers sary in order to maintain accuracy. The reason is
might be examining different characteristics of that our memory for the immediate visual envi-
the task. Evidence of this was found in a study by ronment can "fill in" the gaps if the continuous
Swinnen (1988), who had subjects learn an elbow supply of vision is cut off. For example, suppose
flexion-extension-flexion task with a goal move- you took aim at the bull's-eye in dart throwing
ment time (MT) of 650 ms. Following 60 trials of and the room lights suddenly went out just before
practice (with knowledge of results, KR), no-KR you started moving the dart. How would perfor-
retention tests were given after intervals from mance be affected? Research using experiments
10 min to five months. Swinnen analyzed sepa- that closely resemble this situation suggests that
rately the retention of absolute timing (related performance would depend on the length of time
to the movement parameter) and relative timing you were in the dark before throwing the dart.
(related to the generalized motor program, GMP) Studies by Elliott and his colleagues suggest
and found that absolute timing decayed rapidly, that motor performance deteriorates quickly
supporting much of the research in this area for because persistence of the visual information
discrete tasks. In contrast, the GMP information (the icon) fades rapidly from sensory memory
472 Motor Control and Learning

(Sperling, 1960). For example, in a study by Elliott

and Madalena (1987), subjects moved a stylus to 20
a target under various conditions of available
room light. A control condition provided subjects
with continuous visual feedback of the target
and stylus. In another condition, the room lights e.§. 15

were extinguished as the subjects initiated their ...

e...
movements; thus the entire movement (durations QI
QI
of 200-500 ms) was made in the absence of any "ti

direct visual information. The other three condi- aii 10

tions also involved movements without visual E
as
information available; however, these movements ...
iii
were made after the room lights had been extin- ~ 5
guished for 2, 5, or 10 s.
As figure 14.6 shows, subjects could perform
the aiming movements well without visual
information if the entire movement was com-
Full vision 0 2 5 10
pleted within half a second after the room lights Delay conditions (seconds before movement)
were turned off. Performance was markedly
disrupted, however, after a wait in the dark of 2 s
FIGURE 14.6 Total amplitude error in aiming under
or more. Elliott and Madalena (1987) interpreted conditions of vision and without vision under various delay
these findings to suggest that a very short-lived conditions.
memory for visual information can support per- Data from Elliott and Madalena 198Z

formance rather accurately (see chapter 3; reviews

by Elliott, 1990, 1992; also Farrell & Thomson,
1998). However, the information is prone to memory paradigm (chapter 3). This involves the
forgetting due to a decay of the icon-a process presentation of a movement, followed by recall
whereby rapid information loss is attributable to of that movement after very brief time intervals,
the passage of time. often only a few seconds in duration. These
The findings of Elliott and Madalena (1987) studies used methods that paralleled methods
and others (e.g., Binsted, Rolheiser, & Chua, in experiments in memory for verbal materials,
2006) indicate that motor performance can be early investigations having been conducted by
supported for a brief time by a short-term sen- Brown (1958) and Peterson and Peterson (1959).
sory store, which loses information quite rapidly. In one of the first motor studies, Adams and
These findings suggest a process similar to that Dijkstra (1966) had subjects move to a stop that
proposed in the oldest theory of forgetting, the defined a target position, then return to a start-
trace-decay theory. It is a passive theory of memory ing location for a retention interval, and finally,
loss caused by disuse-information is forgotten estimate the defined target position but with the
because it is not practiced and therefore" decays" stop removed. Subjects were blindfolded and
with time. The memory of an item, event, or skill not given KR about their movement accuracy. In
is thought to be represented as a trace in the cen- addition, subjects were given various numbers
tral nervous system, with the strength of this trace of "reinforcements," whereby movement to the
weakening over time. When the information or target position was presented 1, 6, or 15 times
skill in memory is needed at some future time, before the retention interval.
performance accuracy is related to the current The major findings are presented in figure
strength of the trace. This idea accounts well for 14.7. The absolute errors on the recall trials
the common effects of disuse and, of course, for are presented as a function of the number of
the fact that time, per se, seems to be a strong "reinforcements" and the length of the retention
factor in forgetting. interval. As the length of the retention interval
Considerable research on trace-decay effects increased, the error in recall also increased, with
in slow, linear-positioning tasks has been con- the increases being nearly maximized by the
ducted using what is called the short-term motor time the retention interval was 80 s in length and
 Retention and Transfer 473

30

26

'E 22

..
.§.
0
co 18
!::::s
0 14
t/I
.c
Ill
c:
Ill
co 10
:a

20 40 60 80 100 120
Retention interval (s)

FIGURE 14.7 Mean absolute error in positioning as a function of the retention-interval length and the number of "reinforce-
rnents."
Reprinted, by permission, from J.A. Adams and S. Dijkstra, 1966, "Short-term memory and motor responses," Journal of Experimental Psychology 71: 317. Copyright© 1966 by
the American Psychological Association.

with no important increases thereafter. Similar & Hindorff, 1986; see also Rosenbaum, 2009).
to memory for verbal items, memory for these The subject's task is easy to simulate: The basic
linear-positioning movements appears to have requirement is to speak aloud as many letters
a forgetting process that is nearly completed in as possible in 10 s, alternating between a loud
about 1 min. Also, the rate of forgetting appears voice and a soft voice with each spoken letter.
to be slowed by "reinforcements," or practice; For example, in one condition the subject would
the errors were systematically smaller with more shout the letter A, then softly speak the letter b,
repetitions of the target position. then shout C, softly speak d, then start over again
One interpretation of these results is that by shouting A, and so on (AbCdAbCd . . .). Notice
the movement to the stop created a short-term that a loud vocalization was always required for
memory representation of the feedback quali- the letters A and C, and a soft vocalization was
ties of the correct position. Further, it appeared always required for b and d. And this is true for
that, although this representation was weakened any even-numbered memory set. Now compare
over the course of the empty retention interval, this to an odd-numbered memory set, such as
it was strengthened by repetition. These factors AbCaBcAbCaBc. Notice now that the stress on a
combined to determine the "strength" of the specific letter switches to the opposite stress on
representation against which the feedback was each repeated cycle. This feature is consistent for
compared at the retention test-weakened by all odd-numbered memory sets.
time, but strengthened by repetitions. It is also The speed and error data from Rosenbaum
possible that forgetting can occur by means other and colleagues' (1986) experiment are presented
than trace decay. This idea is presented in the in figure 14.8. As you would expect, more letters
next example. in the even-numbered memory sets (2, 4, 6, 8)
were produced in 10 s than in the odd-numbered
Brief Postmovement Memory sets (3, 5, 7, 9) (figure 14.8a). Also, trials on which
Now consider a very different and clever memory- errors occurred were more frequent for the odd-
related paradigm, developed by Rosenbaum and numbered than the even-numbered sets (figure
his colleagues (Rosenbaum, Weber, Hazelett, 14.8b).
474 111 Motor Control and Learning

42 50
Ill

s 38
"C

40
5l
~ 34 !!!
.5
"C
..e
Q)

B 30
:5 30
::I ·~

..
"C
0 Ill
iii
..
c..
Ill
Q)
26 :5
'E 20
ii
Q)

-..
0
Q)
.c
22
l:!Q)
D..

10
E 18
::I
z

2 3 4 5 6 7 8 9 2 3 4 5 6 7 8 9
a Memory-set length b Memory-set length

FIGURE 14.8 (a) Total number of letters spoken and (b) percentage of trials with errors as a function of memory set length.
Adapted from Journal of Memory and Language, Vol. 25, D.A. Rosenbaum, R.J. Weber, W.M. Hazelett, and V. Hindorll, "The parameter remapping effect in human performance:
Evidence from tongue twisters and finger fumblers;' pg. 713, Copyright 1986, with permission of Elsevier.

What do these findings suggest about motor letter. We expect that if the length of the memory
memory? One view is that this task required sets had been extended even further (e.g., to
subjects to vocalize letters (all having different 25 and 26 letters), the performance differences
learned GMPs) with different parameteriza- between the odd- and even-numbered letter
tions-in this case the specific parameter of strings might have been eliminated completely.
interest was whether the letter is spoken loudly This finding suggests a weakening of the influ-
or softly. Once the letter was produced, the ence of a previous performance on selecting a
parameter used for that instance was retained parameter for a subsequent performance, which
in memory. If the next vocalization of that same is dependent on the memory-set size.
letter required the same parameter (i.e., as in an Two possible influences seem to be occurring
even-numbered memory set), the accurate rep- in Rosenbaum and colleagues' (1986) study. As
resentation that remained in memory facilitated the length of the memory set increased, the time
performance. However, if the opposite param- between any two vocalizations of the same letter
eter was required (i.e., as in an odd-numbered increased, resulting in a decay of the representa-
memory set), the memory of the previous param- tion for the previous parameter. The mere pas-
eter for that letter interfered with performance sage of time is not all that happened, though,
because the remembered parameter needed to because as the memory-set size increased, more
be changed. Thus, a strong memory representa- intervening letters were spoken, which caused
tion either facilitated or degraded performance, more interference with the memory for any specific
depending on the task demands. previously spoken letter. Thus, another cause of
But notice something else in the speed and forgetting may have had something to do with
error data in figure 14.8. As the length of the these events, rather than mere passage of time as
memory set increased, the size of the performance trace-decay theory would have it.
difference between the even- and odd-numbered Interference theory suggests that memory is
sets was reduced. The memory-set effect, which actively degraded by other events. Such inter-
previously had either facilitated or degraded ference, according to the theory, can be of two
performance, was reduced when more letters basic kinds: proactive interference and retroac-
intervened between the repetitions of any one tive interference (Underwood, 1957). The most
 Retention and Transfer 475

common research method involves an experi-

mental paradigm in which the interfering event
occurs between the time of the storage of the 60
to-be-remembered information and the time of
the attempted recall-that is, during the reten- ~ 50
tion interval. The term retroactive implies that the E

interference "works backward" on the memory;

s
~ 40
of course, it does not work backward at all, but Gi
it does nevertheless serve to disrupt the recall of !
:I 30
something that occurred before the interference. 1 0I/)
.c
Interference can also occur in a less obvious <( 20
way when something that happens before the
criterion memory task causes interference with
10
the recall of that criterion information. The term
proactive implies that the information already "in
memory" interferes with more recently acquired 0 2 4
information. # prior positions
Using the short-term motor memory para-
digm described in chapter 3, experimenters have FIGURE 14.9 Mean absolute error in positioning as a
attempted to assess the mechanisms causing function of the retention-interval length and the number of
forgetting in relation to interference theory. With previous positions.
respect to proactive interference, neither Adams and Adapted, by permission, from K.M. Ascoli and R.A. Schmidt, 1969, "Proactive
interference in short-term motor retention;' Journal of Motor Behavior 1: 29-35,
Dijkstra (1966) nor Posner and Konick (1966) found adapted with permission of the publisher (Taylor & Francis Ltd, http://www.
informaworld, com).
evidence that later positions to be remembered in
a sequence were less accurate than earlier ones,
which would be expected if the proactive interfer- for constant error, with increased prior positions
ence from the earlier movements were disrupting making the movements systematically too short.
the memory of the later positions. Such findings The data can be interpreted to mean that proactive
had been shown in verbal behavior. One reason interference is a factor in the retention of these
these proactive effects may not have occurred in positioning movements, supporting the interfer-
the motor studies is that the intertrial intervals ence theory (see also Stelmach, 1969).
were very long (2 min in Adams & Dijkstra's With respect to retroactive interference, some
study; figure 3.21 on p. 90), possibly providing an earlier researchers failed to find effects of activi-
opportunity for forgetting of an earlier movement ties placed between the presentation and the
before a later movement could be presented. recall of the test movements, casting serious
Ascoli and Schmidt (1969) studied proactive doubt on the application of interference theory
effects by concentrating the prior movements into to memory for movements. But none of these
a short period of time. They presented either zero, studies reported constant errors, and the finding
two, or four positions just prior to the presenta- that proactive interference had its major effects on
tion of a criterion movement (the movement to constant error raised the possibility that retroac-
be remembered). A retention interval of either 10 tive effects would be seen in the same way. In a
or 120 s followed the criterion movement, then reanalysis of earlier data, Pepper and Herman
recall of the criterion movement was attempted, (1970) found that movements produced during
and finally a recall of the preliminary movements the retention interval tended to have negative
(if any) was done. Figure 14.9 presents absolute effects on movement accuracy when measured
errors in recall for the two retention intervals and in terms of constant error. Subsequently, Patrick
for the various numbers of prior movements. (1971) and Milone (1971) also provided evidence
Errors increased as the length of the retention for retroactive interference.
interval increased. But of more interest was the
finding that the four-prior-position condition Cue-Separation Techniques
showed more error than either the zero- or two- What does the performer remember and recall
prior-position condition. A major effect was seen in these positioning tasks? One possibility is that
476 Motor Control and Learning

the person remembers the sensory qualities of condition, the subject was asked to move the
the target position and attempts to match these same distance as in the presentation movement,
sensations through a closed-loop process during so the location of the presentation movement was
the recall movement. That is, the person might be unreliable to the subject for recall.
attempting to move to that position that is recog- Laabs' major findings were that accuracy was
nized as correct (see "Schema Theory," p. 441). far greater in the condition in which the location
Another possibility, however, is that the person cue was recalled than in the one in which the
remembers the distance moved, rather than the distance cue was recalled. Subsequent research
location of the target, and remembers a motor has suggested that subjects have a difficult time
program that will move the limb a certain dis- remembering cues about movement distance
tance. These two possible cues (location vs. and that positioning movements are probably
distance cues) were confounded in the earlier based on some memory of location. However,
experiments on motor short-term memory. How- retroactive-interference effects for location and
ever, Keele and Ells (1972), Marteniuk (1973), and distance information may occur in complex ways
Laabs (1973) used a simple, but clever, method in some instances (Imanaka & Abernethy, 1991,
for unraveling these two potential cues (see figure 1992; Imanaka,Abernethy, & Quek, 1998; Walsh,
14.10 for an illustration). Russell, Imanaka, & James, 1979).
For example, Laabs (1973) had subjects move to
a stop for the presentation of the stimulus materi- The Preselection Effect
als (as in the Adams & Dijkstra study). Then he In the usual paradigm for motor short-term
formed two different conditions for recall. In both memory studies, the subject is asked to move to
of these conditions, subjects began at a different a stop that is defined by the experimenter; thus
starting position for the recall movement. In one the subject does not have any advance knowledge
condition, subjects were asked to recall the same about where the movement end point is located
location on the curvilinear track as before, so the until she contacts the stop. Marteniuk (1973)
distance of the recall movement was different and Stelmach, Kelso, and Wallace (1975) intro-
from that of the presentation movement, render- duced a new method when they asked subjects
ing memory for distance unreliable. In the other to choose their own movement end points. In

FIGURE 14.10 Illustration of the motor short-term memory paradigm used to separate the effects of end-location and dis-
tance cues.
Adapted, by permission, from K. lmanaka, B. Abernethy, and J.J. Quek, 1998, The locus of distance-location interference in movement reproduction: Do we know any more 25
years on? In Motor behavior and human skill, edited by J.P. Piek (Champaign, IL: Human Kinetics), 33.
 Retention and Transfer 477

effect, the instruction was to move to a position ing is similar to the contextual-interference effect
of the subject's choice (a stop was not provided); discussed in chapter 11, suggesting that common
then the subject returned to the starting position underlying factors may be involved.
and was asked to reproduce the position after
a retention interval. This so-called preselection Warm-Up Decrement
method led to much more accurate recall than the
To this point in the chapter, the focus has been
experimenter-defined method. Note that these
on memory losses. But as mentioned earlier, not
findings have some similarity to recently studied
all decrements seen in a retention test are due to
effects in learning when subjects are allowed to
memory losses, as evidenced by such temporary
regulate their own practice schedule (chapter 11)
factors as loss of motivation, day-to-day fluctua-
or augmented feedback presentations (chapter
tions in performance, effects of drugs, and illness.
12). The key commonality may be related to the
Many of these have been discussed with respect
active involvement of the learner in remembering
to the measurement of performance (chapter
and learning processes.
2) and learning (chapter 10), and they are all
When the subject is faced with these repro-
involved in motor retention as well. But a special
duction situations, it is likely that the nature of
kind of decrement in motor performance has a
the methods will influence the way in which the
small literature of its own, and it deserves special
person stores the information. For example, if
mention. This effect is called warm-up decrement.
the person does not know where the target will
The phenomenon can be easily introduced with
be (in the standard paradigm), this could force
an example. Adams (1952, 1961) studied a large
the individual to process sensory cues about the
group of subjects on the pursuit rotor task, pro-
target location, perhaps leading to a strategy
viding thirty-six 30 s trials per day for five days;
wherein the recall of the movement is produced
through closed-loop processes. In the preselection the results are shown in figure 14.11. The typical
method, however, the performer can generate a improvement with practice during a session of
movement plan in advance, perhaps program- trials is seen, but also seen is a relatively large
ming it, and thus can ignore the sensory conse- decrement in performance after each 24 h rest
quences of the movement-simply rerunning period. This decrement appears to be quite severe,
the program at the retention test. This may also and it is equivalent in size to the gains experienced
suggest that memory for programs or parameter- in 5 to 10 trials. It is also rather short-lived, being
izations may be more stable than memory about eliminated in only a few practice trials. The phe-
the feedback for correct locations. nomenon has been known for a long time and
has been found in nearly every motor task that
Spacing of Repetitions has been studied (see Adams, 1961, for a review).
Earlier we presented the findings of the Adams This decrement was thought to be related in some
and Dijkstra (1966) study, in which many rep- way to the need to "warm up" (probably not in
etitions of the movement reduced the loss of the usual sense of warming up the muscles) for
information during the retention interval. These the task again after the rest, and the phenomenon
findings have been replicated often (reviewed in came to be called warm-up decrement. It can be of
Lee & Weeks, 1987), suggesting that a memory potential importance when people are asked to
representation is stronger or more resistant to perform after a rest period, as occurs with the
forgetting with "practice." A curious finding, worker operating a dangerous machine after a
however, is that the repetition effect is enhanced coffee break, the athlete going into the game from
if the repetitions themselves do not occur imme- the bench, or a surgeon's first operation of the day
diately but instead are spaced apart-especially (Kahol, Satava, Ferrara, & Smith, 2009).
so if some interference occurs between these rep- Two major classes of explanation for warm-up
etitions (e.g., Lee & Weeks, 1987; Weeks, Reeve, decrement can be described. A forgetting hypothesis
Dornier, & Fober, 1991). One explanation for this holds that the loss in skill is due to forgetting
spacing effect is that the forgetting that occurs of the type mentioned in the previous sections.
between repetitions actually serves to enhance On the other hand, various versions of the set
memory on the retention test (seep. 379, "When hypothesis argue that the loss in skill is due to a
Forgetting Improves Remembering"). This find- relatively temporary loss of bodily adjustments
478 Motor Control and Learning

Day 1 Day 2 Day 3 Day4 Day5

§: 6
-a;
~5
sc:
0 4
CJ)
E
j:: 3

12 24 12 24 12 24 12 24 12 24
Trials Trials Trials Trials Trials

FIGURE 14.11 Mean performance on the pursuit rotor task for five days. (The decrements in performance from the end of
one day until the beginning of the next are termed "warm-up decrement.")
Reprinted from J.A. Adams, 1961, "The second facet of forgetting: A review of warm-up decrement;' Psychological Bulletin 58: 260.

or states. These views and the evidence for them tion of attention to the feedback channel that is
are contrasted in the following sections. relevant for the task (e.g., vision vs. kinesthesis),
adjustments in emotional state, and many more.
Warm-Up Decrement as Forgetting
According to this view, warm-up decrement is
One major hypothesis, and probably the earliest caused by the loss (or disruption) of these adjust-
and simplest explanation to be considered, is that ments (set) over the rest period. The hypothesis
warm-up decrement is simply another form of says that memory of the skill is not lost over the
forgetting-that is, the loss of memory for the rest period; or perhaps very small memory losses
skill. In this view, the rest period allows certain do occur, but they are far too small to account
forgetting processes to occur, with the initial for the large decrements seen. With practice
phases of these processes being relatively rapid. resumed on the task after the rest, performance
These account for the rather large performance is improved because the internal set (or adjust-
decrements seen with only a few minutes of rest. ments) that supports the skill is reinstated.
The improvements in performance with resumed
practice are, in this view, due to relearning of the Early Evidence on the Set Hypothesis The set
task whose memory was weakened over the rest hypothesis seemed reasonable for many years, as
period. This view does not seem to hold well for it is easy to imagine how such a process might
continuous skills, which as we have discussed, disrupt skills with rest, especially in the face of
are retained well for long periods of time but the nearly perfect retention of skills like those in
also show substantial warm-up decrements. In the pursuit rotor task. Yet no evidence existed for
general, there appears to be little support for a these set-loss phenomena until Irion's (1948) data
memory-loss explanation of warm-up decrement with verbal skills suggested a way to study the
(Stratton, Liu, Hong, Mayer-Kress, & Newell, problem. Irion' s idea was that it should be possi-
2007). ble to reinstate a lost set through certain activities
that are related to the action in question but that
Warm-Up Decrement as a Loss of Set cannot be thought of as contributing specifically
In another view, the loss of skill is related to the to the memory for it. Irion used verbal learning
loss of set-one or more temporary internal states as the main task, with two groups; both practiced
that underlie and support the skill in question. the verbal task, then had a rest, then resumed
Set could consist of postural adjustments, orienta- practice again. One of the groups remained
 Retention and Transfer 479

inactive during the rest period. The other group the activities presented in the 10 min rest period.
engaged in color naming during the end of the One group (Rest) was allowed to rest for 10 min.
rest period-an activity presented on the same Another group (Exp) had 5 min of rest, followed
apparatus and having the same rhythms as the by 5 min of another force-estimation task; this
verbal-learning task but using none of the learned task, though, involved the left arm rather than
items from the main task. If the set hypothesis is the right arm, elbow flexors rather than the grip-
correct, color naming should reinstate the lost ping action, and a different level of force (9 kg).
set produced by the rest, and the initial perfor- So it could not be argued that this task would
mance on the verbal-learning task should be more contribute to the memory of the right-hand grip
accurate than for the group that simply rested. It task. After 18 trials of this task with the same
was. Because color naming cannot be argued to intertrial interval and KR, subjects were shifted
increase memory strength for the verbal task, the immediately to the right-hand grip task for the
implication is that color naming reinstated the retention test.
lost set, in some way preparing the subjects for The absolute errors in the main (right-hand
the upcoming verbal task. gripping) task are shown in figure 14.12 for the
Numerous studies were done to evaluate the two groups before and after the rest period. The
set hypothesis with motor skills, but with few group that simply rested (Rest) for 10 min showed
successes. In one such investigation, Ammons the typical warm-up decrement after the rest;
(1951) used the pursuit rotor; during the rest but the group with the left-hand activities (Exp)
he had subjects watch another active subject or showed very little warm-up decrement, suggest-
follow the target area with the finger, for example, ing that the activities in the rest period reinstated
in an attempt to eliminate warm-up decrement. the lost set. Similar findings have been shown for
No procedures were found that would eliminate a linear-positioning task (with a positioning task
it (see Adams, 1955). These data seemed to say as the warm-up task) by Nacson and Schmidt
that either (a) the set hypothesis was wrong (1971; Schmidt & Nacson, 1971), and by Schmidt
for motor behavior or (b) the appropriate non- and Wrisberg (1971) using a movement-speed
memory-set-reinstating activities had not been task (with another movement-speed task as the
studied. In either case, the set hypothesis was warm-up task). These data also argue against the
not well supported. This evidence is reviewed hypothesis that warm-up decrement is simply
more completely by Adams (1961, 1964) and by forgetting; a forgetting hypothesis cannot explain
Nacson and Schmidt (1971). why a different warm-up task (which seems to
Recent Evidence on the Set Hypothesis Nacson have no memory elements in common with the
and Schmidt (1971) tested the set hypothesis and main task) should produce improvements in
provided considerable support for it. Their idea main-task performance.
was that during practice, various supportive Other data (Schmidt & Nacson, 1971) showed
mechanisms are adjusted constantly so that per- that the reinstated set was rather transient in
formance is maximized; then, during rest, these nature. If as few as 25 s of rest were inserted
functions are adjusted to levels most compatible between the reinstatement of the set and the
with resting, leading to an ineffective pattern of resumption of practice on the main task, the set
adjustment when the task is resumed. Practicing was completely lost again. Also, activities can
a task requiring the same adjustments (set) as be designed that will increase warm-up decre-
the main task just before returning to it should ment even more than resting does. For example,
reinstate those adjustments, leading to a reduc- Schmidt and Nacson (1971) showed that a grip
tion in warm-up decrement, just as Irion (1948) strength task (with maximum force) performed
had found with color naming. just before the resumption of practice on a linear-
The task used by Nacson and Schmidt (1971) positioning task caused a very large increase in
involved a right-hand force production; the sub- error on the first postrest trial, suggesting that
ject had to learn to squeeze a handle with a 21 kg the maximum-grip task required a set that was
force, with KR given after each trial and 10 s rest incompatible with the set for linear positioning.
between trials. After trial 20, a 10 min rest was Other experiments indicate that imagery practice
given, and then practice resumed for another 10 of the task just prior to the resumption of per-
trials. The independent variable was the nature of formance can reduce the warm-up decrement,
480 Motor Control and Learning

20

'E 15
.s.
..
0
Cl)
g
::J 10
0Ul
,Q
IO
c:
IO
Cl)
:ii! 5

5 10 15 20 25 30
Pre rest -------I I~ Postrest ---+-I
Trials

FIGURE 14.12 Absolute error in a force-estimation task for original learning and after a 10 min rest. Group Rest rested
during the interval, and group Exp performed a left-hand force-estimation task; error is measured as a polygraph pen dis-
placement.
Adapted, by permission, from J. Nacson and R.A. Schmidt, 1971, ''The activity-set hypothesis for warm-up decrement;' Journal of Motor Behavior 3: 1-15, adapted with
permission of the publisher (Taylor & Francis Ltd, http://www.informaworld.com).

although the nature of the reduction seems to be professional basketball players before they take
task specific (Ainscoe & Hardy, 1987; Anshel & a free throw. Most players carry out a "preshot
Wrisberg, 1988, 1993; Wrisberg & Anshel, 1993). routine"-a sequence of actions and thoughts that
In sum, these findings suggest that warm-up are specific to each athlete, but done consistently
decrement is caused by some loss of internal by that athlete from shot to shot. It is tempting to
adjustments (or set) over the rest period. These suggest that the preshot routine is a method that
adjustments are critical to effective performance reinstates the set and helps to overcome warm-up
in the task, but they are not a part of the memory decrement (Boutcher & Crews, 1987), and some
for it. Just as a race car needs to attain the proper evidence exists to support the contention (Mack,
temperature before maximal performance can 2001). However, much more research could be
be achieved, so too, it appears, must the human done to more fully investigate the idea.
be brought into the proper state of (temporary)
adjustment for high-level skilled performance. Consolidation
It is not clear exactly what is being adjusted in An old concept in motor learning research, which
these experiments, but probable candidates are dates back well over a century (see McGaugh,
the level of arousal, the rhythm and timing for 2000, for a historical review), has received
the trial cycle, attention to the proper focus and renewed interest in recent years. Much of the
sources of feedback, and so on. current work is being conducted at the cellular
These findings have considerable relevance level of analysis with animals, and is beyond the
for high-level performances, especially after scope of discussion here. However, a significant
performance is interrupted by rest or when major amount of research has also been conducted at the
changes in tasks are required. For example, in behavioral level, with intriguing results.
golf, there are probably different sets for driv- The basic idea is that practice produces a
ing and putting, each of which must be rees- memory for motor skill that is unstable for a
tablished before each shot. Watch professional period of time but that stabilizes, or consoli- /1

golfers before they execute a swing; or watch dates," during a critical period afterward. A fre-
 Retention and Transfer 1111 481

quently used method to examine consolidation is the use of drills, in basketball for example. The
uses a variant of the retroactive-interference instructor usually does not really care whether
paradigm discussed earlier. In this paradigm, one the student can perform these drills, per se, well;
group learns task A, then immediately practices rather, the instructor assumes that, by practicing
a second task (B). Another group undergoes the them, the student will learn something that will
same learning procedures, except that a time transfer to some other task that is of primary inter-
interval is inserted prior to learning task B, which est (e.g., performance in a basketball game). For
presumably allows for the consolidation of task drills to be successful, one must be certain that
A. Retention of task A is measured later for both what is learned in practice on the drill transfers
groups. to performance of the desired criterion task.
This paradigm was used, for example, by Another example is the common method
Walker, Brakefield, Hobson, and Stickgold whereby the task is broken down into its compo-
(2003) to examine the retention characteristics of nents for practice. The assumption is that practice
a finger-sequencing task. Subjects who learned on the parts will transfer to the whole task (see
a different sequence (task B) immediately after chapter 11). Still another example is the use of
practicing an initial sequence (task A) performed simulators of various kinds, such as a pitching
much less skillfully 24 h later in the retention machine to simulate a "real" pitcher in baseball,
trials of task A than subjects who delayed prac- a dummy for training resuscitation skills, or a
tice of task B by 6 h after initial practice of task simulator to duplicate an aircraft cockpit. Does
A. This suggested that the 6 h rest allowed some practice on these simulators result in improved
consolidation of task A, rendering it less vulner- performance on the criterion task-that is, do
able to interference from task B. Moreover, these learning skills using the simulator transfer? The
consolidation effects appear to be larger if the choices about whether or not to use these meth-
consolidation interval includes a period of sleep ods, and about how they should be structured,
(Stickgold & Walker, 2006). if used, depend heavily on an understanding
These findings are not without some contro- of transfer of learning. We consider some of the
versy, however, as failures to find consolidation principles of motor transfer next.
effects appear to be related to task-specific dif-
ferences and experimental design issues (e.g., Basic Principles of Transfer
Criscimagna-Hemminger & Shadmehr, 2008; Many studies using different techniques and
Krakauer & Shadmehr, 2006). These consolidation tasks have produced a vast array of different
effects are also difficult to reconcile with variabil- and sometimes contradictory findings on trans-
ity-of-practice effects discussed in chapter 11, in fer (see Cormier & Hagman, 1987, for a review).
which retention and transfer following practice Two major points emerge from the work on
on a single task are less effective than with practice motor skills. First, the amount of transfer seems
on multiple tasks (e.g., Shea & Kohl, 1991). Nev- to be quite small and positive unless the tasks
ertheless, the renewed interest in these retention are practically identical. Second, the amount of
issues has escalated motor learning research in a transfer depends on the "similarity" between the
number of experimental laboratories, represent- two tasks (Schmidt & Young, 1987).
ing a current "hot topic" in the literature.
Motor Transfer Is Small
When the transfer from one task to a completely
Transfer of Learning different task-sometimes called intertask trans-
fer-is studied, we typically find that the transfer
A number of decisions about the design of prac- is small or negligible. Such evidence comes from
tice sessions are based heavily on an understand- studies concerned with attempts to train some
ing of transfer of learning-the gain (or loss) in behavior or trait in one situation by providing
proficiency in one skill as a result of practice on presumably related experiences in different
some other skill. Often, the task actually practiced situations. For example, investigations by Linde-
in a session is not the activity of primary inter- burg (1949) and Blankenship (1952) showed that
est, the real concern being for some other task "quickening exercises" (various laboratory tasks
believed to be related to this activity. One example that require rapid decision and action) provided
482 !I! Motor Control and Learning

no transfer to other tasks that required quickness. And, as can be seen in figure 14.13, both groups
This is certainly not surprising in light of what required considerable practice on day 2 to achieve
is known about the specificity of motor abilities the same level of performance as attained by the
(see chapter 9), as the activities in the quickening 6~0 group at the end of day 1 practice. Narnikas
exercises probably used different motor abilities and Archer (1960), using the same procedures,
than the task to which the exercises were sup- found somewhat higher transfer, ranging from
posed to have contributed (see "The Myth of 42% to 64%. Remember that in these experiments
General Vision Training" on p. 486). Evidence the transfer is between the pursuit rotor and itself,
suggests that general traits such as quickness, with only the speed of rotation changed to define
balance, and coordination cannot be improved by the different "tasks." It is somewhat surprising
the use of different activities supposedly involv- that the transfer is so small, but numerous other
ing that trait; and we would not expect that an experiments show essentially the same thing.
ability would be improved by practice anyway. These generally small transfer effects seem to fit
What if the tasks are more similar? Here, the with a number of other phenomena that we have
transfer among tasks tends to be higher than for discussed already. First, the transfer findings coin-
the previous situation, but still the amount of cide with the ideas about individual differences.
transfer is typically small. For example, figure An important concept in chapter 9 was that motor
14.13 presents results from Lordahl and Archer abilities are both numerous and specific, and that
(1958). Different groups of subjects practiced the even similar tasks appear to correlate very weakly
pursuit rotor task on one day at 40, 60, or 80 rpm with each other (with the possible exception of
for 30 trials. All groups then switched to the 60 rpm timing skills). If so, then in transfer experiments
version of the task for evaluation of the transfer when the task is changed in even a small way (e.g.,
effects on the next day. The group that had 60 rpm changing the turntable speed of the pursuit rotor),
in both the training trials and the transfer trials was it is likely that different and unrelated abilities are
used as the standard against which the transfer in called into play. Thus, there might be low transfer
the other two groups was assessed (i.e., it served among even very similar tasks because the abilities
the role of group II in figure 14.2). Using the calcu- involved are almost completely different.
lation for the percentage transfer introduced ear- These findings also fit well with the GMP
lier in this chapter, the transfer from the 40 and 80 notion. In chapter 6, a major idea was that two
rpm versions of the task to the 60 rpm version was tasks with different relative timing characteristics
12% and 31 %, respectively, on the very first trial. were assumed to be governed by different GMPs.

70
'lii
~ 60
sc
0 50
fl)
E
:: 40
cfl)
~ 30
a.
c
m20
:e
10

Day 1 Day2
Blocks of 5 trials

FIGURE 14.13 Mean time on target in pursuit tracking. Separate groups practiced on day 1 at speeds of 40, 60, or BO rpm
and transferred to 60 rpm on day 2.
Data from Lordahl and Archer 1958.
 Retention and Transfer 483

If a shift in conditions requires subjects to aban- literature is that motor transfer is still not well
don one GMP in favor of another, then they will understood at all (Schmidt & Young, 1987).
be performing two different GMPs in the two
Negative Transfer
different variations of the" same" motor task. This
is analogous to speeding up a treadmill so that We have mentioned that transfer is not always
jogging is substituted for walking, each activity positive and that losses can occur in one skill as
having its own program (e.g., Shapiro, Zernicke, a result of experience on another. This is called
Gregor, & Diestel, 1981). It is difficult to say how negative transfer. Many people believe that nega-
wide the range of conditions produced by a given tive transfer is relatively common and that the skill
GMP might be, but we suspect that many GMPs losses it produces can be quite large. Almost cliche
exist and that they are shifted rather freely when is the story that tennis in the summer ruined the
the conditions change. Viewed in this way, it is person's badminton game in the winter, presum-
not surprising that the tasks do not transfer to ably because the two tasks are quite similar yet
each other very strongly. somewhat incompatible (e.g., the wrist action in
the two strokes is different). But the research on
Transfer Depends on Similarity transfer nearly always shows low but positive
A second and related concept is that transfer transfer; negative transfer is seldom the outcome.
depends on the similarity of the two tasks being However, negative transfer can be produced if the
considered. The idea of similarity is certainly not proper conditions are presented, such as those
new, as Thorndike (1906) and Woodworth (1901) provided by Lewis, McAllister, and Adams (1951).
proposed that transfer depends on the number Lewis and colleagues used the Mashburn task, in
of "identical elements" that exist in common which a two-dimensional arm control and a foot
between two tasks. If one task had elements that control are operated simultaneously to match
were totally different from the elements in another the positions of lights on a display. After subjects
task, then no transfer would be expected. Transfer practiced for a varying number of trials (either 10,
would be 100% if the two tasks had all their ele- 30, or 50) with the usual configuration of the task,
ments in common. The problem with this theory they were switched to a condition in which the
was that it never specified what an" element" was control-display relationships were reversed. For
and how it could be operationalized, so the theory example, in order to move the light on the display
cannot be put to empirical test. In the previous to the left, the lever had to be moved to the right
paragraphs, the implication is that the" elements" rather than to the left as had been the case before.
could be (a) abilities in common between the two All three dimensions of the task (right-left, back-
tasks, (b) GMPs that are used for the two tasks, or ward-forward, right foot-left foot) were reversed.
(c) both. And other possibilities exist. This is analogous to driving a car in which the
The theories of transfer have been improved "normal" movements of the controls are suddenly
considerably since the publication of this next backward (e.g., steering wheel turned clockwise
idea. A major contribution was Osgood's (1949) to go left, brake pedal released to stop). After 10,
transfer surface, which provided a description of 20, 30, or 50 trials on this reversed task, subjects
the amount of transfer of verbal learning as a joint were switched back to the original configuration
function of the similarity of the stimulus elements of the task to examine whether skill on it had been
and the response elements. Holding (1976) pre- lost or gained. This is a retroactive-transfer design
sented a related idea for motor skills. In all these (as shown in table 14.3).
cases, the notion of similarity is a dominant theme, The differences on the main task between the
as it always has been. But these recent theories are number of matches before and after reversed-
not completely satisfactory, as a large number of task practice are plotted in figure 14.14 (see p.
transfer phenomena do not appear to be explained 485). A decrement score of zero means that the
by them. The problem seems to be related to our standard task was performed just as well after the
lack of understanding about what "similarity" is reversed task as before, meaning that no negative
and what the "elements" are that are supposedly (or positive) retroactive transfer occurred; larger
transferred across various tasks. Perhaps research decrement scores imply more negative transfer.
with abilities and motor programs will contribute Transfer was generally negative, and negative
to this area, but to date this possibility has not transfer increased as the number of reversed-task
been realized. The conclusion from a look at this trials increased. This is what one might expect,
484 Motor Control and Learning

The Myth of General Vision Training

A quicksearch of the Internet will reveal a growing industry that markets various "training programs"
des~gnedto improve vision. Some of these programs make the further claim that improvements in
vislon Will transfer to improvements in performance, most notably sport performance. These claims
are rather impressive, if not surprising, given that the amount of motor transfer between two tasks
is normally smaH and restri.cted to training tasks that are highly similar to the transfer task (e.g.,
Lqrdahl &Archer, 1958; Schmidt & Young, 1987). However, a close look at the "evidence" provided
· iri support of t.hese programs quickly reveals it to be weak, biased, and perhaps even fraudulent.
Sport yision training programs generally make the following claims: (1) Superior athletes have
· superiqrVisuaf.skiHs; (2) visual skills can be improved with training; and (3) visual skills that are trained
in sport vision programs will result in superior sport performance (Abernethy & Wood, 2001 ; Starkes,
Helsen, & Jack, 2001; Williams & Grant, 1999). The first claim, that superior athletes have superior
visual skills, has little to no support. Instead, the evidence suggests that superior athletes often have
a perceptual advantage-that is, experts process specific sport-related perceptual information faster
and more precisely than less skilled athletes (Starkes et al., 2001; Williams & Ward, 2003). Experts are
simileir to less skilled athletes in speed and precision in processing perceptual information (and visual
infbrmationingeneral) that is not specific to the nature of their expertise (Starkes & Ericsson, 2003).
The second claim, that general visual skills can be trained, is misleading. There does appear
to be evidence that some improvement can be gained from general visual skills training, but this
benefit is limited to individuals with visual defects. In their review of the literature, Abernethy and
Wood•(2001} conclude that there is no evidence that visual skills can be improved in athletes as
appears to be the case for individuals with compromised vision.
The last claim, that general visual skills training programs can improve sport performance, appears
highly.suspect or fraudulent. The ''strongest" support is provided by case testimonials, usually by
.athletes who. have undergone the training program. However, testimonials are not experimental
evidence, and any perceived benefit could be due to expected improvements (i.e., a Hawthorne
effect); As we have suggested many times in this book, transfer is a highly selective and specific
process. The.re is no evidence at all, for example, that intensive "training" to respond to a stimulus
light in the mid.st .of a compllcated array will facilitate auto racing performance. Tracking a swinging
ball with ocular and finger pursuit movements will not improve forearm shots in tennis. And, trying
to identify an alphanumeric character presented in a tachistoscopic display will never help a batter
to distinguish between a fastball and a curveball. The conclusion regarding this third claim-that
general Vision training can improve sport performance-appears to be an overwhelming "no!" based
on theory and empirical evidence (Abernethy & Wood, 2001).

as the amount of interference from this reversed not have had much to do with motor negative
task should be larger if it is learned more com- transfer. The reversed conditions probably left
pletely. (There was also an effect of the number of the subjects confused about what to do (which
original-practice trials of the task with standard way to move) and may not have disrupted the
controls, but it is far from clear what this means; motor control processes in the task at all. This
see Schmidt, 1971a, for a more complete discus- argument is not strong, though, as it is difficult
sion of this effect.) This is an example of clear to know what the relevant motor and cognitive
and unmistakable negative retroactive transfer; processes are in such tasks. Yet it seems logical
similar findings have been produced in other to assume that a major portion of the problem
studies using similar procedures (see Lewis, 1953; for the subjects on returning to the standard task
Schmidt, 1971a; Schmidt & Young, 1987). was confusion about what the limbs controlling
However, the negative transfer produced in each of the three dimensions of the task were
these studies seemed mainly cognitive and may supposed to do.
 Retention and Transfer mri 485

18

16

14
UI
Cl)
.c 12
....u
I'll
E 10
.E
E 8
Cl)
E 6
...u
Cl)

Cl)
"O 4
c
I'll
Cl) 2
:a:
0

-2

10 20 30 40 50
Number of reversed-task trials

FIGURE 14.14 Retroactive negative transfer (interference) as a function of amount of practice on the reversed task and the
amount of original practice on the standard task.
Reprinted from D. Lewis, D.E. McAllister, and J.A. Adams, 1951, "Facilitation of interference in performance on the modified mashburn apparatus: I. The effects of varying the
amount of original learning;• Journal of Experimental Psychology 41: 53.

Negative Transfer of Timing Similar effects for learning new coordination-

Some studies suggest, however, that negative timing patterns were described in chapter 13.
transfer of limb control can be quite large. For Strong negative-transfer effects are exerted by
example, Shapiro (1977, 1978) had subjects learn the existing, stable patterns (in-phase and anti-
complex patterns of movements with a particular phase patterns) when one attempts to learn a
relative timing. Later, subjects were instructed to new pattern, such as a 90° relative-phase coor-
speed up the movement, maintaining the same dination (Zanone & Kelso, 1992; Lee, Swinnen,
relative timing, which they had no trouble doing. & Verschueren, 1995). This suggests that some
But when they were told to ignore the temporal pat- negative transfer can result from the experiences
tern they had learned earlier, subjects had a great that subjects bring into the laboratory (i.e., before
deal of difficulty producing a new temporal struc- learning any specific task). Certainly much more
ture. Instead, they sped up the original temporal can be discovered about negative transfer effects
structure, more or less as one would speed up a from this kind of research.
phonograph record. (These studies are discussed Another example involves second-language
in more detail in chapter 6.) This can be seen as a learners; here we consider the production of a
kind of negative transfer, where the prior experi- particular language's speech sounds (but not
ence with the old" temporal structure interfered
/1 its grammar or vocabulary) as a motor skill.
with producing the "new" pattern at maximal Common experience tells us that the difficulty in
speed. This might turn out to be an important producing a particular speech sound in English,
finding for understanding transfer. Schmidt and for example, is critically related to the speaker's
Young (1987) suggest that tasks whose relative first language. The same acoustic goal is often
timing and sequencing are the same will tend to produced differently by speakers whose native
transfer to each other positively; two tasks whose language is French versus German; these dif-
sequencing is the same, but whose timing is differ- ficulties represent negative transfer from French
ent, will tend to transfer to each other negatively; (or German) to English. (One of us, R.A.S., was
most tasks with neither sequencing nor timing in never able to perform the common "ui" sound in
common transfer to each other hardly at all. Dutch [e.g., "bruin"], despite much practice.) If
486 111 Motor Control and Learning

negative transfer were not occurring, these pro- the shot, or the limb actions in the shot itself. The
nunciation difficulties would not be common to point is that whether or not two tasks transfer
a particular language group, and we would not positively or negatively might depend on a kind
expect to find, for example, French accents in of "balance sheet" on which the elements that
English. Yet such accents are clearly differenti- transfer positively are "weighed" against those
ated from German accents, and are remarkably that transfer negatively. This is not an adequate
persistent across many years of speaking English. theory of transfer, but it may help to conceptu-
These phenomena seem to represent some of our alize some of the things that happen when two
strongest evidence for negative transfer. tasks interact.
Finally, it seems reasonable to think that two
tasks, each containing a number of "elements," Simulation and Transfer
may have some similar elements leading to posi- An important and commonly used method for
tive transfer and have other, dissimilar elements training people in motor (and cognitive) tasks is
contributing to negative transfer. In Shapiro's simulation. The main feature of simulations is that
(1977, 1978) studies, at the same time negative they provide a practice task that is (supposedly)
transfer of relative timing occurred, positive related to some criterion task (whose performance
transfer of sequencing might also have been is the overall goal of the learning process) in some
occurring. Other aspects of the task might not way. For example, pilots may practice procedural
transfer at all-positively or negatively. This skills on ground-based devices that mimic the
idea can be seen in many tasks in sports, for cockpit of the airplane, as seen in figure 14.15.
example handball and racquetball. There appear The reasoning is that the practice of these skills
to be many common elements between these two in the simulator will transfer to the actual skills
games, such as the angles that the ball bounces in the airplane (the criterion task). Many aspects
off the walls of the court and the strategies of of simulators were reviewed in Sweezy and
the game, all of which might lead to positive Andrews (2001).
transfer (e.g., Smeeton, Ward, & Williams, 2004).
Yet at the same time, other elements of the game Physical Simulators
would appear to lead to negative transfer, such Many examples of simulators in learning situations
as the exact positioning of the body just before could be mentioned. At one end of the scale are

FIGURE 14.15 The MD-11 flight simulator.

Copyright © Boeing.
 Retention and Transfer 487

expensive and highly sophisticated devices that training (or both), increased safety, and the
simulate large and complex systems (see figure increased convenience of having the simulator
14.15). For example, the simulators for learning available for use at any time in any weather.
to fly are often elaborate, with very detailed and And yet simulators have a number of serious
specific replications of the cockpit area, instrumen- drawbacks. First, the "worth" of any simulation
tation, and so on. The pilot or learner is often given device has to be measured in terms of the amount
simulated displays showing airport runways; the of transfer that it provides to the criterion task.
instrumentation is complete and functioning; and If the simulator does not provide transfer to the
the "feel" of the controls is as identical as possible criterion task, the device is essentially useless in
to that in the real aircraft. In some simulators, even terms of the purpose for which it was originally
movements of the cockpit as a whole simulate the intended. Thus, the evaluation of simulation
effect of control movements and the movements devices usually places heavy emphasis on trans-
of the aircraft in a storm. In these situations, the fer of learning from the device to the criterion task
information displayed on the gauges and dials (Alessi, 1988; Lintern, Sheppard, Parker, Yates, &
is produced by a computer, and the learner's Nolan, 1989; Schendel, Heller, Finley, & Hawley,
responses are monitored as well; these are then 1985; Sturm et al., 2008).
used to move the simulator, its displays, or both. One point that consistently emerged from our
Comparable devices are used to simulate the earlier discussion is that motor transfer is gener-
behavior of a weapons system, and simulators for ally quite small unless the training and criterion
controlling the behavior of nuclear power plants tasks are so similar as to be practically identical.
have been developed. As you might imagine, these From these basic research findings, as well as
devices are very expensive to produce, operate, from the literature on the specificity of individual
and update. differences (chapter 9) and specificity of learning
Some of the early medical simulators were less (chapter 11), it might be predicted that many sim-
expensive and could be used to train procedural ulators will not transfer well to the criterion tasks
skills, such as resuscitation. Low-fidelity manne- for which they were designed. Certainly a critical
quins such as Resusci-Annie were the precursor part of simulator evaluation is the conduct of a
to higher-fidelity simulators that remain in use transfer experiment, perhaps with various ver-
today for resuscitation and many other types sions of the simulator, to evaluate the amount of
of medical diagnosis and treatment training transfer that is actually produced. Transfer should
(Cooper & Taqueti, 2009; Perkins, 2007). The use increase from the simulator to the criterion task
of minimally invasive surgical procedures (e.g., to the extent that the two are similar. Recognition
laparoscopic surgery) seems to require simulators of this fact has led the designers of simulation
for training; these kinds of simulators are by now devices to make them very realistic-for example,
quite common and generally supported in the the simulated airplane cockpit that moves as the
medical community (Sturm et al., 2008). actual aircraft would if it were in a storm. Much
At the other end of the scale, simulation devices effort is devoted to making the controls feel as
can be made that are relatively simple and inex- they do in the airplane, with proper resistances,
pensive. Many of us learned to drive a car by prac- feedback, and so on to maximize the similarity.
ticing on driver simulators that had not-so-realistic This makes good sense. If differences between
configurations of an automobile's controls, so that simulator and criterion task are too great, it is
we could learn the proper motions before we tried possible that separate motor control mechanisms
them in a real car. Some dental schools still use might be learned in the two situations, producing
plaster-of-Paris models of the jaw-dentists-in- no transfer to the criterion task.
training practice dental skills with the "jaw" on a Simulation devices are usually excellent for
workbench or even in the position it would be in teaching procedural skills, the proper order of a
if it were the upper jaw of a patient. A simulator sequence of activities, and the like. These aspects
can require almost no apparatus; for example, you of the overall task are important, and consider-
can practice golf putting on a living room rug with able time can be "saved" by using simulators at
a glass lying on its side on the floor. early stages of practice, as sequence knowledge
Physical simulators provide a number of appears to be transferable between different effec-
advantages, such as decreased cost or time of tor systems (Fendrich, Healy, & Bourne, 1991;
488 Motor Control and Learning

Keele, Jennings, Jones, Caulton, & Cohen, 1995). student with visual, auditory, and haptic feedback
There is less certainty that the motor elements of as a cut is made through bone or other tissue,
the task are so easily simulated, however. thereby providing numerous sources of aug-
Simulations are often applied rather blindly mented information during training. Although
without regard for the kinds of transfer that will researchers still await conclusive evidence about
be produced. Many examples are seen in athletics, how to optimize virtual reality training (Fialkow
in which certain kinds of behaviors are simulated & Goff, 2009; van der Meijden & Schijven, 2009),
in various drill procedures. The use of blocking we suspect that the attempt to make training as
dummies in American football may be helpful in task specific as possible can only be a positive
the early stages of learning a play when the athletes advance for motor transfer. That said, we know
have questions about where to go and whom to from the literature on augmented feedback (chap-
block, but there would seem to be little utility in ter 12) that complex and sophisticated feedback
using them beyond this point. Players would seem in simulators can be detrimental for learning if it
to require practice in blocking other players who is not used appropriately.
do not wish to be blocked; this is, of course, very In contrast to expensive, high-fidelity training
difficult to simulate. It is difficult to evaluate the simulators, some researchers and practitioners are
effectiveness of these various procedures because now using commercially available hardware and
we have no research about the transfer of these software to explore skill transfer. For example, the
drills to game situations. Our guess is that the faith Nintendo Wii is a hugely popular gaming system
placed in many of these procedures is probably that combines many different types of part- and
overdone. Certainly it would make sense to exam- whole-body movements together with interactive
ine any such drills or simulations very carefully. visual and haptic feedback experiences. Physical
therapists, for example, have employed the Wii to
Virtual Simulators motivate active participation in movement-related
In contrast to the traditional type of physical activities, and have reported positive effects for an
simulators for use in training new skills are simu- individual with cerebral palsy (Deutsch, Borbely,
lators that use computer-based technologies to Filler, Huhn, & Guarrera-Bowlby, 2008). And in
train the perceptual-motor attributes of criterion a rather surprising finding of transfer general-
tasks. Virtual environments often simulate the per- ity, extensive video gaming experience appears
ceptual (visual, auditory, and haptic) demands of to be causally related to enhancements in visual
a task, together with a simulated effector system, attention (Dye, Green, & Bavelier, 2009; Green
and display these on a computer monitor. The & Bavelier, 2003, 2006). The capability for using
actions of a subject can be mapped in terms of inexpensive yet sophisticated gaming systems in
the actions of the simulated effector system, with these studies represents an exciting new develop-
the expected (computer generated) consequences ment for future skills transfer research.
displayed. One advantage of such devices is that
they are much less costly to produce than many
of the physical simulators already discussed. Summary
And, once developed, these computer programs
should be modifiable so that newer versions need Learning, memory, retention, and transfer are
not be built again from scratch. very closely related concepts. Motor memory
In recent years, one of the fields that has been is the persistence of the acquired capability for
developing virtual environments the fastest is the responding, and losses in memory are called
medical field. Arnold and Farrell (2002) critically forgetting. Forgetting is usually measured by
reviewed the early evidence and concluded that performance losses on a retention test, admin-
virtual reality was, at the time, unverified as a istered after a retention interval. Different mea-
positive training aid for surgical motor skills. sures of retention can be computed, although the
The potential for positive motor skills transfer, absolute-retention measure is the most useful.
however, has been elevated by the use of robotic A variant of the learning experiment is the
devices that provide simulated haptic and pro- transfer experiment, in which the effect of prac-
prioceptive feedback (see figure 2.16, p. 44). For ticing one task on the performance of some other
example, robotic devices can provide the medical (criterion) task is evaluated. Transfer is often mea-
 Retention and Transfer l!li 489

sured as a percentage, indicating the proportion Student Assignments

of performance improvement in one task that was
achieved by practice on the other task. Studies
of transfer are important for evaluating training, 1. Prepare to answer the following questions
simulation, and other instructional issues. during class discussion:
Continuous skills are retained nearly perfectly a. Using practical examples of discrete
over long retention intervals. Discrete skills, on the and continuous skills, illustrate the
other hand, can show marked performance losses differences in expected retention char-
during the same retention intervals. The reasons acteristics.
for this difference in retention are not clear, but b. Describe three workplace examples in
they are probably not based on the tendency for which warm-up decrement might be
continuous tasks to be more "motor" than discrete expected to occur after a lunch break.
tasks. Perhaps the difference has its basis in the
idea that continuous tasks, with more practice time c. Suggest a computer simulation game
in a typical experiment, are more resistant to for- that could be used to train physicians
who are learning a microsurgery tech-
getting because they are learned more completely.
nique. Describe three key features of
The loss of information related to motor perfor-
the simulation that should be particu-
mance can occur in various possible ways. Infor-
larly effective for learning.
mation might decay from memory due to a pas-
sive process, or might be lost due to retroactive or 2. Find a research article that was designed to
proactive interference. Warm-up decrement is a examine the short-teim retention character-
retention loss caused by the imposition of a short istics of movement information.
rest in a series of practice trials. Research supports
the set hypothesis to explain it, which holds that
warm-up decrement is a loss, during rest, of a Web Resources
pattern of temporary nonmemory adjustments
critical to performance. Consolidation of motor This Web site provides a history of virtual reality:
memories, a field of study that has recently http://archive.ncsa.illinois.edu/Cyberia/
reemerged, suggests that the interfering effects VETopLevelsNR.History.html
of learning a competing task are time dependent.
Two basic principles of transfer are (a) that More on virtual reality applications:
motor transfer is usually small but positive and
http://human-factors.arc.nasa.gov/web/
(b) that motor transfer depends on the similarity
hf101 /reference.html
between tasks. Considerable difficulty exists in
understanding the underlying basis of similar-
ity, however. Negative transfer can be produced Notes
under certain conditions, but it is probably mostly
cognitive in nature. Devices such as simulators 1 Robert Bjork tells us that Stanford University's first
and virtual environments provide promise for president, David Starr Jordan, who was an ichthyologist
positive transfer, although their value seems to (i.e., he studied fish), once said that every time he learned
be highly specific to the similarity between the the name of a new student he forgot the name of a fish-a
training and transfer tasks. clear example of retroactive interference.
 APPENDIX
logarithms to the Base 2
Number .0 .1 .2 .3 .4 .5 .6 .7 .8 .9
0. -3.32 -2.32 -1.74 -1.32 -1.00 -0.74 -0.51 -0.32 -0.15
1. 0.00 0.14 0.26 0.38 0.49 0.58 0.68 0.77 0.85 0.93
2. 1.00 1.07 1.14 1.20 1.26 1.32 1.38 1.43 1.49 1.54
3. 1.58 1.63 1.68 1.72 1.77 1.81 1.85 1.89 1.93 1.96
4. 2.00 2.04 2.07 2.10 2.14 2.17 2.20 2.23 2.26 2.29
5. 2.32 2.35 2.38 2.41 2.43 2.46 2.49 2.51 2.54 2.56
6. 2.58 2.61 2.63 2.66 2.68 2.70 2.72 2.74 2.77 2.79
7. 2.81 2.83 2.85 2.87 2.89 2.91 2.93 2.94 2.96 2.98
8. 3.00 3.02 3.04 3.05 3.07 3.09 3.10 3.12 3.14 3.15
9. 3.17 3.19 3.20 3.22 3.23 3.25 3.26 3.28 3.29 3.31
10. 3.32 3.34 3.35 3.36 3.38 3.39 3.41 3.42 3.43 3.45
11. 3.46 3.47 3.49 3.50 3.51 3.52 3.54 3.55 3.56 3.57
12. 3.58 3.60 3.61 3.62 3.63 3.64 3.66 3.67 3.68 3.69
13. 3.70 3.71 3.72 3.73 3.74 3.75 3.77 3.78 3.79 3.80
14. 3.81 3.82 3.83 3.84 3.85 3.86 3.87 3.88 3.89 3.90
15. 3.91 3.92 3.93 3.94 3.94 3.95 3.96 3.97 3.98 3.99
16. 4.00 4.01 4.02 4.03 4.04 4.04 4.05 4.06 4.07 4.08
17. 4.09 4.10 4.10 4.11 4.12 4.13 4.14 4.15 4.15 4.16
18. 4.17 4.18 4.19 4.19 4.20 4.21 4.22 4.22 4.23 4.24
19. 4.25 4.26 4.26 4.27 4.28 4.29 4.29 4.30 4.31 4.31
20. 4.32 4.33 4.34 4.34 4.35 4.36 4.36 4.37 4.38 4.39
21. 4.39 4.40 4.41 4.41 4.42 4.43 4.43 4.44 4.45 4.45
22. 4.46 4.47 4.47 4.48 4.49 4.49 4.50 4.50 4.51 4.52
23. 4.52 4.53 5.54 4.54 4.55 4.55 4.56 4.57 4.57 4.58
24. 4.58 4.59 4.60 4.60 4.61 4.61 4.62 4.63 4.63 4.64
25. 4.64 4.65 4.66 4.66 4.67 4.67 4.68 4.68 4.69 4.69
26. 4.70 4.71 4.71 4.72 4.72 4.73 4.73 4.74 4.74 4.75
27. 4.75 4.76 4.77 4.77 4.78 4.78 4.79 4.79 4.80 4.80
28. 4.81 4.81 4.82 4.82 4.83 4.83 4.84 4.84 4.85 4.85
29. 4.86 4.86 4.87 4.87 4.88 4.88 4.89 4.89 4.90 4.90
30. 4.91 4.91 4.92 4.92 4.93 4.93 4.94 4.94 4.94 4.95
31. 4.95 4.96 4.96 4.97 4.97 4.98 4.98 4.99 4.99 5.00
32. 5.00 5.00 5.01 5.01 5.02 5.02 5.03 5.03 5.04 5.04
33. 5.04 5.05 5.05 5.06 5.06 5.07 5.07 5.07 5.08 5.08
34. 5.09 5.09 5.10 5.10 5.10 5.11 5.11 5.12 5.12 5.13
35. 5.13 5.13 5.14 5.14 5.15 5.15 5.15 5.16 5.16 5.17
36. 5.17 5.17 5.18 5.18 5.19 5.19 5.19 5.20 5.20 5.21
37. 5.21 5.21 5.22 5.22 5.22 5.23 5.23 5.24 5.24 5.24
38. 5.25 5.25 5.26 5.26 5.26 5.27 5.27 5.27 5.28 5.28
39. 5.29 5.29 5.29 5.30 5.30 5.30 5.31 5.31 5.31 5.32
40. 5.32 5.33 5.33 5.33 5.34 5.34 5.34 5.35 5.35 5.35

Note. To find the Log, (23.5), for example, enter the row labeled 23, then move to the right under the column headed .5; the result is 4.55.

491
 GLOSSARY

abilities-Stable characteristics or traits, genetically average knowledge of results-A type of summary-KR

defined and unmodifiable by practice or experience, that method that presents the results of two or more trials as
underlie certain skilled performances. a statistical average.
absolute constant error (I CE I )-The absolute value of average velocity-The speed of a movement, or the move-
CE for each subject; a measure of amount of bias without ment distance divided by the movement time.
respect to its direction. bandwidth KR-Tolerance limits on errors that define
absolute error (AE)-The average absolute deviation of a when to provide qualitative or quantitative KR.
set of scores from a target value; a measure of overall error. Bernstein perspective-A view of motor control and
absolute frequency of knowledge of results-The abso- learning that originated with the work of Russian physi-
lute number of presentations of KR given in a sequence ologist Nikolai Bernstein (1896-1966).
of trials. bit-The amount of information required to reduce the
absolute retention-A measure of retention based on the original amount of uncertainty by half; short for binary
level of performance on the retention test. digit.
accumulated feedback-Information presented after a blocked practice-A practice sequence in which all of
series of movements that represents a summary of those the trials on one task are done together, uninterrupted
performances. by practice on any of the other tasks; low contextual
action-centered interference-A view of attention that interference.
localizes interference effects at the response-selection stage. capacity interference-Interference between tasks caused
Adams' theory-A closed-loop theory of motor learning by limitations in attention.
proposed by Adams (1971), focusing heavily on the learn- ceiling effect-A limitation, imposed either by the scor-
ing of slow positioning movements. ing system or by physiological-psychological sources,
adjacent-trial effect-With intertrial correlation matrices, that places a maximum on the score that a performer can
the tendency for the correlations between adjacent trials achieve in a task.
to increase with practice. central pattern generators-Mechanisms in the spinal
alpha motor neuron-Large efferent neuron responsible cord, capable of providing oscillatory behavior, thought to
for innervation of the extrafusal fibers of the skeletal be involved in the control of locomotion and other tasks.
musculature. change blindness-A failure to perceive changes in the
anti-phase-A coordination pattern in which two move- environment as a consequence of directing attention to
ment components oscillate in 180° relative phase. another object or event.
arousal-An internal state of alertness or excitement. changing component abilities hypothesis-The hypoth-
associative phase-The second of three phases of learn- esis that the set of abilities underlying a skill shifts sys-
ing proposed by Fitts, in which learners establish motor tematically as practice continues.
patterns. choice reaction time-RT for a task in which the response
attention-A limited central resource that can be allocated to be made is dependent on the stimulus presented.
to a few processes at a time. chunking-The combining of individual elements in
augmented feedback-Feedback added to that typically memory into larger units.
received in the task (also called extrinsic feedback). clock variance-The variance in timing due to central time-
automatic processing-Information processing that is keeping processes; from the Wing-Kristofferson model.
relatively fast, that is done in parallel with other processes, closed-loop system-A control system employing feed-
and that requires minimal effort and attention (compare back, a reference of correctness, a computation of error,
with controlled processing). and subsequent correction in order to maintain a desired
autonomous phase-The third of three phases of learning state; sometimes called a servomechanism or servo.
proposed by Fitts, in which learners have greatly reduced closed skills-Skills that are performed in stable or pre-
the attention demands of the task. dictable environmental settings.

493
494 r11 Glossary

cocktail party problem-The phenomenon, described pressure cuff, rendering the afferent neurons anoxic so
by Cherry (1953), whereby humans can attend to a single that they cannot deliver sensory information.
conversation at a noisy gathering, neglecting other inputs. current or contemporary control-Woodworth's idea
cognitive phase-The first of three phases of learning that the later portions of a movement were controlled
proposed by Fitts, in which learners' performances are by a feedback-based "homing in" process that allowed a
heavily based on cognitive or verbal processes. target to be achieved.
cognitive psychology-A psychological tradition in deafferentation-Eliminating, usually by surgery (dorsal
which the behavior of the individual is described in terms rhizotomy), the sensory input to the spinal cord while
of information processes. leaving efferent output intact.
component interaction-A characteristic of some tasks degrees of freedom-The number of separate indepen-
in which the adjustment on one part of the task requires dent dimensions of movement in a system that must be
an adjustment of some other part. controlled.
concurrent feedback-Feedback that is presented simul- degrees of freedom problem-The difficulty in explain-
taneously with an action. ing the simultaneous control of multiple, independently
consciousness-The mechanism or process by which moving body parts.
humans are aware of sensations, elements in memory, deliberate practice-Identified by Ericsson as practice
actions, or internal events. that is not inherently enjoyable and is undertaken for the
constant error (CE)-With respect to sign, the average sole purpose of improving performance.
error of a set of scores from a target value; a measure of
differential approach-That approach to the study of
average bias.
behavior that focuses on individual differences, abilities,
contextual interference-The interference in performance and prediction.
and learning that arises from performing one task in the
difficulty-Depending on the particular paradigm, either
context of other tasks.
the ratio of the amplitude to the target widths (Fitts, 1954)
continuation-A task in which the subject's goal is to or the ratio of the movement amplitude to MT (Schmidt,
continue periodic finger taps after the removal of a pacing Zelaznik, & Frank, 1978; Schmidt, Zelaznik, Hawkins,
signal, usually following a series of synchronization taps. Frank, & Quinn, 1979).
continuous skills-Skills that appear to have no recogniz- directional preference-In bimanual coordination, the
able and inherent beginning or end. tendency to synchronize the timing of two limbs in the
controlled processing-Information processing that is same direction in a plane of motion.
relatively slow, that is done serially with other processes, discrete skills-Skills that have an inherent and definite
and that requires effort and attention (compare with beginning and end.
automatic processing).
discrimination reaction time-RT for a task in which a
coordination-Behavior of two or more degrees of free-
number of stimuli can be presented, but where a response
dom in relation to each other to produce skilled activity.
is made only if a particular stimulus occurs.
correct rejection-An outcome that results when the
distributed practice-A sequence of practice and rest
operator correctly decides that a signal was absent; from
periods in which the time in practice is often equal to or
signal-detection theory.
less than the time at rest.
correlation coefficient (r)-A statistical measure of the
degree of linear association between two variables. dorsal rhizotomy-The cutting of the dorsal roots at
various segmental levels of the spinal cord, resulting in
cost-benefit analysis-A method by which the benefits deafferentation from the associated areas of the body.
from anticipating correctly can be weighed against the
"cost" of anticipating incorrectly. dorsal root-The collection of nerve fibers from the
periphery into a bundle near the posterior side of the
criterion-A cutoff point set by an operator for deciding spinal cord at each spinal level; the major peripheral
when a signal will be judged as present; related to beta sensory input to the cord.
in signal-detection theory.
dorsal visual stream-Processing involving the posterior
criterion variable-In studies of prediction, the variable
parietal cortex; provides information for the visual control
or score that is predicted from the predictor variables;
of movement; sometimes termed ambient vision.
the "best" obtainable measure of the construct that is to
be predicted. d-prime (d')-A measure of sensitivity in decision
making; from signal-detection theory.
Crossman-Goodeve theory-A theory about Fitts' law
that assumed a series of constant-duration movements, dynamic-pattern theory-A view that describes coordi-
each interspersed with feedback-based corrections; an nation as a self-organizing process of pattern formation.
intermittent-control theory of rapid movement. early responding-Processing all of the aspects of a
cuff technique-A method of temporary deafferentation movement in advance so that the movement can occur
in which blood flow to the limb is eliminated by a blood at or before the stimulus.
 Glossary 495

ecological viewpoint-A point of view emphasizing the feedback-Sensory information that results from move-
study of movement in natural environments. ment.
effective target width (W,)-The size of the target area feedforward control-The sending of information ahead
that the performer actually uses in a series of aiming to ready a part of the system for incoming sensory feed-
movements, calculated as the standard deviation of the back or for a future motor command.
movement end points. Fitts' law-Mathematical description of the speed-accu-
effector anticipation-Predicting the duration of internal racy trade-off in which the average MT is linearly related
processes and subsequent external movement so that to Log2 (2A/W).
some portion of the movement can be made coincident floor effect-A limitation, imposed either by the scor-
with some anticipated external event. ing system or by physiological-psychological limits,
egocentric preference-In bimanual coordination, the that places a minimum on the score that a performer can
tendency to synchronize the timing of two limbs toward achieve in a task.
and away from the center of the body. force variability-The within-subject variability in a
elaborative processing hypothesis-A view of random- series of forces produced either in static or in dynamic
blocked practice effects that emphasizes the comparative contractions.
and contrastive value of tasks in short-term memory. foreperiod-The interval between a warning signal and
electromyography (EMG)-Recording of the electrical the presentation of the stimulus to respond.
activity from muscles. forgetting-The loss of memory, or the loss of the
empirical equation-An equation describing the outcome acquired capability for responding.
of an experiment in which a functional mathematical functional magnetic resonance imaging (fMRI)-A tech-
relationship is estimated from empirical observations. nique used to detect areas of the brain that have increased
ensemble-The combination of the various sources of blood flow, indicating active information processing.
sensory information that enable accurate perception of functional stretch reflex-A stretch reflex with a latency
movement and position. of from 50 to 80 ms, modified by instruction and mediated
equilibrium point-For a given level of muscle activa- in higher brain centers; sometimes called the long-loop
tion, the hypothetical joint angle at which the torques from reflex.
the two opposing muscle groups are equal and opposite. gain-The relationship between the amount of input to a
equilibrium-point models (ex and A.)-Limb-control system and the output produced by it; usually expressed
models in which a movement end point is produced as a ratio.
through the specification of an equilibrium point between gamma motor neurons-Small efferent neurons that
the agonist and the antagonist muscle groups. innervate the intrafusal muscle fibers of the muscle
ergonomics-The study of human beings in work envi- spindle.
ronments (also called human factors). gearshift analogy-An idea about the learning of motor
error in execution-An error in which the planned programs, analogous to learning to shift gears in an
spatial-temporal goal of a movement is appropriate, but automobile.
the movement deviates from the desired path because of generalized motor program-A motor program whose
factors occurring during execution. expression can be varied depending on the choice of
error in selection-An error in which the chosen spa- certain parameters.
tial-temporal goal is inappropriate given the nature of general motor ability-An early concept in which a single
the environment. ability was thought to account for major portions of the
expected sensory consequences-A construct in schema individual differences in motor behavior.
theory; the anticipated feedback sensations that should goal setting-A motivational technique in which subjects
be received if the movement is correct. are encouraged to set performance goals.
extrafusal fibers-The muscle fibers of the major skeletal Golgi tendon organs-Small stretch receptors located
muscles, exclusive of the fibers in the muscle spindles. at the musculotendinous junctions, providing precise
factor analysis-A complex statistical procedure wherein information about muscle tension.
a large number of separate tests are grouped into a smaller guidance-Techniques in which the behavior of the
number of factors, each of which is thought to represent learner is limited or controlled by various means to
an underlying ability. prevent errors.
factor loading-In factor analysis, the statistical values habit-The acquired capability for moving; an unobserv-
indicating the extent to which the tests measure the vari- able internal state that underlies skilled performance.
ous factors. Haken-Kelso-Bunz (HKB) model-A self-organizational
false alarm-An outcome that results when the operator model of coordination as patterns of degrees of freedom
incorrectly decides that a signal was present, when in that spontaneously organize, dissolve, and reorganize as
truth the signal was absent; from signal-detection theory. a function of environmental interactions.
496 111 Glossary

Hick's law-A mathematical statement that choice RT is interneurons-Neurons originating and terminating
linearly related to the Log2 of the number of stimulus- wholly within the spinal cord that connect various seg-
response alternatives, or to the amount of information ments of it; some are thought to be involved in the spinal
that must be processed in order to choose a response. generators.
hierarchical control model-The idea that with practice, intertrial correlation matrix-A table or matrix of cor-
the control of the response shifts systematically from relations among scores on all pairs of trials in a practice
attention-demanding higher levels to less attention- sequence.
demanding motor program levels. intertrial interval-The interval of time between one
hit-An outcome that results when the operator correctly movement and the next in the KR paradigm.
decides that a signal was present; from signal-detection intrafusal fibers-The small muscle fibers lying at the
theory.
polar ends of the muscle spindle.
impulse-From physics, the aggregate of forces applied
invariant characteristic-The relationship between joint
over time; the area under a force-time curve, or the inte-
position and joint torque established by the central ner-
gral of force over time.
vous system.
impulse-timing model-A model of motor program-
inverted-U principle-A description of the relationship
ming in which movement trajectory is controlled by
between arousal and performance that resembles an
impulses that determine the amplitude and timing of
inverted U when graphed.
applied forces.
ironic effect-A result in which the specific intention
impulse-variability theory-A theory of rapid actions
to avoid a motor behavior produces that very behavior.
in which the variability in the muscular impulses leads
directly to the variations or errors in movement control. joint receptors-Common term for a number of different
inattention blindness-A failure to perceive objects in receptors that are located in the joint capsules, presumably
the environment as a consequence of directing attention providing information about joint position.
to other objects or events. kinematic feedback-Feedback about the movement
index of difficulty (ID)-In Fitts' law, the Log2 (2A/W), characteristics or movement pattern produced.
or the theoretical "difficulty" of a movement. kinetic feedback-Feedback about the force character-
individual differences-Stable, enduring differences istics of a movement.
among individuals on some variable or task. knowledge of performance (KP)-Augmented feedback
individual differences in learning-Stable, enduring based on some aspect of the movement's kinematics.
differences among individuals in the amount of or rate knowledge of results (KR)-Augmented feedback
of skill acquisition. related to the nature of the result produced in relation to
information-The content of a message that serves to the environmental goal.
reduce uncertainty. KR delay-The interval between the production of a
information-processing viewpoint-The study of move- movement and the presentation of KR.
ment in which the human is viewed as a processor of law of practice-The finding that the log of the perfor-
information, focusing on storage, coding, retrieval, and mance measure tends to change linearly with the log of
transformation of information. the amount of practice.
inherent feedback-That feedback normally received lead-up tasks-Tasks or activities that are typically pre-
in the conduct of a particular task (also called intrinsic sented to prepare learners for a more important or more
feedback). complex task or activity.
inhibition of return (IOR)-A delay in responding to a learning-A set of internal processes associated with
previously cued (orienting) stimulus, when the SOA is practice or experience leading to relatively permanent
300 ms or longer. changes in the capability for skill.
initial adjustment-Woodworth's term for the initial learning curve-A label sometimes applied to the perfor-
open-loop portion of an aiming movement. mance curve, in the belief that the changes in performance
initial conditions-A construct in schema theory; the mirror changes in learning.
nature of the task and environment prior to the produc- learning score-A difference score, computed as the dif-
tion of a movement. ference between the initial and final levels of performance
in-phase-A coordination pattern in which two move- on a task; sometimes used in estimating the amount of
ment components oscillate in 0° relative phase. change in performance as a result of practice.
intercept (a)-One of the constants for linear empirical learning variable-An independent variable that affects
equations; the value on the Y-axis when Xis zero. learning.
interference theory-A theory that forgetting is caused length-tension diagram-A graph of the tension pro-
by interference from other learned materials. duced by a contracting muscle as a function of its length.
 Glossary 111 497

Log2 (N)-The power to which the base, 2, must be raised motor development-A field of study concerning the
to achieve N. changes in motor behavior occurring as a result of growth,
long-loop reflex-A stretch reflex with a latency of from 50 maturation, and experience.
to 80 ms, modified by instructions and mediated in higher motor learning-A set of internal processes associated
brain centers; sometimes called the functional stretch reflex. with practice or experience leading to relatively perma-
long-term memory-A functionally limitless memory nent changes in the capability for motor skill.
store for abstractly coded information, facts, concepts, motor memory-The memory for movement or motor
and relationships; presumably storage for movement information.
programs. motor neuron pools-Collections of alpha motor neuron
magnet effect-Identified by von Holst as the tendency cell bodies in the gray matter of the cord that serve motor
of one effector's rhythmic oscillation to become interde- units in the same, or anatomically related, muscles.
pendent with that of another oscillating limb. motor program-An abstract representation that, when
maintenance tendency-Identified by von Holst as the initiated, results in the production of a coordinated move-
tendency of one effector to maintain an independent ment sequence.
rhythmic oscillation while another effector is oscillating. motor reaction time-The interval between the first
massed practice-A sequence of practice and rest periods change in EMG and the movement's initiation.
in which the rest time is much less than the practice time. motor variance-The variance in timing due to motor-
memory-The persistence of habit; the acquired capabil- implementation processes; from the Wing-Kristofferson
ity for skill. model.
memory trace-A construct in Adams' closed-loop movement-Changes in joint angles, the position of the
theory; a modest motor program for determining and entire body, or both.
initiating the movement. movement outcome-A construct in schema theory;
mental practice-A practice method in which perfor- the result of the movement in the environment, usually
mance on the task is imagined or visualized without overt signaled by intrinsic feedback or KR.
physical practice. movement time (MT)-The interval between the initia-
mesencephalic preparation-A surgical preparation in tion of a movement and its termination.
which the spinal cord is cut at the midbrain, essentially multiple correlation-A statistical procedure in which
separating higher centers from the spinal cord; the Shik the weightings of predictor variables are adjusted so that
preparation. their (weighted) sum correlates maximally with some
miss-An outcome that results when the operator incor- criterion variable.
rectly decides that a signal was absent, when in truth the muscle spindle-Small spindle-shaped structures,
signal was present; from signal-detection theory. located in parallel with the extrafusal fibers, that provide
modeling-A technique for demonstrating the task to information about muscle length.
be practiced. negative transfer-The loss in capability for one task
modular-A view of individual differences that organizes as a result of practice or experience in some other task.
brain activities in terms of functions (such as timekeeping) objectivity-The aspect of measurement related to the
rather than tasks. extent to which two observers assign the same score.
moment of inertia-A physical quantity defining an occlusion method-Technique used to assess perceptual
object's resistance to rotational forces; the mass of the skill by deleting selective spatial or temporal information
object multiplied by the square of the distance of the in a display.
center of the object's mass from the point of rotation. open-loop system-A control system with prepro-
monosynaptic stretch reflex-A segmental reflex pro- grammed instructions to a set of effectors; it does not
duced by stretch of a muscle and its spindles connecting use feedback information and error-detection processes.
monosynaptically with the alpha motor neurons of the operant techniques-Methods for learning in which
same muscle; it has a latency of about 30 to 50 ms in certain behaviors are reinforced or rewarded, leading to
humans. an increase in the probability that they will occur again.
motivation-An internal state that tends to direct or optimized-submovement model-A view of the speed-
energize the system toward a goal. accuracy trade-off that optimizes the duration of an initial
motor behavior-An area of study stressing primarily impulse and, if necessary, one or more corrective impulses.
the principles of human skilled movement generated at parallel processing-A type of information processing
a behavioral level of analysis. in which at least two processes occur simultaneously.
motor control-An area of study dealing with the under- parameter-A value specified to the generalized motor
standing of the neural, physical, and behavioral aspects program that defines the particular expression of the
of biological (e.g., human) movement. pattern of activity.
498 111 Glossary

part practice-The learning technique in which the task selected movements is stronger than for experimenter-
is broken down into its parts for separate practice. selected movements.
perceptual anticipation-Anticipation of the arrival of a proactive interference-In the interference theory, a
signal through internal mechanisms or processes. source of forgetting caused by learning imposed before
perceptual narrowing-The focusing of attention so the original learning of some to-be-remembered task.
that specific sources of information are more likely to be probe technique-A secondary-task method that uses
received but rare, or peripheral, events are more likely RT to assess the attention demands of some primary task.
to be missed. progression-regression hypothesis-The idea that
perceptual trace-A construct in Adams' closed-loop learning produces a progression to more complex control
theory; a reference of correctness in memory that has strategies and that stress or forgetting produces a regres-
been learned from feedback at the correct target position. sion to more simple levels.
performance curve-A plot of the average performance psychological refractoriness-The delay in the response
of a group of subjects for each of a number of practice to the second of two closely spaced stimuli.
trials or blocks of trials. random practice-A practice sequence in which the tasks
performance variable-An independent variable that being practiced are ordered (quasi-) randomly across
affects performance temporarily. trials; high contextual interference.
perturbation-An unexpected physical event that reachlgrasp action-An action that coordinates the limb
changes the movement or the movement goal. transport component with the opening and closing of the
phase transition-An abrupt shift from one coordination grasp component.
pattern to another. reaction time (RT)-The interval between the presenta-
phasing-The temporal structure of a sequence, usually tion of an unexpected stimulus and the initiation of a
measured by the ratios of element durations and the response.
overall movement duration; relative timing. recall schema-A construct in schema theory; the
point-light display-Technique used to assess perceptual relationship among past parameters, past initial condi-
identification based on limited information. tions, and the movement outcomes produced by these
combinations.
point-to-point computation-Models of limb control in
which the coordinates of each point in a limb's trajectory receptor anticipation-Anticipation of the arrival of a
are achieved sequentially by the motor system at the time stimulus based on sensory information indicating its
of response execution. time of arrival.
polyrhythm-The rhythm produced when two effectors recognition schema-A construct in schema theory; the
simultaneously produce their own, nonharmonic rhythms relationship among past initial conditions, past movement
(e.g., three beats with one finger combined with two beats outcomes, and the sensory consequences produced by
of another finger). these combinations.
positive transfer-The gain in capability on one task reconstruction hypothesis-A view of random-blocked
as a result of practice or experience on some other task. practice effects that emphasizes the role of previously for-
gotten action plans or "solutions" to the motor problem.
post-KR delay-The interval of time between the presen-
tation of KR and the next movement. reflex-reversal phenomenon-The phenomenon by which a
given stimulus can produce two different reflexive responses
precision of KR-The level of accuracy with which KR depending on the function of the limb in a movement.
describes the movement outcome produced.
regression line-The line of best fit in plots of two vari-
prediction-The process in which the score on a criterion ables, whose slope and intercept are usually determined
variable is estimated from one or more predictor variables by regression analysis.
based on the association between them.
relative force-An invariant feature of the motor program
predictor variable-The variable(s) from which a crite- that defines the relationships among the forces produced
rion variable is predicted. in the various actions in a movement.
premotor reaction time-The interval from the stimulus relative frequency of KR-The percentage of trials for
presentation to the initial change in EMG. which KR is provided; the absolute frequency divided by
preparation-Reorganization of attention and infor- the number of trials.
mation processing so that a signal can be received and relative phase-A measure of temporal coordination that
responded to quickly. expresses the position of one limb within its cycle relative
preprogramming-The process of preparing a motor to the other limb within its cycle.
program for initiation. relative retention-Measures of retention in which the
preselection effect-In short-term motor memory work, performance on the retention test is evaluated in relation
the phenomenon whereby the memory for subject- to the level of performance reached in original learning.
 Glossary 499

reliability-The aspect of measurement related to the which processing activities are arranged sequentially in
repeatability of a score. time.
remoteness effect-In intertrial correlation matrices, the serial tasks-Movements combining a series of discrete
tendency for trials that are progressively more separated elements, with the order of elements being important.
in the practice sequence to correlate systematically lower set-Anonmemory pattern of adjustments that supports
with each other. performance.
response-chaining hypothesis-A movement control set hypothesis-A hypothesis holding that warm-up
theory which holds that each element in a sequence is trig- decrement is caused by loss of set.
gered by movement feedback from a previous element.
short-term memory-A memory store with a capacity
response-programming stage-A stage of information of about seven elements, capable of holding moderately
processing in which the previously chosen response is abstract information for up to 30 s; analogous to con-
transformed into overt muscular action. sciousness; a "work space" for processing.
response-selection stage-A stage of information pro- short-term sensory store-A functionally limitless
cessing in which the response associated with the pre- memory store for holding literal information for only
sented stimulus is selected. about 1 s.
response time-The interval from the presentation of a signal detection theory (SDT)-A method of analyzing
stimulus to the completion of a movement; the sum of two-choice decisions, resulting in two types of correct
reaction time and movement time. decision (hits and correct rejections) and two types of
retention interval-The interval between the end of errors (misses and false alarms).
original learning and the retention test. similarity-A construct in most theories of transfer,
retention test-A performance test administered after a indicating the extent to which certain aspects of two tasks
retention interval for the purpose of assessing learning. are the same.
retroactive interference-In interference theory, a source Simon effect-A type of stimulus-response compatibility
of forgetting caused by practice imposed between the effect in which irrelevant directional or location informa-
original learning and the retention test for a to-be- tion interferes with the action.
remembered task.
simple reaction time-Reaction time from a task in
root mean square error (RMSE)-The square root of the which a single action is produced in response to a single
average squared deviations of a set of values from a target presented stimulus.
value; typically used as a measure of tracking proficiency. simplification hypothesis-The idea that the factor struc-
savings score-A statistic used in transfer experiments, ture of a skill becomes progressively simpler with practice.
representing the "savings" in practice time on one task
simulator-A training device in which certain features of
resulting from experience on some other task.
a task are duplicated, allowing for practice that resembles
scattergram-A graph on which subjects' scores on two the transfer task.
tests are jointly represented as data points.
single-channel hypothesis-A theory of attention sug-
schema-The basis for schema theory; a rule, concept, or gesting that the system can process only a single stimulus
relationship formed on the basis of experience. leading to a response at any given time.
secondary-task method-A collection of experimental skills-Movements that are dependent on practice and
methods whereby learning on a main task can be esti- experience for their execution, as opposed to being geneti-
mated by the use of simultaneous secondary measures cally defined.
of performance.
slope (b)-One of the constants of a linear equation; the
selective attention-A mechanism for directing attention inclination of the line.
or capacity to a given stimulus input.
spacing effect-In memory experiments, the finding
self-organization-From dynamic-pattern theory; a that repetitions of the criterion task that are increasingly
view that describes motor control as emerging from the separated in time are remembered more effectively.
interaction of the components of the movement system.
spatial anticipation-The anticipation of which of several
self-regulation-Technique used in motor learning possible stimuli will occur; also called event anticipation.
studies in which the learners determine how to schedule
spatial-temporal goal-A subgoal for the performer
practice or feedback.
in which a pattern of limb movement defined in terms
sensitivity-That aspect of measurement dealing with the of both space and time is selected; the major product of
likelihood of detecting changes in a dependent measure running a motor program.
in relation to varying experimental conditions.
specificity of individual differences-Henry's theory of
sequencing-An invariant feature of motor programs in the structure of motor abilities, according to which motor
which the order of elements is fixed. tasks are thought to be composed of many independent
serial processing-A style of information processing in abilities.
500 Glossary

specificity of learning-The concept that the similar- tau (T)-Time-to-contact information based on the rate
ity of the environmental conditions in practice to those of expansion of the approaching object on the retina of
in transfer, has a strong positive influence on transfer the eye.
performance. taxonomy-A system of classification.
speed-accuracy trade-off-The general principle describ- temporal anticipation-The anticipation of when a given
ing a person's tendency to decrease the accuracy of a stimulus will arrive or when a movement is to be made.
movement when its speed is increased.
temporal variability-The inconsistency of some event
startle reaction-A rapid (<100 ms) reaction to an unex- with respect to time.
pected, often very loud, stimulus; used to study the
involuntary release of motor programs. terminal feedback-Feedback given after the move-
ment's completion.
state anxiety-A temporary state of worry or concern
about a particular situation or activity. time to contact (T)-Information about the time remain-
ing until a moving object arrives at the eye.
static contraction-Contraction in which the muscle is
not changing length as it is producing force; sometimes total variability (E)-The standard deviation of a set of
called isometric contraction. scores about a target value; a measure of overall accuracy.
stiffness-A characteristic of muscles and springs defined trace-decay theory-A theory holding that forgetting
as the change in tension divided by the change in length; is caused by the spontaneous "decay" or weakening of
the slope of the length-tension relationship. memory over time.
stimulus-identification stage-A stage of information trait anxiety-A general tendency to be anxious or
processing in which the stimulus is identified and features stressed that is characteristic of a particular individual.
or patterns are abstracted; often divided into separate transcranial magnetic stimulation (TMS)-A technique
encoding and identification stages. used to excite or inhibit processing in a specific area of
stimulus onset asynchrony (SOA)-The interval of the brain.
time between the onsets of two stimuli, as in the double- transfer-appropriate processing-The concept that prac-
stimulation paradigm. tice is most effective when arranged so that the processing
stimulus-response compatibility-The degree to which capability learned is appropriate for that in some goal
the stimuli and associated responses in a set are "natu- criterion task or conditions.
rally" related to each other. transfer design-An experimental design for measuring
stimulus-response viewpoint-A tradition in psychol- learning effects, in which all treatment groups are trans-
ogy and motor behavior stressing the responses produced ferred to a common level of the independent variable.
as a function of stimuli presented, without regard to the trials-delay technique-A procedure in which the presen-
intervening mental events or processes. tation of KR for a movement is delayed; during this inter-
storage problem-A problem with early notions of motor val the learner practices one or more other movements.
programming in which the number of necessary programs triggered reaction-A coordinated response to an envi-
was so large that their storage in the central nervous ronmental stimulus whose latency is shorter than RT yet
system seemed impossible. longer than that of the long-loop reflex.
stress-A motivational state that tends to direct the indi- unit of action-A "piece" of behavior that can be utilized
vidual away from some particular situation. repeatedly in various actions, producing essentially the
Stroop effect-A delay in responding due to competition same movements (but scaled to the environment) each
from automatically processed information. time.
structural interference-Interference among tasks caused validity-That aspect of measurement related to the
by the simultaneous use of the same receptors, effectors, extent to which a test measures what the experimenter
or processing systems. wanted it to measure.
subjective reinforcement-A construct in Adams' variability in practice-A prediction of schema theory;
closed-loop theory; the subject's self-generated error transfer is predicted to be facilitated when goals are
signal, based on comparing feedback against a reference systematically varied from trial to trial during practice.
of correctness. variable error (VE)-The standard deviation of a set of
summary KR-Augmented information about each scores about the subject's own average score; a measure
of a set of performance trials presented after the set is of movement consistency.
completed. variable error in timing (VE 1)-The within-subject stan-
synchronization-A task in which the subject's goal is dard deviation of the duration of some process or event.
to make periodic finger taps that coincide with those of ventral visual stream-Processing involving the infero-
a pacing signal. temporal cortex; responsible for providing cognitive
task analysis-A process of determining the underlying information about objects in the environment; sometimes
abilities and structure of a task or occupation. called focal vision.
 Glossary 501

verbal pretraining-The presentation of stimulus or dis- warm-up decrement-The decrement in performance

play elements of the task in isolation so that they can be occurring when practice is resumed after a brief rest
more easily responded to later in whole-task performance. period.
vestibular apparatus-The receptors in the inner ear that whole practice-The learning technique in which the
are sensitive to the orientation of the head with respect to task is practiced in its entirety (i.e., not broken into parts).
gravity, to rotation of the head, and to balance. wineglass effect-A slip of a held object through the
visual illusion-A perceptual effect that causes an inac- fingertips that triggers an increased grip force within
curate visual representation of a display. approximately 30 ms.
visual proprioception-Gibson's concept that vision can Wing-Kristofferson model-An open-loop, hierarchical
serve as a strong basis for perception of movements and model of timing performance involving a central time-
positions of the body in space. keeper level and a motor implementation level.
 REFERENCES

Abbs, J.H., & Gracco, V.L. (1983). Sensorimotor actions Abrams, R.A., Meyer, D.E., & Kornblum, S. (1989).
in the control of multi-movement speech gestures. Speed and accuracy of saccadic eye movements:
Trends in Neuroscience, 6, 391-395. Characteristics of impulse variability in the
Abbs, J.H., Gracco, V.L., & Cole, K.J. (1984). Control oculomotor system. Journal of Experimental
of multi-movement coordination: Sensorimotor Psychology: Human Perception and Performance,
mechanisms in speech motor programming. Journal 15, 529-543.
of Motor Behavior, 16, 195-232. Abrams, R.A., Meyer, D.E., & Kornblum, S. (1990). Eye-
Abbs, J.H., & Winstein, C.J. (1990). Functional hand coordination: Oculomotor control in rapid
contributions of rapid and automatic sensory-based aimed limb movements. Journal of Experimental
adjustments to motor output. In M. Jeannerod Psychology: Human Perception and Performance,
(Ed.), Attention and performance XIII (pp. 627-652). 16, 248-267.
Hillsdale, NJ: Erlbaum. Ackerman, P.L. (1988). Determinants of individual
Abernethy, B. (1988). Dual-task methodology and motor differences during skill acquisition: Cognitive
skills research: Some applications and methodological abilities and information processing. Journal of
constraints. Journal of Human Movement Studies, Experimental Psychology: General, 117, 288-318.
14, 101-132. Ackerman, P.L. (1989). Individual differences and skill
Abernethy, B. (2001). Attention. In R.N. Singer, H.A. acquisition. In P.L. Ackerman, R.J. Sternberg, & R.
Hausenblas, & C.M. Janelle (Eds.), Handbook of sport Glaser (Eds.), Learning and individual differences:
psychology (2nd ed.) (pp. 53-85). New York: Wiley. Advances in theory and research (pp. 165-217). New
York: Freeman.
Abernethy, B., & Burgess-Limerick, R. (1992). Visual
information for the timing of skilled movements: Ackerman, P.L. (1990). A correlational analysis of
A review. In J.J. Summers (Ed.), Approaches to the skill specificity: Learning, abilities, and individual
study of motor control and learning (pp. 343-384). differences. Journal of Experimental Psychology:
Amsterdam: Elsevier. Learning, Memory, and Cognition, 16, 883-901.
Abernethy, B., Burgess-Limerick, R., & Parks, S. (1994). Ackerman, P.L. (1992). Predicting individual differences
Contrasting approaches to the study of motor in complex skill acquisition: Dynamics of ability
expertise. Quest, 46, 186-198. determinants. Journal of Applied Psychology, 77,
598-614.
Abernethy, B., Hanna, A., & Plooy, A. (2002). The
attentional demands of preferred and non-preferred Ackerman, P.L., & Cianciolo, A.T. (2000). Cognitive,
gait patterns. Gait & Posture, 15, 256-265. perceptual-speed, and psychomotor determinants of
Abernethy, B., Neal, R.J., & Koning, P. (1994). Visual- individual differences during skill acquisition. Journal
perceptual and cognitive differences between expert, of Experimental Psychology: Applied, 6, 259-290.
intermediate, and novice snooker players. Applied Adam, J.J., Mol, R., Pratt, J., & Fischer, M.H. (2006).
Cognitive Psychology, 8, 185-211. Moving farther but faster: An exception to Fitts's law.
Abernethy, B., & Russell, D.G. (1987). Expert-novice Psychological Science, 17, 794-798.
differences in an applied selective attention task. Adamovich, S.V., & Feldman, A.G. (1984). Model of
Journal of Sport Psychology, 9, 326-345. the central regulation of the parameters of motor
Abernethy, B., & Wood, J.M. (2001). Do generalized trajectories. Biophysics, 29, 338-342.
visual training programmes for sport really work? Adams, J.A. (1952). Warm-up decrement in performance
An experimental investigation. Journal of Sports on the pursuit-rotor. American Journal of Psychology,
Sciences, 19, 203-222. 65, 404-414.
Abernethy, B., Wood, J.M., & Parks, S. (1999). Can the Adams, J.A. (1953). The prediction of performance at
anticipatory skills of experts be learned by novices? advanced stages of training on a complex psychomotor
Research Quarterly for Exercise and Sport, 70, 313- task (Research Bulletin 53-49). Lackland Air Force
318. Base, TX: Human Resources Research Center.

503
504 ll1 References

Adams, J.A. (1955). A source of decrement in Adams, J.A., & Reynolds, B. (1954). Effect of shift
psychomotor performance. Journal of Experimental in distribution of practice conditions following
Psychology, 49, 390-394. interpolated rest.Journal of Experimental Psychology,
Adams,J.A. (1956).Anevaluationoftestitemsmeasuring 47, 32-36.
motor abilities (Research Rep. AFPTRC-TN-56-55). Adrian, E.D., & Buytendijk, F.J.J. (1931). Potential
Lackland Air Force Base, TX: Human Resources changes in the isolated brain stem of the goldfish.
Research Center. Journal of Physiology, 71, 121-135.
Adams, J.A. (1957). The relationship between certain Aglioti, S., DeSouza, J.F.X., & Goodale, M.A. (1995). Size
measures of ability and the acquisition of a contrast illusions deceive the eye but not the hand.
psychomotor criterion response. Journal of General Current Biology, 5, 679-685.
Psychology, 56, 121-134. Aiken, L.R. Jr. (1964). Reaction time and the expectancy
Adams, J.A. (1961). The second facet of forgetting: hypothesis. Perceptual and Motor Skills, 19, 655-661.
A review of warm-up decrement. Psychological Ainscoe, M., & Hardy, L. (1987). Cognitive warm-up in
Bulletin, 58, 257-273. a cyclical gymnastics skill. International Journal of
Adams, J.A. (1964). Motor skills. Annual Review of Sport Psychology, 18, 269-275.
Psychology, 15, 181-202.
Al-Ameer,H., & Toole, T. (1993). Combinations of blocked
Adams, J.A. (1967). Human memory. New York: and random practice orders: Benefits to acquisition
McGraw-Hill. and retention. Journal of Human Movement Studies,
Adams, J.A. (1968). Response feedback and learning. 25, 177-191.
Psychological Bulletin, 70, 486-504. Albaret, J.-M., & Thon, B. (1998). Differential effects
Adams, J.A. (1971). A closed-loop theory of motor of task complexity on contextual interference in a
learning. Journal of Motor Behavior, 3, 111-150. drawing task. Acta Psychologica, 100, 9-24.
Adams, J.A. (1976a). Issues for a closed-loop theory of Alessi, S.M. (1988). Fidelity in the design of instructional
motor learning. In G.E. Stelmach (Ed.), Motor control: simulators. Journal of Computer-Based Instruction,
Issues and trends (pp. 87-107). New York: Academic 15, 40-47.
Press. Alexander, R.M. (2003). Principles of animal locomotion.
Adams, J.A. (1976b ). Learning and memory: An Princeton, NJ: Princeton University Press.
introduction. Homewood, IL: Dorsey. Allami, N., Paulignan, Y., Brovelli, A., & Boussaoud,
Adams, J.A. (1977). Feedback theory of how joint D. (2008). Visuo-motor learning with combination
receptors regulate the timing and positioning of a of divergent rates of motor imagery and physical
limb. Psychological Review, 84, 504-523. practice. Experimental Brain Research, 184, 105-113.
Adams, J.A. (1978). Theoretical issues for knowledge Allport, A. (1987). Selection for action: Some behavioral
of results. In G.E. Stelmach (Ed.), Information and neurophysiological considerations of attention
processing in motor control and learning (pp. 229- and action. In H. Heuer & A.F. Sanders (Eds.),
240). New York: Academic Press. Perspectives on perception and action (pp. 395-419).
Adams, J.A. (1986). Use of the model's knowledge Hillsdale, NJ: Erlbaum.
of results to increase the observer's performance. Allport, A. (1993). Attention and control: Have we been
Journal of Human Movement Studies, 12, 89-98. asking the wrong questions? In D.E. Meyer & S.
Adams, J.A. (1987). Historical review and appraisal of Kornblum (Eds.), Attention and performance XIV.
research on the learning, retention, and transfer of Cambridge, MA: MIT Press.
human motor skills. Psychological Bulletin, 101, 41- Allport, D.A., Antonis, B., & Reynolds, P. (1972). On the
74. division of attention: A disproof of the single channel
Adams, J.A., & Bray, N.W. (1970). A closed-loop theory hypothesis. Quarterly Journal of Experimental
of paired-associate verbal learning. Psychological Psychology, 24, 225-235.
Review, 77, 385-405. Alm, H., & Nilsson, L. (1995). The effects of a
Adams, J.A., & Creamer, L.R. (1962). Anticipatory mobile telephoning task on driver behaviour in
timing of continuous and discrete responses. Journal a car following situation. Accident Analysis and
of Experimental Psychology, 63, 84-90. Prevention, 27, 707-715.
Adams, J:A., & Dijkstra, S. (1966). Short-term memory Almeida, Q.J., Wishart, L.R., & Lee, T.D. (2002). Bimanual
for motor responses. Journal of Experimental coordination deficits with Parkinson's disease: The
Psychology, 71, 314-318. influence of movement speed and external cueing.
Adams, J.A., & Hufford, L.E. (1962). Contributions of a Movement Disorders, 17, 30-37.
part-task trainer to the learning and relearning of a Amazeen, P.G. (2002). Is dynamics the content of a
time-shared flight maneuver. Human Factors, 4, 159- generalized motor program for rhythmic interlimb
170. coordination? Journal of Motor Behavior, 34, 233-251.
 References 111 505

Amazeen, P.G., Amazeen, E.L., & Turvey, M.T. (1998). Anshel, M.H., & Wrisberg, C.A. (1988). The effect
Breaking the reflectional symmetry of interlimb of arousal and focused attention on warm-up
coordination dynamics. Journal of Motor Behavior, decrement. Journal of Sport Behavior, 11, 18-31.
30, 199-216. Anshel, M.H., & Wrisberg, C.A. (1993). Reducing warm-
Amazeen, P.G., Schmidt, R.C., & Turvey, M.T. (1995). up decrement in the performance of the tennis serve.
Frequency detuning of the phase entrainment Journal of Sport and Exercise Psychology, 15, 290-303.
dynamics of visually coupled rhythmic movements. Anson, G., Elliott, D., & Davids, K. (2005). Information
Biological Cybernetics, 72, 511-518. processing and constraints-based views of skill
Ambrose, A. (1997). Cellular telephones and traffic acquisition: Divergent or complementary? Motor
accidents (correspondence). New England Journal of Control, 9, 217-241.
Medicine, 337, 127-129. Anzola, G.P., Bertolini, G., Buchtel, H.A., & Rizzolatti, G.
Ammons, R.B. (1950). Acquisition of motor skill: III. (1977). Spatial compatibility and anatomical factors in
Effects of initially distributed practice on rotary simple and choice reaction times. Neuropsychologica,
pursuit performance. Journal of Experimental 15, 295-302.
Psychology, 40, 777-787. Arbib, M.A. (1981). Perceptual structures and distributed
Ammons, R.B. (1951). Effects of pre-practice activities on motor control. In V. Brooks (Ed.), Handbook of
rotary pursuit performance. Journal of Experimental physiology: Section 1: The nervous system. Vol. 2.
Psychology, 41, 187-191. Motor control, part 2 (pp. 1449-1480). Baltimore:
Ammons, R.B. (1988). Distribution of practice in motor American Physiological Society.
skill acquisition: A few questions and comments. Archer, E.J., Kent, G.W., & Mote, F.A. (1956). Effect of
Research Quarterly for Exercise and Sport, 59, 288-290. long-term practice and time-on-target information
Anderson, D.I., Magill, R.A., & Sekiya, H. (1994). A feedback on a complex tracking task. Journal of
reconsideration of the trials-delay of knowledge of Experimental Psychology, 51, 103-112.
results paradigm in motor skill learning. Research Armstrong, T.R. (1970a). Feedback and perceptual-motor
Quarterly for Exercise and Sport, 65, 286-290. skill learning: A review of information feedback and
Anderson, D.I., Magill, R.A., & Sekiya, H. (2001). manual guidance training techniques (Tech. Rep.
Motor learning as a function of KR schedule and No. 25). Ann Arbor, MI: University of Michigan,
characteristics of task intrinsic feedback. Journal of Department of Psychology.
Motor Behavior, 33, 59-66. Armstrong, T.R. (1970b). Training for the production of
Anderson, D.I., Magill, R.A., Sekiya, H., & Ryan, G. memorized movement patterns (Tech. Rep. No. 26).
(2005). Support for an explanation of the guidance Ann Arbor, MI: University of Michigan, Department
effect in motor skill learning. Journal of Motor of Psychology.
Behavior, 37, 231-238. Arnold, P., & Farrell, M.J. (2002). Can virtual reality be
Anderson, J.R. (1982). Acquisition of cognitive skill. used to measure and train surgical skills? Ergonomics,
Psychological Review, 89, 369-406. 45, 362-379.
Anderson, J.R. (1990). Cognitive psychology and its Arps, G.F. (1920). Work with knowledge of results versus
implications (3rd ed.). New York: Freeman. work without knowledge of results. Psychological
Anderson, J.R. (1995). Learning and memory: An Monographs, 28, 1-41.
integrated approach. New York: Wiley. Arutyunyan, G.A., Gurfinkel, V:S., & Mirskii, M.L.
Andres, R.0., & Hartung, K.J. (1989). Prediction of head (1968). Investigation of aiming at a target. Biophysics,
movement time using Fitts' Law. Human Factors, 31, 13, 536-538.
703-713. Arutyunyan, G.A., Gurfinkel, V.S., & Mirskii, M.L. (1969).
Angel, R.W. (1977). Antagonist muscle activity during Organization of movements on execution by man of
rapid arm movements: Central versus proprioceptive an exact postural task. Biophysics, 14, 1162-1167.
influences. Journal of Neurology, Neurosurgery, and Asatryan, D.G., & Feldman, A.G. (1965). Biophysics
Psychiatry, 40, 683-686. of complex systems and mathematical models.
Angel, R.W., & Higgins, J.R. (1969). Correction of false Functional tuning of nervous system with control of
moves in pursuit tracking. Journal of Experimental movement or maintenance of a steady posture-I.
Psychology, 82, 185-187. Mechanographic analysis of the work of the joint on
Annett, J. (1959). Leaming a pressure under conditions execution of a postural task. Biophysics, 10, 925-935.
of immediate and delayed knowledge of results. Aschersleben, G. (2002). Temporal control of movements
Quarterly Journal of Experimental Psychology, 11, in sensorimotor synchronization. Brain and
3-15. Cognition, 48, 66-79.
Annett, J. (1969). Feedback and human behavior. Aschersleben, G., & Miisseler, J. (1999). Dissociations in
Middlesex, England: Penguin. the timing of stationary and moving stimuli. Journal
506 References

of Experimental Psychology: Human Perception and Badets, A., & Blandin, Y. (2005). Observational learning:
Performance, 25, 1709-1720. Effects of bandwidth knowledge of results. Journal of
Ascoli, KM., & Schmidt, R.A. (1969). Proactive Motor Behavior, 37, 211-216.
interference in short-term motor retention. Journal of Badets, A., Blandin, Y., Wright, D.L., & Shea, C.H. (2006).
Motor Behavior, 1, 29-35. Error detection processes during observational
Ashmead, D.H., Davis, D.L., & Northington, A. (1995). learning. Research Quarterly for Exercise and Sport,
Contribution of listeners' approaching motion to 77, 177-184.
auditory distance perception. Journal of Experimental Bahrick, H.P., Fitts, P.M., & Briggs, G.E. (1957). Learning
Psychology: Human Perception and Performance, 21, curves-facts or artifacts? Psychological Bulletin, 54,
239-256. 256-268.
Assaiante, C., Marchand, A.R., & Amblard, B. (1989). Baird, KM., Hoffmann, E.R., & Drury, C.G. (2002).
Discrete visual samples may control locomotor The effects of probe length on Fitts' law. Applied
equilibrium and foot positioning in man. Journal of Ergonomics, 33, 9-14.
Motor Behavior, 21, 72-91. Balakrishnan, R. (2004). "Beating" Fitts' law: Virtual
Atkeson, C.G., & Hollerbach, J.M. (1985). Kinematic enhancements for pointing facilitation. International
features of unrestrained vertical arm movements. Journal of Human-Computer Studies, 61, 857-874.
Journal of Neuroscience, 5, 2318-2330. Balasubramaniam, R. (2006). Trajectory formation in
Atkinson, R.C., & Shiffrin, R.M. (1971). The control of timed repetitive movements. In M.L. Latash & F.
short-term memory. Scientific American, 225, 82-90. Lestienne (Eds.), Progress in motor control IV (pp.
47-54). New York: Springer.
Attneave, F. (1959). Applications of information theory to
psychology: A summary of basic concepts, methods, Balasubramaniam, R., & Turvey, M.T. (2005). The
and results. New York: Holt, Rinehart & Winston. lambda-based equilibrium point (EP) hypothesis
1990-1999. In M.F. Levin (Ed.), Forty years of the
Augustyn, J.S., & Rosenbaum, D.A. (2005). Metacognitive
equilibrium point hypothesis (pp. 375-378). Quebec:
control of action: Preparation for aiming reflects
Tristar.
knowledge of Fitts's law. Psychonomic Bulletin &
Review, 12, 911-916. Balasubramaniam, R., Wing, A.M., & Daffertshofer, A.
(2004). Keeping with the beat: Movement trajectories
Baars, B.J. (1986). The cognitive revolution in psychology.
contribute to movement timing. Experimental Brain
New York: Guilford.
Research, 159, 129-134.
Baars, B.J. (1997). In the theatre of consciousness: The
Baldissera, F., Cavallari, P., & Civaschi, P. (1982).
workspace of the mind. New York: Oxford University
Preferential coupling between voluntary movements
Press.
of ipsilateral limbs. Neuroscience Letters, 34, 95-100.
Bachman, J.C. (1961). Specificity vs. generality in
Baldissera, F., Cavallari, P., Marini, G., & Tassone, G.
learning and performing two large muscle motor
(1991). Differential control of in-phase and anti-phase
tasks. Research Quarterly, 32, 3-11.
coupling of rhythmic movements of ipsilateral hand
Bachman, J.C. (1966). Influence of age and sex on and foot. Experimental Brain Research, 83, 375-380.
the amount and rate of learning two motor tasks.
Baldissera, F., Cavallari, P., & Tesio, L. (1994).
Research Quarterly, 37, 176-186.
Coordination of cyclic coupled movements of hand
Bachrach, A.J. (1970). Diving behavior. In Scripps and foot in normal subjects and on the healthy side
Institute of Oceanography, Human performance of hemiplegic patients. In S.P. Swinnen, H. Heuer, J.
and scuba diving. Proceedings of the Symposium Massion, & P. Casaer (Eds.), Interlimb coordination:
on Underwater Physiology, La Jolla, CA. Chicago: Neural, dynamical, and cognitive constraints (pp.
Athletic Institute. 229-242). San Diego: Academic Press.
Baddeley, A. (2003). Working memory: Looking back Baldo, M.V.C., Ranvaud, R.D., & Morya, E. (2002). Flag
and looking forward. Nature Reviews: Neuroscience, errors in soccer games: The flash-lag effect brought to
4, 829-839. real life. Perception, 31, 1205-1210.
Baddeley, A.D., & Longman, D,J.A. (1978). The influence Band, G.P.H., & van Boxtel, G.J.M. (1999). Inhibitory
of length and frequency of training session on the motor control in stop paradigms: Review and
rate of learning to type. Ergonomics, 21, 627-635. reinterpretation of neural mechanisms. Acta
Baddeley, AD., & Weiskrantz, L. (Eds.). (1993).Attention: Psychologica, 101, 179-211.
Selection, awareness, and control: A tribute to Donald Band, G.P.H., van der Molen, M.W., & Logan, G.D.
Broadbent. Oxford: Clarendon Press. (2003). Horse-race model simulations of the stop-
Badets, A., & Blandin, Y. (2004). The role of knowledge signal procedure. Acta Psychologica, 112, 105-142.
of results frequency in learning through observation. Bandura, A. (1969). Principles of behavior modification.
Journal of Motor Behavior, 36, 62-70. New York: Holt, Rinehart & Winston.
 References 507

Bandura, A., Blanchard, E.B., & Ritter, B. (1969). Battig, W.F., & Shea, J.B. (1980). Levels of processing
Relative efficacy of desensitization and modeling of verbal materials: An overview. In P. Klavora & J.
approaches for inducing behavioral, affective, and Flowers (Ed.), Motor learning and biomechanical
attitudinal changes. Journal of Personality and Social factors in sport (pp. 24-33). Toronto: University of
Psychology, 13, 173-199. Toronto.
Bard, C., Turrell, Y., Fleury, M., Teasdale, N., Lamarre, Baumeister, R.F. (1984). Choking under pressure: Self-
Y., & Martin, 0. (1999). Deafferentation and pointing consciousness and paradoxical effects of incentives
with visual double-step perturbations. Experimental on skillful performance. Journal of Personality and
Brain Research, 125, 410-416. Social Psychology, 46, 610-620.
Barnett, M.L., Ross, D., Schmidt, R.A., & Todd, B. (1973). Bayley, N. (1935). The development of motor abilities
Motor skills learning and the specificity of training during the first three years. Monographs of the
principle. Research Quarterly, 44, 440-447. Society for Research in Child Development, 1, 1-26.
Barnsley, R.H., Thompson, A.H., & Legault, P. (1992). Bean, C.H. (1912). The curve of forgetting. Archives of
Family planning: Football style. The relative age effect Psychology, 3, 1-45.
in football. International Review for the Sociology of Bear, M.F., Connors, B.W., & Paradiso, M.A. (2001).
Sport, 27, 77-86. Neuroscience: Exploring the brain (2nd ed.).
Barrett, N.C., & Glencross, D.J. (1989). Response Baltimore: Lippincott Williams & Wilkins.
amendments during manual aiming movements to Beatty, J., & Wagoner, B.L. (1978). Pupillometric signs
double-step targets. Acta Psychologica, 70, 205-217. of brain activation vary with level of cognitive
Bartlett, F.C. (1932). Remembering: A study in processing. Science, 199, 1216-1218.
experimental and social psychology. Cambridge: Beauchet, 0., Annweiler, C., Dubost, V., Allali, G.,
Cambridge University Press. Kressig, R.W., Bridenbaugh, S., Berrut, G., Assal, F., &
Bartlett, F.C. (1958). Thinking: An experimental and Herrmann, F.R. (2009). Stops walking when talking:
social study. New York: Basic Books. A predictor of falls in older adults? European Journal
of Neurology, 16, 786-795.
Bartlett, N., & Bartlett, S. (1959). Synchronization of a
motor response with an anticipated sensory event. Beek, P.J., & Lewbel, A. (1995, November). The science of
Psychological Review, 66, 203-218. juggling. Scientific American, 273, 92-97.

Bartlett, R., & Darling, W.G. (2002). Opposite effects on Beer, R.S. (2009). Beyond control: The dynamics of brain-
perception and action induced by the Ponzo illusion. body-environment interaction in motor systems. In
Experimental Brain Research, 146, 433-440. D. Sternad (Ed.), Progress in motor control (pp. 7-24).
Berlin: Springer.
Bartram, D., Banerji, N., Rothwell, D., & Smith, P. (1985).
Beevor, C.E., & Horsely, V. (1887). A minute analysis
Task parameters affecting individual differences in
(experimental) of the various movements produced
pursuit and compensatory tracking performance.
by stimulating in the monkey different regions of
Ergonomics, 28, 1633-1652.
the cortical centre for the upper limb as defined by
Basmajian, J.V. (1963). Control and training of individual Professor Ferrier. Philosophical Transactions, 178, 153.
motor units. Science, 141, 440-441.
Beevor, C.E., & Horsely, V. (1890). A record of the results
Basmajian, J.V. (1989). Biofeedback: Principles and obtained by electrical excitation of the so-called
practice for clinicians (3rd ed.). Baltimore: Williams motor cortex and internal capsule in the orangutang.
& Wilkins. Philosophical Transactions, 181, 129.
Battig, W.F. (1956). Transfer from verbal pretraining Beilock, S.L. (2008). Math performance in stressful
to motor performance as a function of motor task situations. Current Directions in Psychological
complexity. Journal of Experimental Psychology, 51, Science, 17, 339-343.
371-378.
Beilock, S.L. (2010). Choke: What the secrets of the brain
Battig, W.F. (1966). Facilitation and interference. In reveal about getting it right when you have to. New
E.A. Bilodeau (Ed.), Acquisition of skill. New York: York: Free Press.
Academic Press. Beilock, S.L., Bertenthal, B.I., McCoy, A.M., & Carr, T.H.
Battig, W.F. (1972). Intratask interference as a source of (2004). Haste does not always make waste: Expertise,
facilitation in transfer and retention. In R.F. Thompson direction of attention, and speed versus accuracy
& J.F. Voss (Eds.), Topics in learning and performance in performing sensorimotor skills. Psychonomic
(pp. 131-159). New York: Academic Press. Bulletin & Review, 11, 373-379.
Battig, W.F. (1979). The flexibility of human memory. Beilock, S.L., & Carr, T.H. (2001). On the fragility of
In L.S. Cermak & F.I.M. Craik (Eds.), Levels of skilled performance: What governs choking under
processing in human memory (pp. 23-44). Hillsdale, pressure? Journal of Experimental Psychology:
NJ: Erlbaum. General, 130, 701-725.
508 Iii! References

Beilock, S.L., Carr, T.H., MacMahon, C., & Starkes, J.L. Bilodeau, E.A., & Bilodeau, I.M. (1961). Motor-skills
(2002). When paying attention becomes counter- learning. Annual Review of Psychology, 12, 243-280.
productive: Impact of divided versus skill-focused Bilodeau, E.A., Bilodeau, I.M., & Schumsky, D.A.
attention on novice and experienced performance (1959). Some effects of introducing and withdrawing
of sensorimotor skills. Journal of Experimental knowledge of results early and late in practice.
Psychology: Applied, 8, 6-16. Journal of Experimental Psychology, 58, 142-144.
Beilock, S.L., Kulp, C.A., Holt, L.E., & Carr, T.H. (2004). Bilodeau, I.M. (1956). Accuracy of a simple positioning
More on the fragility of performance: Choking under response with variation in the number of trials by
pressure in mathematical problem solving.Journal of which knowledge of results is delayed. American
Experimental Psychology: General, 133, 584-600. Journal of Psychology, 69, 434-437.
Belen'kii, V.Y., Gurfinkel, V.S., & Pal'tsev, Y.I. (1967). Bilodeau, I.M. (1966). Information feedback. In E.A.
Elements of control of voluntary movements. Bilodeau (Ed.), Acquisition of skill (pp. 255-296).
Biofizika, 12, 135-141. New York: Academic Press.
Benedetti, C., & McCullagh, P. (1987). Post-knowledge Bilodeau, I.M. (1969). Information feedback. In E.A.
of results delay: Effects of interpolated activity on Bilodeau (Ed.), Principles of skill acquisition (pp. 255-
learning and performance. Research Quarterly for 285). New York: Academic Press.
Exercise and Sport, 58, 375-381.
Binsted, G., Rolheiser, T.M., & Chua, R. (2006). Decay in
Bennett, D.M., & Simmons, R.W. (1984). Effects of
visuomotor representations during manual aiming.
precision of knowledge of results on acquisition and
Journal of Motor Behavior, 38, 82-87.
retention of a simple motor skill. Perceptual and
Motor Skills, 58, 785-786. Bizzi, E. (1974). The coordination of eye-head
movements. Scientific American, 231(4), 100-106.
Berkinblit, M.B., & Feldman, A.G. (1988). Some problems
of motor control. Journal of Motor Behavior, 20, 369- Bizzi, E.,Accornero, N., Chapple, W., & Hogan, N. (1982).
373. Arm trajectory formation in monkeys. Experimental
Brain Research, 46, 139-143.
Berkinblit, M.B., Feldman, A.G., & Fukson, 0.I. (1986).
Adaptability of innate motor patterns and motor Bizzi, E., Hogan, N., Mussa-Ivaldi, F.A., & Giszter, S.
control mechanisms. Behavioral and Brain Sciences, (1992). Does the nervous system use equilibrium-
9, 585-638. point control to guide single and multiple joint
movements? Behavioral and Brain Sciences, 15, 603-
Bernstein, N.A. (1947). On the structure of movements.
613.
Moscow: State Medical Publishing House.
Bizzi, E., Kalil, R.E., & Tagliasco, V. (1971). Eye-head
Bernstein, N.A. (1967). The co-ordination and regulation
coordination in monkeys: Evidence for centrally
of movements. Oxford: Pergamon Press.
patterned organization. Science, 173, 452-454.
Bernstein, N.A. (1996). On dexterity and its development.
Bizzi, E., Polit, A., & Morasso, P. (1976). Mechanisms
In M.L. Latash & M.T. Turvey (Eds.), Dexterity and its
development (pp. 3-244). Mahwah, NJ: Erlbaum. underlying achievement of final head position.
Journal of Neurophysiology, 39, 435-444.
Bernstein, N.A., & Popova, T.S. (1930). Studies on the
biodynamics of the piano strike: Paper 1. Studies of the Bjork, R.A. (1994). Institutional impediments to effective
rhythmic octave strike using the kymocyclographic training. In D. Druckman & R.A. Bjork (Eds.),
method. Proceedings of the piano-methodological Learning, remembering, believing (pp. 295-306).
section of the State Institute of Music Science (Vol. Washington, DC: National Academy Press.
1) (pp. 5-47). Moscow: Muzgiz. (In Russian; English Blakemore, S.-J., Frith, C.D., & Wolpert, D.M. (1999).
translation published 2003 in Motor Control, 7, 3-45) Spatio-temporal prediction modulates the perception
Biguer, B., Jeannerod, M., & Prablanc, C. (1982). The of self-produced stimuli. Journal of Cognitive
coordination of eye, head, and arm movements Neuroscience, 11, 551-559.
during reaching at a single visual target. Experimental Blakemore, S.-J., Wolpert, D.M., & Frith, C.D. (1998).
Brain Research, 46, 301-304. Central cancellation of self-produced tickle sensation.
Biguer, B., Prablanc, C., & Jeannerod, M. (1984). The Nature Neuroscience, 1, 635-640.
contribution of coordinated eye and head movements Blakemore, S.-J., Wolpert, D.M., & Frith, C.D. (2000).
in hand pointing accuracy. Experimental Brain Why can't you tickle yourself? Neuroreport, 11,
Research, 55, 462-469. Rll-Rl6.
Bilodeau, E.A. (Ed.). (1966). Acquisition of skill. New Blandin, Y., Lhuisset, L., & Proteau, L. (1999). Cognitive
York: Academic Press. processes underlying observational learning of motor
Bilodeau, E.A., & Bilodeau, I.M. (1958). Variable skills. Quarterly Journal of Experimental Psychology,
frequency of knowledge of results and the learning 52A, 957-979.
of a simple skill. Journal of Experimental Psychology, Blandin, Y., & Proteau, L. (2000). On the cognitive basis of
55, 379-383. observational learning: Development of mechanisms
 References 509

for the detection and correction of errors. Quarterly Bourne, L.E. Jr., & Bunderson, C.V. (1963). Effects
Journal of Experimental Psychology, 53A, 846-867. of delay of informative feedback and length of
Blandin, Y., Toussaint, L., & Shea, C.H. (2008). Specificity postfeedback interval on concept identification.
of practice: Interaction between concurrent sensory Journal of Experimental Psychology, 65, 1-5.
information and terminal feedback. Journal of Bourne, L.E. Jr., Guy, D.E., Dodd, D.H., & Justesen, D.R.
Experimental Psychology: Learning, Memory, and (1965). Concept identification: The effects of varying
Cognition, 34, 994-1000. length and informational components of the intertrial
Blankenship, W.C. (1952). Transfer effects in neuro- interval. Journal of Experimental Psychology, 69, 624-
muscular responses involving choice. Unpublished 629.
master's thesis, University of California, Berkeley. Boutcher, S.H., & Crews, D.J. (1987). The effect of a
preshot attentional routine on a well-learned skill.
Blix, M. (1892-1895). Die Lange und Spannung des
International Journal of Sport Psychology, 18, 30-39.
Muskels. Skandinavische Archiv Physiologie, 3, 295-
318; 4, 399-409; 5, 150-206. Bowditch, H.P., & Southard, W.F. (1882). A comparison
of sight and touch. Journal of Physiology, 3, 232-244.
Blouin, J., Gauthier, G.M., Vercher, J.L., & Cole, J. (1996).
The relative contribution of retinal and extraretinal Boyce, B.A. (1992). Effects of assigned versus participant-
signals in determining the accuracy of reaching set goals on skill acquisition and retention of a
movements in normal subjects and a deafferented selected shooting task.Journal of Teaching in Physical
patient. Experimental Brain Research, 109, 148-153. Education, 11, 220-234.
Bogacz, S. (2005). Understanding how speed affects Boyce, B.A., & Del Rey, P. (1990). Designing applied
performance of polyrhythms: Transferring control as research in a naturalistic setting using a contextual
speed increases. Journal of Motor Behavior, 37, 21-34. interference paradigm. Journal of Human Movement
Studies, 18, 189-200.
Bogaerts, H, & Swinnen, S.P. (2001). Spatial interactions
during bimanual coordination patterns: The effect of Boyd, I.A., & Roberts, T.D.M. (1953). Proprioceptive
directional compatibility. Motor Control, 5, 183-199. discharges from stretch-receptors in the knee-joint of
the cat. Journal of Physiology, 122, 38-59.
Bonnet, C., Carello, C., & Turvey, M.T. (2009). Diabetes
and postural stability: Review and hypotheses. Brady, F. (1998). A theoretical and empirical review of
Journal of Motor Behavior, 41, 172-190. the contextual interference effect and the learning of
motor skills. Quest, 50, 266-293.
Book, W.F. (1908). The psychology of skill. University of
Brady, F. (2004). Contextual interference: A meta-analytic
Montana Studies in Psychology (Vol. 1). (Reprinted,
study. Perceptual and Motor Skills, 99, 116-126.
New York: Gregg, 1925)
Bransford, J.D., Franks, J.J., Morris, C.D., & Stein, B.S.
Bootsma, R.J., Marteniuk, R.G., MacKenzie, C.L., &
(1979). Some general constraints on learning and
Zaal, F.T.J.M. (1994). The speed-accuracy trade-
memory research. In L.S. Cermak & F.I.M. Craik
off in manual prehension: Effects of movement
(Eds.), Levels of processing in human memory (pp.
amplitude, object size and object width on kinematic
331-354). Hillsdale, NJ: Erlbaum.
characteristics. Experimental Brain Research, 98, 535-
541. Brebner, J., Shepard, M., & Cairney, P. (1972).
Spatial relationships and S-R compatibility. Acta
Bootsma, R.J., & van Wieringen, P.C.W. (1990). Timing
Psychologica, 36, 1-15.
an attacking forehand drive in table tennis. Journal
of Experimental Psychology: Human Perception and Brenner, J. (1974). A general model of voluntary control
Performance, 16, 21-29. applied to the phenomenon of learned cardiovascular
change. In P.A. Obrist, A.H. Black, J. Brenner, & L.V.
Boring, E.G. (1950). A history of experimental
DiCara (Eds.), Cardiovascular psychophysiology:
psychology. New York: Appleton-Century-Crofts.
Current issues in response mechanisms, biofeedback,
Bartoli, L., Robazza, C., Durigon, V., & Carra, C. and methodology (pp. 365-391). Chicago: Aldine.
(1992). Effects of contextual interference on learning Breslin, G., Hodges, N.J., Kennedy, R., Hanlon, M., &
technical sports skills. Perceptual and Motor Skills, Williams, A.M. (2010). An especial skill: Support for
75, 555-562. a learned parameters hypothesis. Acta Psychologica,
Boucher, J.L. (1974). Higher processes in motor learning. 134, 55-60.
Journal of Motor Behavior, 6, 131-137. Bridgeman, B. (1996). Extraretinal signals in visual
Boucher, L., Palmeri, T.J., Logan, G.D., & Schall, J.D. orientation. In W. Prinz & B. Bridgeman (Eds.),
(2007). Inhibitory control in mind and brain: An Handbook of perception and action. Vol. 1: Perception
interactive race model of countermanding saccades. (pp. 191-223). San Diego: Academic Press.
Psychological Review, 114, 376-397. Bridgeman, B., Gemmer, A., Forsman, T., & Huemer,
Bourne, L.E. Jr., & Archer, E.J. (1956). Time continuously V. (2000). Processing spatial information in the
on target as a function of distribution of practice. sensorimotor branch of the visual system. Vision
Journal of Experimental Psychology, 51, 25-33. Research, 40, 3539-3552.
51 O References

Bridgeman, B., Kirch, M., & Sperling, A. (1981). Brown, T.G. (1911). The intrinsic factors in the act of
Segregation of cognitive and motor aspects of visual progression in the mammal. Proceedings of the Royal
information using induced motion. Perception & Society of London. Series B, Containing Papers of a
Psychophysics, 29, 336-342. Biological Character, 84, 308-319.
Briem, V, & Hedman, L.R. (1995). Behavioural effects Bruce, D. (1994). Lashley and the problem of serial order.
of mobile telephone use during simulated driving. American Psychologist, 49, 93-103.
Ergonomics, 38, 2536-2562. Bruno, N. (2001). When does action resist visual
Briggs, G.E., & Brogden, W.J. (1954). The effect of illusions? Trends in Cognitive Sciences, 5, 379-382.
component practice on performance of a lever- Bruno, N., Bernardis, P., & Gentilucci, M. (2008). Visually
positioning skill. Journal of Experimental Psychology, guided pointing, the Miiller-Lyer illusion, and
48, 375-380. the functional interpretation of the dorsal-ventral
Briggs, G.E., & Waters, L.K. (1958). Training and transfer split: Conclusions from 33 independent studies.
as a function of component interaction. Journal of Neuroscience & Biobehavioral Reviews, 32, 423-437.
Experimental Psychology, 56, 492-500.
Bruno, N., & Franz, V.H. (2009). When is grasping
Brisson, T.A., & Alain, C. (1996a). Should common affected by the Miiller-Lyer illusion? A quantitative
optimal movement patterns be identified as the review. Neuropsychologia, 47, 1421-1433.
criterion to be achieved? Journal of Motor Behavior,
Bryan, W.L., & Harter, N. (1897). Studies in the
28, 211-223.
physiology and psychology of the telegraphic
Brisson, T.A., & Alain, C. (1996b). Optimal movement language. Psychological Review, 4, 27-53.
pattern characteristics are not required as a reference
Bryan, W.L., & Harter, N. (1899). Studies on the
for knowledge of performance. Research Quarterly
telegraphic language: The acquisition of a hierarchy
for Exercise and Sport, 67, 458-464.
of habits. Psychological Review, 6, 345-375.
Broadbent, D.E. (1958). Perception and communication.
Buchanan, J.J., & Kelso, J.A.S. (1993). Posturally induced
London: Pergamon Press.
transitions in rhythmic multijoint limb movements.
Brodie, E.E., & Ross, H.E. (1985). Jiggling a lifted Experimental Brain Research, 94, 131-142.
weight does aid discrimination. American Journal of
Buekers, M.J., & Magill, R.A. (1995). The role of task
Psychology, 98, 469-471.
experience and prior knowledge for detecting invalid
Broker, J.P., Gregor, R.J., & Schmidt, R.A. (1993). Extrinsic augmented feedback while learning a motor skill.
feedback and the learning of kinetic patterns in Quarterly Journal of Experimental Psychology, 48A,
cycling. Journal of Applied Biomechanics, 9, 111-123. 84-97.
Brookhuis, K.A., de Vries, G., & de Waard, D. (1991). The Buekers, M.J., Magill, R.A., & Hall, K.G. (1992). The effect
effects of mobile telephoning on driving performance. of erroneous knowledge of results on skill acquisition
Accident Analysis and Prevention, 23, 309-316. when augmented information is redundant.
Brooks, VB. (1975). Roles of cerebellum and basal ganglia Quarterly Journal of Experimental Psychology, 44A,
and control of movements. Le Journal Canadien Des 105-117.
Sciences Neurologiques, 2, 265-277. Buekers, M.J., Magill, R.A., & Sneyers, K.M. (1994).
Brooks, VB. (1979). Motor programs revisited. In Resolving a conflict between sensory feedback and
R.E. Talbott & D.R. Humphrey (Eds.), Posture and knowledge of results, while learning a motor skill.
movement (pp. 13-49). New York: Raven Press. Journal of Motor Behavior, 26, 27-35.
Brooks, V.B. (1986). The neural basis of motor control. Burgess, P.R., & Clark, F.J. (1969). Characteristics of knee
New York: Oxford University Press. joint receptors in the cat. Journal of Physiology, 203,
Brown, I.D. (1962). Measuring the "spare mental 317-335.
capacity" of car drivers by a subsidiary auditory task. Burgess-Limerick, R., Neal, R.J., & Abernethy, B. (1992).
Ergonomics, 5, 247-250. Against relative timing invariance in movement
Brown, I.D. (1967). Measurement of control skills, kinematics. Quarterly Journal of Experimental
vigilance, and performance on a subsidiary task Psychology, 44A, 705-722.
during 12 hours of car driving. Ergonomics, 10, 665- Burton, D., Naylor, S., & Holliday, B. (2001). Goal setting
673. in sport: Investigating the goal effectiveness paradox.
Brown, I.D., Tickner, A.H., & Simmons, D.C.V (1969). In R.N. Singer, H.A. Hausenblas, & C.M. Janelle
Interference between concurrent tasks of driver and (Eds.), Handbook of sport psychology (2nd ed.) (pp.
telephoning. Journal of Applied Psychology, 53, 419- 497-528). New York: Wiley.
424. Butler, M.S., Reeve, T.G., & Fischman, M.G. (1996). Effects
Brown, J. (1958). Some tests of the decay theory of imme- of the instructional set in the bandwidth feedback
diate memory. Quarterly Journal of Experimental paradigm on motor skill acquisition. Research
Psychology, 10, 12-21. Quarterly for Exercise and Sport, 67, 335-359.
 References 511

Byblow, W.D., Lewis, G.N., Stinear, J.W., Austin, N.J., & Vision and motor control (pp. 3-31). Amsterdam:
Lynch, M. (2000). The subdominant hand increases Elsevier.
the efficacy of voluntary alterations in bimanual coor- Carlton, L.G. (1994). The effects of temporal-precision
dination. Experimental Brain Research, 131, 366-374. and time-minimization constraints on the spatial
Byblow, W.D., Summers, J.J., Lewis, G.N., & Thomas, J. and temporal accuracy of aimed hand movements.
(2002). Bimanual coordination in Parkinson's disease: Journal of Motor Behavior, 26, 43-50.
Deficits in movement frequency, amplitude, and Carlton, L.G., Carlton, M.J., & Kim, K.H. (1997).
pattern switching. Movement Disorders, 17, 20-29. Visuo-motor delays with changing environmental
Byrne, J. (Ed.). (2008). Learning and memory: A conditions. Unpublished manuscript.
comprehensive reference (4 vols.). Oxford: Elsevier.
Carlton, L.G., Chow, J.W., Ekkekakis, P., Shim, J.,
Caird, J.K., Willness, C.R., Steel, P., & Scialfa, C. (2008). Ichiyama, R., & Carlton, M.J. (1999). A Web-based
A meta-analysis of the effects of cell phones on driver digitized video image system for the study of
performance. Accident Analysis and Prevention, 40, motor coordination. Behavior Research Methods,
1282-1293. Instruments & Computers, 31, 57-62.
Camachon, C., Jacobs, D.M., Huet, M., Buekers, M., & Carlton, L.G., & Newell, K.M. (1988). Force variability
Montagne, G. (2007). The role of concurrent feedback and movement accuracy in space-time. Journal of
in learning to walk through sliding doors. Ecological Experimental Psychology: Human Perception and
Psychology, 19, 367-382. Performance, 14, 24-36.
Canic, M.J., & Franks, I.M. (1989). Response preparation Carnahan, H. (1992). Eye, head and hand coordination
and latency in patterns of tapping movements. during manual aiming. In L. Proteau & D. Elliott
Human Movement Science, 8, 123-139. (Eds.), Vision and motor control (pp. 179-196).
Card, S.K., English, W.K., & Burr, B.J. (1978). Evaluation Amsterdam: Elsevier.
of mouse, rate-controlled isometric joystick, step Carnahan, H., & Marteniuk, R.G. (1991). The temporal
keys, and text keys for text selection on a CRT. organization of hand, eye, and head movements
Ergonomics, 21, 601-613. during reaching and pointing. Journal of Motor
Cardinal, B.J., & Thomas, J.R. (2005). The 75th Behavior, 23, 109-119.
anniversary of Research Quarterly for Exercise Carroll, W.R., & Bandura, A (1990). Representational
and Sport: An analysis of status and contributions. guidance of action production in observational
Research Quarterly for Exercise and Sport, 76 (Suppl. learning: A causal analysis. Journal of Motor
2), S122-S134. Behavior, 22, 85-97.
Carey, D.P. (2001). Do action systems resist visual Carron, AV. (1967). Performance and learning in a
illusions? Trends in Cognitive Sciences, 5, 109-113. discrete motor task under massed versus distributed
Carlsen, AN., Chua, R., Inglis, J.T., Sanderson, D.J., & conditions. Unpublished doctoral dissertation,
Franks, I.M. (2004). Prepared movements are elicited University of California, Berkeley.
early by startle. Journal of Motor Behavior, 36, 253-264. Carron, AV. (1969). Performance and learning in a
Carlsen, AN., Dakin, C.J., Chua, R., & Franks, I.M. discrete motor task under massed vs. distributed
(2007). Startle produces early response latencies practice. Research Quarterly, 40, 481-489.
that are distinct from stimulus intensity effects.
Carron, AV., Loughhead, T.M., & Bray, S.R. (2005). The
Experimental Brain Research, 176, 199-205.
home advantage in sport competitions: Courneya
Carlson, R.A, & Yaure, R.G. (1990). Practice schedules and Carron's (1992) conceptual framework a decade
and the use of component skills in problem solving. later. Journal of Sports Sciences, 23, 395-407.
Journal of Experimental Psychology: Learning,
Carson, R.G., Byblow, W.D., Abernethy, B., & Summers,
Memory, and Cognition, 16, 484-496.
J.J. (1996). The contribution of inherent and incidental
Carlton, L.G. (1979). Control processes in the production constraints to intentional switching between patterns
of discrete aiming responses. Journal of Human of bimanual coordination. Human Movement
Movement Studies, 5, 115-124. Science, 15, 565-589.
Carlton, L.G. (1981a). Processing visual feedback Carson, R.G., Goodman, D., Kelso, J.AS., & Elliott, D.
information for movement control. Journal of (1995). Phase transitions and critical fluctuations in
Experimental Psychology: Human Perception and rhythmic coordination of ipsilateral hand and foot.
Performance, 7, 1019-1030. Journal of Motor Behavior, 27, 211-224.
Carlton, L.G. (1981b). Visual information: The control Carson, R.G., Riek, S., Smethurst, C.J., Parraga, J.F.,
of aimmg movements. Quarterly Journal of & Byblow, W.D. (2000). Neuromuscular-skeletal
Experimental Psychology, 33A, 87-93. constraints upon the dynamics of unimanual and
Carlton, L.G. (1992). Visual processing time and the bimanual coordination. Experimental Brain Research,
control of movement. In L. Proteau & D. Elliott (Eds.), 131, 196-214.
512 References

Carter, M.C., & Shapiro, D.C. (1984). Control of Chalmers, D. J. (1995, December). The puzzle of conscious
sequential movements: Evidence for generalized experience. Scientific American, 273(6), 80-86.
motor programs. Journal of Neurophysiology, 52, Chamberlin, C.J., & Coelho, A.J. (1993). The perceptual
787-796. side of action: Decision-making in sport. In J.L.
Carter, M.C., & Smith, J.L. (1986). Simultaneous Starkes & F. Allard (Eds.), Cognitive issues in motor
control of two rhythmical behaviors. I. Locomotion expertise (pp. 135-157). Amsterdam: Elsevier.
and paw-shake response in normal cat. Journal of Chambers, J.W. Jr., & Schumsky, D.A. (1978). The
Neurophysiology, 56, 171-183. compression block technique: Use and misuse in the
Caserta, R.J., Young, J., & Janelle, C.M. (2007). Old study of motor skills. Journal of Motor Behavior, 10,
dogs, new tricks: Training the perceptual skills of 301-311.
senior tennis players. Journal of Sport and Exercise Chan, T., & Chan, K. (1995). Effect of frequency ratio and
Psychology, 29, 479-497. environmental information on spatial coupling: A
Castellote, J.M., Kumm, H., Queralt, A., & Valls-Sole, J. study of attention. Ecological Psychology, 7, 125-144.
(2007). A startle speeds up the execution of externally Chapanis, A. (1951). Theory and methods for analyzing
guided saccades. Experimental Brain Research, 177, errors in man-machine systems. Annals of the New
129-136. York Academy of Sciences, 51, 1179-1203.
Casteneda, B., & Gray, R. (2007). Effects of focus of Chapanis, A. (1965). Man-machine engineering.
attention on baseball batting performance in players Belmont, CA: Wadsworth.
of differing skill levels. Journal of Sport and Exercise
Chapanis, A. (1999). The Chapanis chronicles: 50 years
Psychology, 29, 60-77.
of human factors research, education, and design.
Castiello, U. (1996). Grasping a fruit: Selection for Santa Barbara, CA: Agean.
action. Journal of Experimental Psychology: Human
Chapman, S. (1968). Catching a baseball. American
Perception and Performance, 22, 582-603.
Journal of Physics, 36, 868-870.
Castiello, U. (2005). The neuroscience of grasping.
Chase, W.G., & Simon, H.A. (1973). Perception in chess.
Nature Reviews: Neuroscience, 6, 726-736.
Cognitive Psychology, 4, 55-81.
Castiello, U., Bennett, K.M.B., & Stelmach, G.E. (1993).
Cheng,D.T.,Luis,M.,&Tremblay,L.(2008).Randomizing
Reach to grasp: The natural response to perturbation
visual feedback in manual aiming: Reminiscence of
of object size. Experimental Brain Research, 94, 163-
the previous trial condition and prior knowledge of
178.
feedback availability. Experimental Brain Research,
Castiello, U., & Umilta, C. (1988). Temporal dimensions 189, 403-410.
of mental effort in different sports. International Cherry, E.C. (1953). Some experiments on the recognition
Journal of Sport Psychology, 19, 199-210. of speech, with one and two ears. Journal of the
Catalano, J.F., & Kleiner, B.M. (1984). Distant transfer in Acoustical Society of America, 25, 975-979.
coincident timing as a function of practice variability. Chisholm, S.L., Caird, J.K., & Lockhart, J. (2008). The
Perceptual and Motor Skills, 58, 851-856. effects of practice with MP3 players on driving
Cattell, J.M. (1886). The time it takes to see and name performance. Accident Analysis and Prevention, 40,
objects. Mind, 11, 63-65. 704-713.
Cattell, J.M. (1893). Aufmerksamkeit und reaction. Chiviacowsky, S., & Wulf, G. (2002). Self-controlled
Philosophische Studien, 8, 403-406. English feedback: Does it enhance learning because
translation in R.S. Woodworth (1947), Psychological performers get feedback when they need it? Research
research (Vol. 1) (pp. 252-255). Lancaster, PA: Science Quarterly for Exercise and Sport, 73, 408-415.
Press. Chiviacowsky, S., & Wulf, G. (2005). Self-controlled
Cauraugh, J.H., Chen, D., & Singer, R.N. (1993). Graphic feedback is effective if it is based on the learner's
versus numeric knowledge of results: Which mode? performance. Research Quarterly for Exercise and
Research Quarterly for Exercise and Sport, 64, 213-216. Sport, 76, 42-48.
Cavallo, V., & Laurent, M. (1988). Visual information and Chiviacowsky, S., & Wulf, G. (2007). Feedback after
skill level in time-to-collision estimation. Perception, good trials enhances learning. Research Quarterly for
17, 623-632. Exercise and Sport, 78, 40-47.
Cerritelli, B., Maruff, P., Wilson, P., & Currie, J. (2000). Cho, Y.S., & Proctor, R.W. (2003). Stimulus and response
The effect of an external load on the force and representations underlying orthogonal stimulus-
timing components of mentally represented actions. response compatibility effects. Psychonomic Bulletin
Behavioural Brain Research, 108, 91-96. & Review, 10, 45-73.
Chabris, C.F., & Simons, D.J. (2010). The invisible gorilla: Choi, Y., Qi, F., Gordon, J., & Schweighofer, N. (2008).
And other ways our intuitions deceive us. New York: Performance-based adaptive schedules enhance motor
Crown. learning. Journal of Motor Behavior, 40, 273-280.
 References 513

Christina, R.W. (1992). The 1991 C.H. McCloy research Cooke, N.J. (2008). Preface to the special 50th anniversary
lecture: Unraveling the mystery of the response issue of Human Factors. Human Factors, 50, 347-350.
complexity effect in skilled movements. Research Cooper, J.M., & Strayer, D.L. (2008). Effects of simulator
Quarterly for Exercise and Sport, 63, 218-230. practice and real-world experience on cell-phone-
Chua, R., & Elliott, D. (1993). Visual regulation of manual related driver distraction. Human Factors, 50, 893-
aiming. Human Movement Science, 12, 365-401. 902.
Cirstea, M.C., & Levin, M.F. (2007). Improvement of arm Cooper, J.B., & Taqueti, V.R. (2009). A brief history of the
movement patterns and endpoint control depends on development of mannequin simulators for clinical
type of feedback during practice in stroke survivors. education and training. Postgraduate Medical
Neurorehabilitation and Neural Repair, 21, 398-411. Journal, 84, 563-570.
Cirstea, M.C., Ptito, A., & Levin, M.F. (2006). Feedback Coreas, D.M. (1984). Two-handed movement control.
and cognition in arm motor skill reacquisition after Research Quarterly for Exercise and Sport, 55, 117-
stroke. Stroke, 37, 1237-1242. 122.
Claxton, G. (1975). Why can't we tickle ourselves? Coreas, D.M., Jaric, S., & Gottlieb, G.L. (1996).
Perceptual and Motor Skills, 41, 335-338. Electromyographic analysis of performance
Cleary, T.J., Zimmerman, B.J., & Keating, T. (2006). enhancement. In H.N. Zelaznik (Ed.), Advances in
Training physical education students to self-regulate motor learning and control (pp. 123-153). Champaign,
during basketball free throw practice. Research IL: Human Kinetics.
Quarterly for Exercise and Sport, 77, 251-262. Cardo, P. J., & N ashner, L.M. (1982). Properties of postural
Cohen, J.D., Dunbar, K., & McClelland, J.L. (1990). On the adjustments associated with rapid arm movements.
control of automatic processes: A parallel distributed Journal of Neurophysiology, 47, 287-302.
processing account of the Stroop effect. Psychological Cormier, S.M., & Hagman, J.D. (Eds.). (1987). Transfer of
Review, 97, 332-361. learning: Contemporary research applications. New
York: Academic Press.
Cohen, J., & Schooler, J.W. (Eds.). (1996). Scientific
approaches to consciousness. Mahwah, NJ: Erlbaum. Courneya, KS., & Carron, A.V. (1992). The home
advantage in sport competitions: A literature review.
Cohen, L. (1971). Synchronous bimanual movements
Journal of Sport and Exercise Psychology, 14, 13-27.
performed by homologous and non-homologous
muscles. Perceptual and Motor Skills, 32, 639-644. Court, M.L.J., Bennett, S.J., Williams, A.M., & Davids,
K. (2005). Effects of attentional strategies and anxiety
Cohen, P.J. (1997). Cellular telephones and traffic
constraints on perceptual-motor organisation of
accidents (correspondence). New England Journal of
rhythmical arm movements. Neuroscience Letters,
Medicine, 337, 127-129.
384, 17-22.
Cole, K.J., & Abbs, J.H. (1988). Grip force adjustments
Crago, P.E., Houk, J.C., & Hasan, Z. (1976). Regulatory
evoked by load force perturbations of a grasped
actions of human stretch reflex. Journal of
object. Journal of Neurophysiology, 60, 1513-1522.
Neurophysiology, 39, 925-935.
Collet, C., Guillot, A., & Petit, C. (2010). Phoning while
Craik, F.I.M., & Jacoby, L.L. (1996). Aging and memory:
driving I: A review of epidemiological, psychological,
Implications for skilled performance. In W.A. Rogers,
behavioural and physiological studies. Ergonomics,
A.D. Fisk, & N. Walker (Eds.), Aging and skilled
53, 589-601.
performance: Advances in theory and applications
Collet, C., Guillot, A., & Petit, C. (2010). Phoning while (pp. 113-137). Mahwah, NJ: Erlbaum.
driving II: A review of driving conditions influence.
Craik, K.J.W. (1948). The theory of the human operator
Ergonomics, 53, 602-616.
in control systems: IL Man as an element in a control
Collier, G.L., & Ogden, R.T. (2004). Adding drift to the system. British Journal of Psychology, 38, 142-148.
decomposition of simple isochronous tapping: An
Cratty, B.J. (1964). Movement behavior and motor
extension of the Wing-Kristofferson model. Journal
learning. Philadelphia: Lea & Febiger.
of Experimental Psychology: Human Perception and
Performance, 30, 853-872. Crawley, S.L. (1926). An experimental investigation of
recovery from work. Archives of Psychology, 13, 85.
Consiglio, W., Driscoll, P., Witte, M., & Berg, W.P. (2003).
Effect of cellular telephone conversations and other Creamer, L.R. (1963). Event uncertainty, psychological
potential interference on reaction time in a braking refractory period, and human data processing.
response. Accident Analysis and Prevention, 35, 495- Journal of Experimental Psychology, 66, 187-194.
500. Creem, S.H., & Proffitt, D.R. (2001). Defining the cortical
Cooke, J.D. (1980). The organization of simple, visual systems: "What", "where", and "how." Acta
skilled movements. In G.E. Stelmach & J. Requin Psychologica, 107, 43-68.
(Eds.), Tutorials in motor behavior (pp. 199-212). Crick, F., & Koch, C. (2003). A framework for
Amsterdam: Elsevier. consciousness. Nature Neuroscience, 6, 119-126.
514 111 References

Criscimagna-Hemminger, S.E., & Shadmehr, R. (2008). Decety, J., & Jeannerod, M. (1996). Mentally simulated
Consolidation patterns of human motor memory. movements in virtual reality: Does Fitts's law hold
Journal of Neuroscience, 28, 9610-9618. in motor imagery? Behavioural Brain Research, 72,
Croll, W.L. (1970). Children's discrimination learning as 127-134.
a function of intertrial interval duration. Psychonornic Dees, V.,&Grindley,G.C. (1951). The effect of knowledge
Science, 18, 321-322. of results on learning and performance: IV. The
Cronbach, L.J. (1957). The two disciplines of scientific direction of the error in very simple skills. Quarterly
psychology. American Psychologist, 12, 671-684. Journal of Experimental Psychology, 3, 36-42.
Cross, E.S., Schmitt, P.J., & Grafton, S.T. (2007). Neural deGroot,A.D. (1946/1978). Thought and choice in chess.
substrates of contextual interference during motor The Hague: Mouton. (Original work published in
learning support a model of active preparation. 1946)
Journal of Cognitive Neuroscience, 19, 1854-1871. DeJaeger, D., & Proteau, L. (2003). The relative efficacy
of different forms of knowledge of results for the
Crossman, E.R.F.W. (1959). A theory of the acquisition of
speed skill. Ergonomics, 2, 153-166. learning of a new relative timing pattern. Quarterly
Journal of Experimental Psychology, 56A, 621-640.
Crossman, E.R.F.W., & Goodeve, P.J. (1963/1983).
De Jong, R., Coles, M.G.H., Logan, G.D., & Gratton, G.
Feedback control of hand-movements and Fitts' law.
(1990). In search of the point of no return: The control
Paper presented at the meeting of the Experimental
of response processes. Journal of Experimental
Psychology Society, Oxford, July, 1963. Published in
Psychology: Human Perception and Performance,
Quarterly Journal of Experimental Psychology, 1983,
16, 164-182.
35A, 251-278.
Del Rey, P. (1989). Training and contextual interference
Cruse, H., Dean, J., Heuer, H., & Schmidt, RA (1990).
effects on memory and transfer. Research Quarterly
Utilization of sensory information for motor control.
for Exercise and Sport, 60, 342-347.
In 0. Neumann & W. Prinz (Eds.), Relationships
between perception and action: Current approaches Del Rey, P., Liu, X., & Simpson, K.J. (1994). Does
(pp. 43-79). Berlin: Springer-Verlag. retroactive inhibition influence contextual
interference effects? Research Quarterly for Exercise
Cuddy, L.J., & Jacoby, L.L. (1982). When forgetting helps and Sport, 65, 120-126.
memory. An analysis of repetition effects. Journal of
Verbal Learning and Verbal Behavior, 21, 451-467. Del Rey, P., & Shewokis, P. (1993). Appropriate summary
KR for learning timing tasks under conditions of high
Custers, E.J.F.M., Regehr, G., McCulloch, W., Peniston, and low contextual interference. Acta Psychologica,
C., & Reznick, R. (1999). The effects of modeling on 83, 1-12.
learning a simple surgical procedure: See one, do one
or see many do one? Advances in Health Sciences Del Rey, P., Whitehurst, M., & Wood, J.M. (1983). Effects
Education, 4, 123-143. of experience and contextual interference on learning
and transfer by boys and girls. Perceptual and Motor
Dail, T.K., & Christina, R.W. (2004). Distribution of Skills, 56, 581-582.
practice and metacognition in learning and retention
of a discrete motor task. Research Quarterly for Del Rey, P., Wughalter, E.H., & Whitehurst, M. (1982).
Exercise and Sport, 75, 148-155. The effects of contextual interference on females
with varied experience in open sport skills. Research
Daniels, G.L., & Newell, K.M. (2003). Attentional Quarterly for Exercise and Sport, 53, 108-115.
focus influences the walk-run transition in human
locomotion. Biological Psychology, 63, 163-178. Dempster, F.N. (1988). The spacing effect: A case in the
failure to apply the results of psychological research.
Darling, W.G., & Cooke, J.D. (1987). Changes in the American Psychologist, 43, 627-634.
variability of movement trajectories with practice.
Denier van der Gon, J.J., & Thuring, J.P. (1965). The
Journal of Motor Behavior, 19, 291-309.
guiding of human writing movements. Kybernetik,
Davids, K., Button, C., & Bennett, S. (2008). Dynamics 2, 145-148.
of skill acquisition: A constraints-led approach.
Department of Transportation. (2000). Denial of Motor
Champaign, IL: Human Kinetics.
Vehicle Defect Petition, DP99-004. Federal Register,
Davies, G.M., & Thomson, D.M. (1988). Memory in 65,No. 83,25026-25037.
context: Context in memory. New York: Wiley.
Desmurget, M., & Grafton, S. (2000). Forward modeling
Davis, R. (1959). The role of "attention" in the allows feedback control for fast reaching movements.
psychological refractory period. Quarterly Journal of Trends in Cognitive Sciences, 4, 423-431.
Experimental Psychology, 11, 211-220. Desmurget, M., Prablanc, C., Arzi, M., Rossetti, Y.,
Deakin, J.M., & Proteau, L. (2000). The role of scheduling Paulignan, Y., & Urquizar, C. (1996). Integrated control
in learning through observation. Journal of Motor of hand transport and orientation during prehension
Behavior, 32, 268-276. movements. Experimental Brain Research, 110, 265-278.
 References 515

Desrosiers, J., Hebert, R., Bravo, G., & Dutil, E. (1995). Drazin, D.H. (1961). Effects of foreperiod, foreperiod
Upper-extremity motor co-ordination of healthy variability, and probability of stimulus occurrence
elderly people. Age and Ageing, 24, 108-112. on simple reaction time. Journal of Experimental
Deutsch, D. (1983). The generation of two isochronous Psychology, 62, 43-50.
sequences in parallel. Perception & Psychophysics, Drewing, K., & Aschersleben, G. (2003). Reduced
34, 331-337. timing variability during bimanual coupling: A
Deutsch, J.E., Borbely, M., Filler,}., Huhn, K., & Guarrera- role for sensory information. Quarterly Journal of
Bowlby, P. (2008). Use of a low-cost, commercially Experimental Psychology, 56A, 329-350.
available gaming console (Wii) for rehabilitation of Drews, P.A., Pasupathi, M., & Strayer, D.L. (2008).
an adolescent with cerebral palsy. Physical Therapy, Passenger and cell phone conversations in simulated
88, 1196-1207. driving. Journal of Experimental Psychology:
Deutsch, J.A., & Deutsch, D. (1963). Attention: Some Applied, 14, 392-400.
theoretical considerations. Psychological Review, 70, Driver, J., Davis, G., Russell, C., Turatto, M., & Freeman,
80-90. E. (2001). Segmentation, attention and phenomenal
Dewhurst, D.J. (1967). Neuromuscular control system. visual objects. Cognition, 80, 61-95.
IEEE Transactions on Biomedical Engineering, 14, Druckman, D., & Bjork, R.A. (1991). In the mind's eye:
167-171. Enhancing human performance. Washington, DC:
National Academy Press.
Dick, M.B., Hsieh, S., Bricker, J., & Dick-Muehlke, C.
(2003). Facilitating acquisition and transfer of a Druckman, D., & Bjork, R.A. (1994). Learning, remem-
continuous motor task in healthy older adults and bering, believing: Enhancing human performance.
patients with Alzheimer's disease. Neuropsychology, Washington, DC: National Academy Press.
17, 202-212. Drury, C.G., & Hoffmann, E.R. (1992). A model
Dickstein, R., & Deutsch, J.E. (2007). Motor imagery in for movement time on data-entry keyboards.
physical therapist practice. Physical Therapy, 87, 942- Ergonomics, 35, 129-147.
953. Drury, C.G., & Woolley, S.M. (1995). Visually-controlled
Diedrich, F.J., & Warren, W.H. Jr. (1995). Why change leg movements embedded in a walking task.
gaits? Dynamics of the walk-run transition. Journal Ergonomics, 38, 714-722.
of Experimental Psychology: Human Perception and Dubrowski, A. Backstein, D., Abughaduma, R., Leidl,
Performance, 21, 183-202. D., & Carnahan, H. (2005). The influence of practice
Diedrich, F.J., & Warren, W.H. Jr. (1998a). The dynamics schedules in the learning of a complex bone-plating
of gait transitions: Effects of grade and load. Journal surgical task. American Journal of Surgery, 190, 359-
of Motor Behavior, 30, 60-78. 363.
Diedrich, F.J., & Warren, W.H. Jr. (1998b). Dynamics Dudink, A. (1994). Birth date and sporting success.
of human gait transitions. In D.A. Rosenbaum & Nature, 368, 592.
C.E. Collyer (Eds.), Timing of behavior: Neural, Duffy, E. (1962). Activation and behavior. New York:
psychological, and computational perspectives (pp. Wiley.
323-343). Cambridge, MA: MIT Press.
Dunbar, K.N., & MacLeod, C.M. (1984). A horse race
Dietz, V., & Duysens, J. (2000). Significance of load of a different color: Stroop interference patterns
receptor input during locomotion: A review. Gait & with transformed words. Journal of Experimental
Posture, 11, 102-110. Psychology: Human Perception and Performance,
Domingo, A., & Ferris, D.P. (2009). Effects of physical 10, 622-639.
guidance on short-term learning of walking on a Duncan-Johnson, C.C., & Donchin, E. (1982). The P300
narrow beam. Gait & Posture, 30, 464-468. component of the event-related brain potential
Donders, F.C. (1969). On the speed of mental processes. as an index of information processing. Biological
In W.G. Koster (Ed. & Trans.), Attention and Psychology, 14, 1-52.
performance II. Amsterdam: North-Holland. Dunlap, K. (1910). Reactions to rhythmic stimuli, with
(Original work published in 1868) attempt to synchronize. Psychological Review, 17,
Donovan, J.J., & Radosevich, D.J. (1999). A meta-analytic 399-416.
review of the distribution of practice effect: Now you Duysens, J., & Van de Crommert, H.W.A.A. (1998).
see it, now you don't. Journal of Applied Psychology, Neural control of locomotion; part 1: The central
84, 795-805. pattern generator from cats to humans. Gait &
Doody, S.G., Bird, A.M., & Ross, D. (1985). The effect of Posture, 7, 131-141.
auditory and visual models on acquisition of a timing Dye, M.W.G., Green, C.S., & Bavelier, D. (2009).
task. Human Movement Science, 4, 271-281. Increasing speed of processing with action video
516 References

games. Current Directions in Psychological Science, Elliott, D., & Madalena, J. (1987). The influence of
18, 321-326. premovement visual information on manual aiming.
Easterbrook, J.A. (1959). The effect of emotion on Quarterly Journal of Experimental Psychology, 39A,
cue utilization and the organization of behavior. 541-559.
Psychological Review, 66, 183-201. Elliott, D., & Roy, E.A. (Eds.). (1996). Manual asymmetries
Easton, T.A. (1972). On the normal use of reflexes. in motor performance. Boca Raton, FL: CRC Press.
American Scientist, 60, 591-599. Elliott, D., Weeks, D.J., & Elliott, C.L. (1987). Cerebral
Easton, T.A. (1978). Coordinative structures-the specialization in individuals with Down's syndrome.
basis for a motor program. In D.M. Landers & R.W. American Journal on Mental Retardation, 92, 263-271.
Christina (Eds.), Psychology of motor behavior and Elliott, D., Zuberec, S., & Milgram, P. (1994). The effects
sport (pp. 63-81). Champaign, IL: Human Kinetics. of periodic visual occlusion on ball catching. Journal
Ebbinghaus, H.D. (1913). Memory: A contribution of Motor Behavior, 26, 113-122.
to experimental psychology (H.A. Ruger & C.E. Ellis, H.C. (1965). The transfer of learning. New York:
Bussenius, Trans.). New York: Teachers Colleges. Macmillan.
(Original work published in 1885) Ellis, R.R., Flanagan, J.R., & Lederman, S.J. (1999).
Eccles, D.W., & Tenenbaum, G. (2004). Why an expert The influence of visual illusions on grasp position.
team is more than a team of experts: A social- Experimental Brain Research, 125, 109-114.
cognitive conceptualization of team coordination
Ells, J.C. (1973). Analysis of temporal and attentional
and communication in sport. Journal of Sport and
aspects of movement control. Journal of Experimental
Exercise Psychology, 26, 542-560.
Psychology, 99, 10-21.
Eimer, M., Nattkemper, D., Schri:iger, E., & Prinz, W.
Elwell, J.L., & Grindley, G.C. (1938). The effect of
(1996). Involuntary attention. In 0. Neumann & A.F.
knowledge of results on learning and performance.
Sanders (Eds.), Handbook of perception and action.
British Journal of Psychology, 29, 39-54.
Vol. 3: Attention (pp. 155-184). San Diego: Academic
Press. Enebo, B., & Sherwood, D.E. (2005). Experience and
practice organization in learning a simulated high-
Elliott, D. (1990). Intermittent visual pickup and goal
velocity low-amplitude task. Journal of Manipulative
directed movement: A review. Human Movement
and Physiological Therapeutics, 28, 33-43.
Science, 9, 531-548.
English, H.B. (1942). How psychology can facilitate
Elliott, D. (1992). Intermittent versus continuous control
military training-a concrete example. Journal of
of manual aiming movements. In L. Proteau & D.
Applied Psychology, 26, 3-7.
Elliott (Eds.), Vision and motor control (pp. 33-48).
Amsterdam: Elsevier. Ericsson, K.A. (Ed.). (1996). The road to excellence: The
acquisition of expert performance in the arts and
Elliott, D., & Allard, F. (1985). The utilization of visual
sciences, sports, and games. Mahwah, NJ: Erlbaum.
feedback information during rapid pointing
movements. Quarterly Journal of Experimental Ericsson, K.A., Chase, W.G., & Faloon, S. (1980).
Psychology, 37A, 407-425. Acquisition of a memory skill. Science, 208, 1181-
1182.
Elliott, D., Binsted, G.,& Heath, M. (1999). The control
of goal-directed limb movements: Correcting errors Ericsson, K.A., Krampe, R.Th., & Tesch-Romer, C. (1993).
in the trajectory. Human Movement Science, 18, 121- The role of deliberate practice in the acquisition of
136. expert performance. Psychological Review, 100, 363-
406.
Elliott, D., Chua, R., & Pollock, B.J. (1994). The influence
of intermittent vision on manual aiming. Acta Espenschade, A. (1940). Motor performance in
Psychologica, 85, 1-13. adolescence including the study of relationships
Elliott, D., Hansen, S., Grierson, L.E.M., Lyons, J., with measures of physical growth and maturity.
Bennett, S.J., & Hayes, S.J. (2010). Goal-directed Monographs of the Society for Research in Child
aiming: Two components but multiple processes. Development, 5, 1-126.
Psychological Bulletin, 136, 1023-1044. Estes, W.K. (1956). The problem of inference from curves
Elliott, D., Helsen, W.F., & Chua, R. (2001). A century based on group data. Psychological Bulletin, 53, 134-
later: Woodworth's (1899) two-component model 140.
of goal-directed aiming. Psychological Bulletin, 127, Evarts, E.V. (1972). Contrasts between activity of
342-357. precentral and postcentral neurons of cerebral cortex
Elliott, D., Lyons, J., & Dyson, K. (1997). Rescaling an during movement in the monkey. Brain Research, 40,
acquired discrete aiming movement: Specific or 25-31.
general motor learning? Human Movement Science, Evarts, E.V. (1973). Motor cortex reflexes associated with
16, 81-96. learned movement. Science, 179, 501-503.
 References 517

Fagg, A.H., & Arbib, M.A. (1998). Modeling parietal- Feldman, A.G., & Levin, M.F. (1995). The origin and use
premotor interactions in primate control of grasping. of positional frames of reference in motor control.
Neural Networks, 11, 1277-1303. Behavioral and Brain Sciences, 18, 723-806.
Fantino, E., & Logan, C.A. (1979). The experimental Feldman, A.G., & Levin, M.F. (2009). The equilibrium-
analysis of behavior: A biological perspective. San point hypothesis - past, present and future. In D.
Francisco: Freeman. Sternad (Ed.), Progress in motor control (pp. 699-
Farnsworth, P.R., & Poynter, W.F. (1931). A case of 726). Berlin: Springer.
unusual ability in simultaneous tapping in two Feldman, A.G., Ostry, D.J., Levin, M.F., Gribble, P.L., &
different times. American Journal of Psychology, 43, Mitnitski, AB. (1998). Recent tests of the equilibrium-
633. point hypothesis (A model). Motor Control, 2, 189-
Farrell, J.E. (1975). The classification of physical 205.
education skills. Quest, 24, 63-68. Feltz, D.L., & Landers, D.M. (1983). The effects of mental
Farrell, M.J., & Thomson, J.A. (1998). Automatic practice on motor skill learning and performance: A
spatial updating during locomotion without vision. meta-analysis. Journal of Sport Psychology, 5, 25-57.
Quarterly Journal of Experimental Psychology, 51A, Feltz, D.L., Landers, D.M., & Becker, B.J. (1988). A revised
637-654. meta-analysis of the mental practice literature on
Farrow, D., & Abernethy, B. (2002). Can anticipatory motor skill learning. In D. Druckman & J.A. Swets
skills be learned through implicit video-based (Eds.), Enhancing human performance: Issues,
perceptual training? Journal of Sports Sciences, 20, theories, and techniques. Background papers (part
471-485. III, chapter 5, pp. 1-65). Washington, DC: National
Academy Press.
Farrow, D., Baker, J., & MacMahon, C. (2008). Developing
sport expertise: Researchers and coaches put theory Fendrich, D.W., Healy, AF., & Bourne, L.E. Jr. (1991).
into practice. New York: Routledge. Long-term repetition effects for motoric and
perceptual procedures. Journal of Experimental
Faugloire, E., Bardy, B.G., & Stoffregen, T.A. (2009). (De)
Psychology: Learning, Memory, and Cognition, 17,
stabilization of required and spontaneous postural
137-151.
dynamics with learning. Journal of Experimental
Psychology: Human Perception and Performance, Ferrari, M. (1996). Observing the observer: Self-
35, 170-187. regulation in the observational learning of motor
skills. Developmental Review, 16, 203-240.
Fecteau, J.H., & Munoz, D.P. (2003). Exploring the
consequences of the previous trial. Nature Reviews: Ferrier, D. (1888). Discussions on cerebral localization.
Neuroscience, 4, 435-443. Transactions of the Congress of American Physicians
and Surgeons, 1, 337-340.
Feigenberg, I.M., & Latash, L.P. (1996). N.A. Bernstein:
The reformer of neuroscience. In M.L. Latash & M.T. Fialkow, M.F., & Goff, B.A. (2009). Training the next
Turvey (Eds.), Dexterity and its development (pp. generation of minimally invasive surgeons. Journal
247-275). Champaign, IL: Human Kinetics. of Minimally Invasive Gynecology, 16, 136-141.
Feldman, A.G. (1966a). Functional tuning of the nervous Fischer, M.H. (2001). How sensitive is hand transport to
system with control of movement or maintenance of illusory context effects? Experimental Brain Research,
a steady posture-II. Controllable parameters of the 136, 224-230.
muscles. Biophysics, 11, 565-578. Fischer, M.H., Pratt, J., & Neggers, S.F.W. (2003).
Feldman, A.G. (1966b ). Functional tuning of the nervous Inhibition of return and manual pointing movements.
system during control of movement or maintenance Perception & Psychophysics, 65, 379-387.
of a steady posture-III. Mechanographic analysis Fischman, M.G. (1984). Programming time as a function
of the execution by man of the simplest motor tasks. of number of movement parts and changes in
Biophysics, 11, 667-675. movement direction. Journal of Motor Behavior, 16,
Feldman, A.G. (1986). Once more on the equilibrium- 405-423.
point hypothesis (A model) for motor control. Journal Fischman, M.G., Christina, R.W., & Anson, J.G. (2008).
of Motor Behavior, 18, 17-54. Memory drum theory's C movement: Revelations
Feldman, A.G. (2009). Origin and advances of the from Franklin Henry. Research Quarterly for Exercise
equilibrium-point hypothesis. In D. Sternad (Ed.), and Sport, 79, 312-318.
Progress in motor control (pp. 637-643). Berlin: Fischman, M.G., Stodden, D.F., & Lehman, D.M. (2003).
Springer. The end-state comfort effect in bimanual grip
Feldman, A.G., & Latash, M.L. (2005). Testing selection. Research Quarterly for Exercise and Sport,
hypotheses and the advancement of science: Recent 74, 17-24.
attempts to falsify the equilibrium point hypothesis. Fitch, H.L., Tuller, B., & Turvey, M.T. (1982). The Bernstein
Experimental Brain Research, 161, 91-103. perspective: III. Tuning of coordinative structures
518 References

with special reference to perception. In J.A.S. Kelso Fleishman, E.A. (1975). Toward a taxonomy of human
(Ed.), Human motor behavior: An introduction (pp. performance. American Psychologist, 30, 1127-1149.
271-281). Hillsdale, NJ: Erlbaum. Fleishman, E.A. (2004). Gold medal award for life
Fitts, P.M. (1954). The information capacity of the achievement in the application of psychology.
human motor system in controlling the amplitude of American Psychologist, 59, 352-354.
movement. Journal of Experimental Psychology, 47, Fleishman, E.A., & Bartlett, C.J. (1969). Human abilities.
381-391. Annual Review of Psychology, 20, 349-380.
Fitts, P.M. (1964). Perceptual-motor skills learning. In Fleishman, E.A., & Hempel, W.E. Jr. (1955). The relation
AW. Melton (Ed.), Categories of human learning between abilities and improvement with practice
(pp. 243-285). New York: Academic Press. in a visual discrimination reaction task. Journal of
Fitts, P.M., Bahrick, H.P., Noble, M.E., & Briggs, G.E. Experimental Psychology, 49, 301-312.
(1959). Skilled performance (Contract No. AF 41 Fleishman, E.A., & Hempel, W.E. Jr. (1956). Factorial
[657]-70). Columbus, OH: Ohio State University, analysis of complex psychomotor performance and
Wright Air Development Center. related skills. Journal of Applied Psychology, 40, 96-
Fitts, P.M., & Deininger, R.L. (1954). S-R compatibility: 104.
Correspondence among paired elements within Fleishman, E.A., & Parker, J.F. (1962). Factors in the
stimulus and response codes. Journal of Experimental retention and relearning of perceptual motor skill.
Psychology, 48, 483-492. Journal of Experimental Psychology, 64, 215-226.
Fitts, P.M., & Jones, R.E. (1947). Analysis of factors Fleishman, E.A., & Rich, S. (1963). Role of kinesthetic and
contributing to 460 "pilot error" experiences in spatial-visual abilities in perceptual-motor learning.
operating aircraft controls (Memorandum Rep. Journal of Experimental Psychology, 66, 6-11.
TSEA4-694-12, Aero Medical Laboratory). Drayton,
OH: Wright Patterson Air Force Base. Fleishman, E.A., & Stephenson, R.W. (1970).
Development of a taxonomy of human performance:
Fitts, P.M., & Peterson, J.R. (1964). Information capacity A review of the third year's progress (Tech. Rep. No.
of discrete motor responses. Journal of Experimental 726-TPR3). Silver Springs, MD: American Institutes
Psychology, 67, 103-112. for Research.
Fitts, P.M., & Posner, M.I. (1967). Human performance. Folk, C.L., & Gibson, B.S. (Eds.). (2001). Attraction,
Belmont, CA: Brooks/Cole. distraction and action: Multiple perspectives on
Fitts, P.M., & Seeger, C.M. (1953). S-R compatibility: attentional capture. Amsterdam: Elsevier.
Spatial characteristics of stimulus and response codes. Fontaine, R.J., Lee, T.D., & Swinnen, S.P. (1997). Learning
Journal of Experimental Psychology, 46, 199-210. a new bimanual coordination pattern: Reciprocal
Flach, J.M., Guisinger, M.A., & Robison, AB. (1996). influences of intrinsic and to-be-learned patterns.
Fitts's law: Nonlinear dynamics and positive entropy. Canadian Journal of Experimental Psychology, 51, 1-9.
Ecological Psychology, 8, 281-325. Ford, P., Hodges, N.J., & Williams, AM. (2005). Online
Flegal, K.E., & Anderson, M.C. (2008). Overthinking attentional-focus manipulations in a soccer-dribbling
skilled motor performance: Or why those who teach task: Implications for the proceduralization of motor
can't do. Psychonomic Bulletin & Review, 15, 927-932. skills. Journal of Motor Behavior, 37, 386-394.
Fleishman, E.A. (1956). Psychomotor selection tests: Ford, P., Hodges, N.J., & Williams, AM. (2007).
Research and application in the United States Air Examining action effects in the execution of a skilled
Force. Personnel Psychology, 9, 449-467. soccer kick by using erroneous feedback. Journal of
Fleishman, E.A. (1957). A comparative study of aptitude Motor Behavior, 39, 481-490.
patterns in unskilled and skilled psychomotor Forssberg, H., Grillner, S., & Rossignol, S. (1975). Phase
performances. Journal of Applied Psychology, 41, dependent reflex reversal during walking in chronic
263-272. spinal cats. Brain Research, 85, 103-107.
Fleishman, E.A. (1964). The structure and measurement Fowler, B., Duck, T., Mosher, M., & Mathieson, B. (1991).
of physical fitness. Englewood Cliffs, NJ: Prentice The coordination of bimanual aiming movements:
Hall. Evidence for progressive desynchronization.
Fleishman, E.A. (1965). The description and prediction Quarterly Journal of Experimental Psychology, 43A,
of perceptual-motor skill learning. In R. Glaser (Ed.), 205-221.
Training research and education (pp. 137-175). New Fowler, B., Meehan, S., & Singhal, A. (2008). Perceptual-
York: Wiley. motor performance and associated kinematics in
Fleishman, E.A. (1967). Individual differences and space. Human Factors, 50, 879-892.
motor learning. In R.M. Gagne (Ed.), Learning and Fowler, C.A., Richardson, M.J., Marsh, K.L., & Shockley,
individual differences (pp. 165-191). Columbus, OH: K.D. (2008). Language use, coordination, and the
Merrill. emergence of cooperative action. In A. Fuchs & V.K.
 References 519

Jirsa (Eds.), Coordination: Neural, behavioral and Fritsch, G., & Hitzig, E. (1870). Ober die elektrische
social dynamics (pp. 261-279). Berlin: Springer. Erregbarkeit des Grosshirns. Archiv Anatomie
Fozard, J.L., Vercruyssen, M., Reynolds, S.L., Hancock, Physiologie, 37, 300-332.
P.A., & Quilter, R.E. (1994). Age differences and Fuchs, A.H. (1962). The progression-regression
changes in reaction time: The Baltimore longitudinal hypothesis in perceptual-motor skill learning. Journal
study of aging. Journal of Gerontology: Psychological of Experimental Psychology, 63, 177-182.
Sciences, 49, P179-P189. Fuchs, A.H. (1998). Psychology and "The Babe." Journal
Fraise, P. (1969). Why is naming longer than reading? of the History of Behavioral Sciences, 34, 153-165.
Acta Psychologica, 30, 96-103. Fuchs, A., & Jirsa, V.K. (2000). The HKB model revisited:
Frak, V.,Paulignan, Y.,&Jeannerod,M. (2001). Orientation How varying the degree of symmetry controls
of the opposition axis in mentally simulated grasping. dynamics. Human Movement Science, 19, 425-449.
Experimental Brain Research, 136, 120-127. Fukson, O.I., Berkinblit, M.B., & Feldman, A.G. (1980).
Frank, J.S., Williams, I.D., & Hayes, K.C. (1977). The The spinal frog takes into account the scheme of its
ischemic nerve block and skilled movement. Journal body during the wiping reflex. Science, 209, 1261-1263.
of Motor Behavior, 9, 217-224. Fukuda, T. (1961). Studies on human dynamic postures
Franks, I.M., & Harvey, T. (1997). Cues for goalkeepers: from the viewpoint of postural reflexes. Acta Oto-
High-tech methods used to measure penalty shot Laryngologica, 161, 1-52.
response. Soccer Journal, 42(3), 30-33, 38. Fullerton, G.S., & Cattell, J. (1892). On the perception
Franks, I.M., Nagelkerke, P., Ketelaars, M., & van of small differences. University of Pennsylvania
Donkelaar, P. (1998). Response preparation and Philosophical Series, No. 2.
control of movement sequences. Canadian Journal of Gabbard, C.P. (2008). Lifelong motor development (5th
Experimental Psychology, 52, 93-102. ed.). San Francisco: Benjamin/Cummings.
Franks, I.M., & Stanley, M.L. (1991). Learning the Gable, C.D., Shea, C.H., & Wright, D.L. (1991). Summary
invariants of a perceptual motor skill. Canadian knowledge of results. Research Quarterly for Exercise
Journal of Psychology, 45, 303-320. and Sport, 62, 285-292.
Franks, I.M., & Wilberg, R.B. (1984). Consistent Gabriele, T.E., Hall, C.R., & Lee, T.D. (1989). Cognition
reproduction of movement sequences during in motor learning: Imagery effects on contextual
acquisition of a pursuit tracking task. Perceptual and interference. Human Movement Science, 8, 227-245.
Motor Skills, 58, 699-709. Gagnon, C., Mathieu, J., & Desrosiers, J. (2004).
Franz, E.A. (1997). Spatial coupling in the coordination of Standardized finger-nose test validity for coordination
complex actions. Quarterly Journal of Experimental assessment in an ataxic disorder. Canadian Journal of
Psychology, SOA, 684-704. Neurological Sciences, 31, 484-489.
Franz, E.A., Eliassen, J.C., Ivry, R.B., & Gazzaniga, M.S. Gallagher, J.D., & Thomas, J.R. (1980). Effects of
(1996). Dissociation of spatial and temporal coupling varying post-KR intervals upon children's motor
in the bimanual movements of callosotomy patients. performance. Journal of Motor Behavior, 12, 41-46.
Psychological Science, 7, 306-310. Gallistel, C.R. (1980). The organization of action: A new
Franz, E.A., Zelaznik, H.N., & McCabe, G. (1991). synthesis. Hillsdale, NJ: Erlbaum.
Spatial topological constraints in a bimanual task. Gandevia, S.C., & Burke, D. (1992). Does the nervous
Acta Psychologica, 77, 137-151. system depend on kinesthetic information to control
Franz, E.A., Zelaznik, H.N., & Smith, A. (1992). Evidence natural limb movements? Behavioral and Brain
of common timing processes in the control of manual, Sciences, 15, 614-632.
orofacial, and speech movements. Journal of Motor Gandevia, S.C., Proske, U., & Stuart, D.G. (Eds.). (2002).
Behavior, 24, 281-287. Sensorimotor control of movement and posture. New
Franz, E.A., Zelaznik, H.N., Swinnen, S.P., & Walter, York: Kluwer Academic/Plenum Press.
C.B. (2001). Spatial conceptual influences on the Canel, T., Tanzer, M., & Goodale, M.A. (2008). A double
coordination of bimanual actions: When a dual task dissociation between action and perception in the
becomes a single task. Journal of Motor Behavior, 33, context of visual illusions: Opposite effects of real
103-112. and illusory size. Psychological Science, 19, 221-225.
Franz, V.H., Bulthoff, H.H., & Fahle, M. (2003). Grasp Gao, L., & Zelaznik, H.N. (1991). The modification of an
effects of the Ebbinghaus illusion: Obstacle avoidance already programmed response: A new interpretation
is not the explanation. Experimental Brain Research, of Henry and Harrison (1961). Journal of Motor
149, 470-477. Behavior, 23, 221-223.
French, K.E., & Thomas, J.R. (1987). The relation of Garvey, W.D. (1960). A comparison of the effects of
knowledge development to children's basketball training and secondary tasks on tracking behavior.
performance. Journal of Sport Psychology, 9, 15-32. Journal of Applied Psychology, 44, 370-375.
520 I!! References

Gauggel, S., & Fischer, S. (2001). The effect of goal Gilbreth, F.B. (1909). Bricklaying system. New York:
setting on motor performance and motor learning Myron C. Clark.
in brain-damaged patients. Neuropsychological Giuffrida, C.G., Shea, J.B., & Fairbrother, J.T. (2002).
Rehabilitation, 11, 33-44. Differential transfer benefits of increased practice
Gawron, V.J., Drury, C.G., Czaja, S.J., & Wilkins, D.M. for constant, blocked, and serial practice schedules.
(1989). A taxonomy of independent variables Journal of Motor Behavior, 34, 353-365.
affecting human performance. International Journal Gladwell, M. (2002). The tipping point: How little things
of Man-Machine Studies, 31, 643-672. can make a big difference. New York: Little, Brown.
Gelfan, S., & Carter, S. (1967). Muscle sense in man.
Gladwell, M. (2008). Outliers: The story of success. New
Experimental Neurology, 18, 469-473.
York: Little, Brown.
Gelfand, l.M., Gurfinkel, V.S., Tomin, S.V., & Tsetlin,
Glencross, D.J. (1973). Temporal organization in a
M.L. (1971). Models of the structural-functional
repetitive speed skill. Ergonomics, 16, 765-776.
organization of certain biological systems.
Cambridge, MA: MIT Press. Glencross, D., & Barrett, N. (1992). The processing of
visual feedback in rapid movements: Revisited. In
Gentile,A.M. (1972). A working model of skill acquisition
J.J. Summers (Ed.), Approaches to the study of motor
with application to teaching. Quest, 17, 3-23.
control and learning (pp. 289-311). Amsterdam:
Gentile, A.M. (1998). Implicit and explicit processes Elsevier.
during acquisition of functional skills. Scandinavian
Glover, S. (2002). Visual illusions affect planning but not
Journal of Occupational Therapy, 5, 7-16.
control. Trends in Cognitive Sciences, 6, 288-292.
Gentile,A.M. (2000). Skill acquisition: Action, movement,
and neuromotor processes. In J.H. Carr & R.H. Glover, S. (2004). Separate visual representations in the
Shepherd (Eds.), Movement science: Foundation for planning and control of action. Behavioral and Brain
physical therapy in rehabilitation (2nd ed.) (pp. 111- Sciences, 27, 3-78.
180). Gaithersburg, MD: Aspen. Glover, S.R., & Dixon, P. (2001). Dynamic illusion
Gentilucci, M., Chieffi, S., Scarpa, M., & Castiello, U. effects in a reaching task: Evidence for separate
(1992). Temporal coupling between transport and visual representations in the planning and control
grasp components during prehension movements: of reaching. Journal of Experimental Psychology:
Effects of visual perturbation. Behavioural Brain Human Perception and Performance, 27, 560-572.
Research, 47, 71-82. Goettl, B.P. (1996). The spacing effect in aircraft
Gentner, D.R. (1987). Timing of skilled motor recognition. Human Factors, 38, 34-49.
performance: Tests of the proportional duration Goggin, N.L., & Meeuwsen, H.J. (1992). Age-related
model. Psychological Review, 94, 255-276. differences in the control of spatial aiming
Ghez, C., & Krakauer, J. (2000). The organization of movements. Research Quarterly for Exercise and
movement. In E.R. Kandel, J.H. Schwartz, & T.M. Sport, 63, 366-372.
Jessell (Eds.), Principles of neural science (pp. 653- Goginski, A.M., & Collins, D. (1996). Research design
673). New York: McGraw-Hill. and mental practice. Journal of Sports Sciences, 14,
Ghez, C., & Thach, W.T. (2000). The cerebellum. In 381-392.
E.R. Kandel, J.H. Schwartz, & T.M. Jessell (Eds.), Goodale, M.A., & Humphrey, G.K. (2001). Separate
Principles of neural science (pp. 832-852). New York: visual pathways for action and perception. In E.B.
McGraw-Hill. Goldstein (Ed.), Blackwell handbook of perception
Gibson, J.J. (1958). Visually controlled locomotion and (pp. 312-343). Malden, MA: Blackwell.
visual orientation in animals. British Journal of Goodale, M.A., & Milner, A.D. (1992). Separate visual
Psychology, 49, 182-194. pathways for perception and action. Trends in
Gibson, J.J. (1966). The senses considered as perceptual Neurosciences, 15, 20-25.
systems. Boston: Houghton Mifflin. Goodale, M.A., Milner, A.D., Jakobson, L.S., & Carey,
Gibson, J.J. (1979). The ecological approach to visual D.P. (1991). A neurological dissociation between
perception. Boston: Houghton Mifflin. perceiving objects and grasping them. Nature, 349,
Gielen, C.C.A.M., Schmidt, R.A., & van den Heuvel, 154-156.
P.J.M. (1983). On the nature of intersensory facilitation Goodale, M.A., & Servos, P. (1996). Visual control of
of reaction time. Perception & Psychophysics, 34, prehension. In H.N. Zelaznik (Ed.), Advances in
161-168. motor control and learning (pp. 87-121). Champaign,
Gielen, C.C.A.M., van den Oosten, K., & ter Gunne, IL: Human Kinetics.
F.P. (1985). Relation between EMG activation and Goode, S., & Magill, R.A. (1986). Contextual interference
kinematic properties of aimed arm movements. effects in learning three badminton serves. Research
Journal of Motor Behavior, 17, 421-442. Quarterly for Exercise and Sport, 57, 308-314.
 References 521

Goodman, D., & Kelso, J.A.S. (1980). Are movements observation and dialogue. Research Quarterly for
prepared in parts? Not under compatible Exercise and Sport, 78, 197-203.
(naturalized) conditions. Journal of Experimental Granit, R. (1970). The basis of motor control. New York:
Psychology: General, 109, 475-495. Academic Press.
Goodman, D., Kobayashi, R.B., & Kelso, J.A.S. (1983). Gray, J.T., Neisser, U., Shapiro, B.A., & Kouns, S. (1991).
Maintenance of symmetry as a constraint in motor Observational learning of ballet sequences: The role
control. Canadian Journal of Applied Sport Science, of kinematic information. Ecological Psychology, 3,
8, 238. 121-134.
Goodwin, G.M., McCloskey, D.I., & Matthews, P.B.C. Gray,R. (2002). "Markov at the Bat": Amodelofcognitive
(1972). The contribution of muscle afferents to processing in baseball batters. Psychological Science,
kinaesthesia shown by vibration induced illusions 13, 543-548.
of movement and by the effects of paralyzing joint
Gray, R. (2004). Attending to the execution of a complex
afferents. Brain, 95, 705-748.
sensorimotor skill: Expertise differences, choking,
Goodwin, J.E., & Meeuwsen, H.J. (1995). Using and slumps. Journal of Experimental Psychology:
bandwidth knowledge of results to alter relative Applied, 10, 42-54.
frequencies during motor skill acquisition. Research
Gray, R. (2009a). A model of motor inhibition for
Quarterly for Exercise and Sport, 66, 99-104.
a complex skill: Baseball batting. Journal of
Gordon, A.M., & Soechting, J.F. (1995). Use of tactile Experimental Psychology: Applied, 15, 91-105.
afferent information in sequential finger movements.
Gray, R. (2009b). How do batters use visual, auditory,
Experimental Brain Research, 107, 281-292.
and tactile information about the success of a baseball
Gordon, J., & Chez, C. (1991). Muscle receptors and swing? Research Quarterly for Exercise and Sport, 80,
spinal reflexes: The stretch reflex. In E.R. Kandel, J.H. 491-501.
Schwartz, & T.M. Jessell (Eds.), Principles of neural
Gray, S., Watts, S., Debicki, D., & Hore, J. (2006).
science (3rd ed.) (pp. 564-580). Amsterdam: Elsevier.
Comparison of kinematics in skilled and unskilled
Gordon, J., Ghilardi, M.F., & Chez, C. (1995). arms of the same recreational baseball players.
Impairments of reaching movements in patients Journal of Sports Sciences, 24, 1183-1194.
without proprioception. I. Spatial errors. Journal of
Green, C.S., & Bavelier, D. (2003). Action video game
Neurophysiology, 73, 347-360.
modifies visual selective attention. Nature, 423, 534-
Gottlieb, G.L. (1998). Rejecting the equilibrium-point 537.
hypothesis. Motor Control, 2, 10-12.
Green, C.S., & Bavelier, D. (2006). Effect of action video
Gottlieb, G.L. (2000). A test of torque-control and games on the spatial distribution of visuospatial
equilibrium-point models of motor control. Human attention. Journal of Experimental Psychology:
Movement Science, 19, 925-931. Human Perception and Performance, 32, 1465-1478.
Gottlieb, G.L., Coreas, D.M., & Agarwal, G.C. (1989). Green, D.M., & Swets, J.A. (1966). Signal detection
Strategies for the control of voluntary movements theory and psychophysics. New York: Wiley.
with one mechanical degree of freedom. Behavioral Green, D.P., Whitehead, J., & Sugden, D.A. (1995).
and Brain Sciences, 12, 189-250. Practice variability and transfer of a racket skill.
Gottsdanker, R. (1970). Uncertainty, timekeeping, and Perceptual and Motor Skills, 81, 1275-1281.
simple reaction time. Journal of Motor Behavior, 2, Green, M. (2000). "How long does it take to stop?"
245-260. Methodological analysis of driver perception-brake
Gottsdanker, R. (1973). Psychological refractoriness times. Transportation Human Factors, 2, 195-216.
and the organization of step-tracking responses. Greene, LS., & Williams, H.G. (1993). Age related
Perception & Psychophysics, 14, 60-70. differences in timing control of repetitive movements:
Gottsdanker, R., & Stelmach, G.E. (1971). The persistence Application of the Wing-Kristofferson model.
of psychological refractoriness. Journal of Motor Research Quarterly for Exercise and Sport, 64, 32-38.
Behavior, 3, 301-312. Greene, P.H. (1972). Problems of organization of motor
Gould, D., & Krane, V. (1992). The arousal-athletic systems. In R. Rosen & F.M. Snell (Eds.), Progress
performance relationship: Current status and future in theoretical biology (Vol. 2). New York: Academic
directions. In T.S. Horn (Ed.), Advances in sport psy- Press.
chology (pp. 119-141). Champaign, IL: Human Kinetics. Greenwald, A.G., & Schulman, H.G. (1973). On doing
Goulding, M. (2009). Circuits controlling vertebrate two things at once: Elimination of the psychological
locomotion: Moving in a new direction. Nature refractory period effect. Journal of Experimental
Reviews: Neuroscience, 10, 507-518. Psychology, 101, 70-76.
Granados, C., & Wulf, G. (2007). Enhancing motor Griffith, C.R. (1931). An experiment on learning to drive
learning through dyad practice: Contributions of a golf ball. Athletic Journal, 11, 11-13.
522 Iii! References

Grillner, S. (1972). The role of muscle stiffness in meeting Haffenden, A.M., Schiff, K.C., & Goodale, M.A. (2001).
the changing postural and locomotor requirements The dissociation between perception and action
for force development by the ankle extensors. Acta in the Ebbinghaus illusion: Nonillusory effects of
Physiologica Scandinavica, 86, 92-108. pictorial cues on grasp. Current Biology, 11, 177-181.
Grillner, S. (1975). Locomotion in vertebrates: Central Hagemann, N., Strauss, B., & Cafial-Bruland, R. (2006).
mechanisms and reflex interaction. Physiological Training perceptual skill by orienting visual attention.
Reviews, 55, 247-304. Journal of Sport and Exercise Psychology, 28, 143-158.
Grillner, S. (2007). Biological pattern generation: The Haggard, P. (2008). Human volition: Towards a
cellular and computational logic of networks in neuroscience of will. Nature Reviews: Neuroscience,
motion. Neuron, 52, 751-766. 9, 934-946.
Grillner, S., & Wallen, P. (1985). Central pattern Haggard, P., Clark, S., & Kalogeras, J. (2002).
generators for locomotion, with special reference to Voluntary action and conscious awareness. Nature
vertebrates. Annual Review of Neuroscience, 8, 233- Neuroscience, 5, 382-385.
261. Haggard, P., & Wing, A. (1995). Coordinated responses
Grillner, S., & Wallen, P. (2002). Cellular bases following mechanical perturbation of the arm during
of a vertebrate locomotion system-steering, prehension. Experimental Brain Research, 102, 483-
intersegmental and segmental co-ordination and 494.
sensory control. Brain Research Reviews, 40, 92-106. Hagman, J.D. (1983). Presentation- and test-trial effects
Grosjean, M., Shiffrar, M., & Knoblich, G. (2007). Fitts's on acquisition and retention of distance and location.
law holds for action perception. Psychological Journal of Experimental Psychology: Learning,
Science, 18, 95-99. Memory, and Cognition, 9, 334-345.
Guadagnoli, M.A., Dornier, L.A., & Tandy, R.D. (1996). Hah, S., &Jagacinski, R.J. (1994). The relative dominance
Optimal length for summary knowledge of results: of schemata in a manual tracking task: Input patterns,
The influence of task-related experience and system dynamics, and movement patterns. Journal of
complexity. Research Quarterly for Exercise and Motor Behavior, 26, 204-214.
Sport, 67, 239-248. Haigney, D., & Westerman, J. (2001). Mobile (cellular)
Guadagnoli, M.A., & Kohl, R.M. (2001). Knowledge phone use and driving: A critical review of research
of results for motor learning: Relationship between methodology. Ergonomics, 44, 132-143.
error estimation and knowledge of results frequency. Haken, H., Kelso, J.A.S., & Bunz, H. (1985). A theoretical
Journal of Motor Behavior, 33, 217-224. model of phase transitions in human hand
Guadagnoli, M.A., & Lee, T.D. (2004). Challenge point: A movements. Biological Cybernetics, 51, 347-356.
framework for conceptualizing the effects of various Hall, K.G., Domingues, D.A., & Cavazos, R. (1994).
practice conditions in motor learning. Journal of Contextual interference effects with skilled baseball
Motor Behavior, 36, 212-224. players. Perceptual and Motor Skills, 78, 835-841.
Guay, M., Salmoni, A., & Lajoie, Y. (1999). The effects Hall, K.G., & Magill, R.A. (1995). Variability of practice
of different knowledge of results spacing and and contextual interference in motor skill learning.
summarizing techniques on the acquisition of a Journal of Motor Behavior, 27, 299-309.
ballistic movement. Research Quarterly for Exercise
Hallett, M. (2007). Transcranial magnetic stimulation: A
and Sport, 70, 24-32.
primer. Neuron, 55, 187-199.
Guay, M., Salmoni, A., & Mcilwain, J. (1992). Summary
knowledge of results for skill acquisition: Beyond Halverson, L.E., Roberton, M.A., & Langendorfer,
S. (1982). Development of the overarm throw:
Lavery and Schmidt. Human Movement Science, 11,
653-673. Movement and ball velocity changes by seventh
grade. Research Quarterly for Exercise and Sport, 53,
Guiard, Y. (1993). On Fitts's and Hooke's laws: Simple 198-205.
harmonic movement in upper-limb cyclical aiming.
Hammerton, M. (1989). Tracking. In D.H. Holding (Ed.),
Acta Psychologica, 82, 139-159.
Human skills (2nd ed.) (pp. 171-195). Chichester:
Guiard, Y. (1997). Fitts' law in the discrete vs. cyclical Wiley.
paradigm. Human Movement Science, 16, 97-131.
Hancock, G.R., Butler, M.S., & Fischman, M.G. (1995).
Gullicksen, H. (1950). Theory of mental tests. New York: On the problem of two-dimensional error scores:
Wiley. Measures and analyses of accuracy, bias, and
Guthrie, E.R. (1952). The psychology of learning. New consistency. Journal of Motor Behavior, 27, 241-250.
York: Harper & Row. Hancock, P.A., Lesch, M., & Simmons, L. (2003). The
Haffenden, A.M., & Goodale, M.A. (1998). The effect distraction of phone use during a critical driving
of pictorial illusion on prehension and perception. maneuver. Accident Analysis and Prevention, 35,
Journal of Cognitive Neuroscience, 10, 122-136. 501-514.
 References 523

Hasan, Z., & Enoka, R.M. (1985). Isometric torque-angle Helsen, W., Starkes, J.L., & Buekers, M.J. (1997). Effects of
relationship and movement-related activity of human target eccentricity on temporal costs of point of gaze
elbow flexors: Implications for the equilibrium-point and the hand in aiming. Motor Control, 1, 161-177.
hypothesis. Experimental Brain Research, 59, 441- Helsen, W.F., Starkes, J.L., & van Winckel, J. (2000).
450. Effects of a change in selection year on success of
Hasher, L., & Zacks, R.T. (1988). Working memory, male soccer players. American Journal of Human
comprehension, and aging: A review and a new view. Biology, 12, 729-735.
In C.H. Bower (Ed.), The psychology of learning Hendrickson, G., & Schroeder, W.H. (1941). Transfer of
and motivation (Vol. 22) (pp. 193-225). New York: training in learning to hit a submerged target. Journal
Academic Press. of Educational Psychology, 32, 205-213.
Hatze, H. (1976). Biomechanical aspects of a successful Henry, F.M. (1953). Dynamic kinesthetic perception and
motion optimization. In P. V. Komi (Ed.), Biomechanics adjustment. Research Quarterly, 24, 176-187.
V-B (pp. 5-12). Baltimore: University Park Press.
Henry, F.M. (1959). Reliability, measurement error, and
Hay, J.F., & Jacoby, L.L. (1996). Separating habit and intraindividual difference. Research Quarterly, 30,
recollection: Memory slips, process dissociations, 21-24.
and probability matching. Journal of Experimental
Henry, F.M. (1961). Reaction time-movement time
Psychology: Learning, Memory, and Cognition, 22,
correlations. Perceptual and Motor Skills, 12, 63-66.
1323-1335.
Henry, F.M. (1968). Specificity vs. generality in learning
Hay, L. (1981). The effect of amplitude and accuracy motor skill. In R.C. Brown & G.S. Kenyon (Eds.),
requirements on movement time in children. Journal Classical studies on physical activity (pp. 331-340).
of Motor Behavior, 13, 177-186. Englewood Cliffs, NJ: Prentice Hall. (Original work
Hayes, S.J., Ashford, D., & Bennett, S.J. (2008). Goal- published in 1958)
directed imitation: The means to an end. Acta Henry, F.M. (1975). Absolute error vs "E" in target
Psychologica, 127, 407-415. accuracy. Journal of Motor Behavior, 7, 227-228.
Haynes, J.D., & Rees, G. (2006). Decoding mental states Henry, F.M. (1980). Use of simple reaction time in motor
from brain activity in humans. Nature Reviews: programming studies: A reply to Klapp, Wyatt, and
Neuroscience, 7, 523-534. Lingo. Journal of Motor Behavior, 12, 163-168.
Haywood, K.M., & Getchell, N. (2009). Life span motor Henry, F.M., & Harrison, J.S. (1961). Refractoriness of a
development (5th ed.). Champaign, IL: Human fast movement. Perceptual and Motor Skills, 13, 351-
Kinetics. 354.
Head, H. (1926). Aphasia and kindred disorders of Henry, F.M., & Rogers, D.E. (1960). Increased response
speech. Cambridge: Cambridge University Press. latency for complicated movements and a "memory
Healy, A.F., Wohldmann, E.L., Sutton, E.M., & drum" theory of neuromotor reaction. Research
Bourne, L.E. Jr. (2006). Specificity effects in training Quarterly, 31, 448-458.
and transfer of speeded responses. Journal of Hepp-Reymond, M.-C., Chakarov, V., Schulte-
Experimental Psychology: Learning, Memory, and Mi:inting, J., Huethe, F., & Kristeva, R. (2009). Role
Cognition, 32, 534-546. of proprioception and vision in handwriting. Brain
Heath, M., & Elliott, D. (1999). Cerebral specialization for Research Bulletin, 79, 365-370.
speech production in persons with Down syndrome. Herrick, C.J. (1924). Origin and evolution of the
Brain and Language, 69, 193-211. cerebellum. Archives of Neurology and Psychiatry,
Heathcote, A., Brown, S., & Mewhort, D.J.K. (2000). The 11, 621-652.
power law repealed: The case for an exponential law Herslund, M.B., & Jmgensen, N.O. (2003). Looked-but-
of practice. Psychonomic Bulletin & Review, 7, 185- failed-to-see-errors in traffic. Accident Analysis and
207. Prevention, 35, 885-891.
Hellebrandt, F.A., Houtz, S.J., Partridge, M.J., & Walters, Heuer, H. (1985). Wie wirkt mentale Obung [How does
C.E. (1956). Tonic reflexes in exercises of stress in man. mental practice operate]? Psychologische Rundschau,
American Journal of Physical Medicine, 35, 144-159. 36, 191-200.
Hellyer, S. (1963). Stimulus-response coding and amount Heuer, H. (1988). Testing the invariance of relative
of information as determinants of reaction time. timing: Comment on Gentner (1987). Psychological
Journal of Experimental Psychology, 65, 521-522. Review, 95, 552-557.
Helmuth, L.L., & Ivry, R.B. (1996). When two hands are Heuer, H. (1991). Invariant relative timing in motor-
better than one: Reduced timing variability during program theory. In J. Fagard & P.H. Wolff (Eds.),
bimanual movement. Journal of Experimental The development of timing control and temporal
Psychology: Human Perception and Performance, organization in coordinated action (pp. 37-68).
22, 278-293. Amsterdam: Elsevier.
524 111 References

Heuer, H. (1996). Coordination. In H. Heuer & S.W. Keele practice behaviours. Human Movement Science, 21,
(Eds.), Handbook of perception and action. Vol. 2: 231-258.
Motor skills (pp. 121-180). San Diego: Academic Press. Hodges, N.J., & Franks, l.M. (2002b ). Modelling coaching
Heuer, H., & Schmidt, RA (1988). Transfer of learning practice: The role of instruction and demonstration.
among motor patterns with different relative timing. Journal of Sports Sciences, 20, 793-811.
Journal of Experimental Psychology: Human Hodges, N.J., Hayes, S., Horn, R.R., & Williams, AM.
Perception and Performance, 14, 241-252. (2005). Changes in coordination, control and outcome
Heuer, H., Schmidt, RA, & Ghodsian, D. (1995). as a result of extended practice on a novel motor skill.
Generalized motor programs for rapid bimanual Ergonomics, 48, 1672-1685.
tasks: A two-level multiplicative-rate model. Hodges, N.J., & Lee, T.D. (1999). The role of augmented
Biological Cybernetics, 73, 343-356. information prior to learning a bimanual visual-
Heyes, C.M., & Foster, C.L. (2002). Motor learning by motor coordination task: Do instructions of the
observation: Evidence from a serial reaction time movement pattern facilitate learning relative to
task. Quarterly Journal of Experimental Psychology, discovery learning? British Journal of Psychology, 90,
55A, 593-607. 389-403.
Hick, W.E. (1952). On th.e rate of gain of information. Hoffmann, E.R., & Sheikh, l.H. (1991). Finger width
Quarterly Journal of Experimental Psychology, 4, 11- corrections in Fitts' Law: Implications for speed-
26. accuracy research. Journal of Motor Behavior, 23,
Hiew, C.C. (1977). Sequence effects in rule learning 259-262.
and conceptual generalization. American Journal of Hoffmann, E.R., Tsang, K.K., & Mu, A (1995). Data-
Psychology, 90, 207-218. entry keyboard geometry and keying movement
Higuchi, T., Takada, H., Matsuura, Y., & Imanaka, K. times. Ergonomics, 38, 940-950.
(2004). Visual estimation of spatial requirements for Hogan, J.C., & Yanowitz, B.A. (1978). The role of
locomotion in novice wheelchair users. Journal of verbal estimates of movement error in ballistic skill
Experimental Psychology: Applied, 10, 55-66. acquisition. Journal of Motor Behavior, 10, 133-138.
Hikosaka, 0., Rand, M.K., Nakamura, K., Miyachi, S., Holden, M.K., Dettwiler, A, Dyar, T., Niemann, G., &
Kitaguchi, K., Sakai, K., Lu, X., & Shimo, Y. (2002). Bizzi, E. (2001 ). Retraining movement in patients with
Long-term retention of motor skill in macaque acquired brain injury using a virtual environment.
monkeys and humans. Experimental Brain Research, Studies in Health Technology and Informatics, 81,
147, 494-504. 192-198.
Hill, L.B. (1934). A quarter century of delayed recall. Holding, D.H. (1965). Principles of training. Oxford:
Pedagogical Seminary and Journal of Genetic Pergamon.
Psychology, 44, 231-238. Holding, D.H. (1970). Learning without errors. In L.E.
Hill, L.B. (1957). A second quarter century of delayed Smith (Ed.), Psychology of motor learning (pp. 59-
recall or relearning at 80. Journal of Educational 74). Chicago: Athletic Institute.
Psychology, 48, 65-68. Holding, D.H. (1976). An approximate transfer surface.
Hill, L.B., Rejall, A.E., & Thorndike, E.L. (1913). Practice Journal of Motor Behavior, 8, 1-9.
in the case of typewriting. Pedagogical Seminary, 20, Holding, D.H., & Macrae, AW. (1964). Guidance,
516-529. restriction, and knowledge of results. Ergonomics, 7,
Hills, B.L. (1980). Vision, visibility and perception in 289-295.
driving. Perception, 9, 183-216. Hollands, M.A., Marple-Horvat, D.E., Henkes, S., &
Hird, J:s., Landers, D.M., Thomas, J.R., & Horan, J.J. Rowan, AK. (1995). Human eye movements during
(1991). Physical practice is superior to mental practice visually guided stepping. Journal of Motor Behavior,
in enhancing cognitive and motor task performance. 27, 155-163.
Jounrnl of Sport and Exercise Psychology, 13, 281-293. Hollerbach, J.M. (1978). A study of human motor control
Hodges, N.J., & Franks, I.M. (2000). Attention focusing through analysis and synthesis of handwriting.
instructions and coordination bias: Implications for Unpublished doctoral dissertation, Massachusetts
learning a novel bimanual task. Human Movement Institute of Technology, Cambridge.
Science, 19, 843-867. Hollerbach, J.M. (1981). An oscillation theory of
Hodges, N.J., & Franks, I.M. (2001). Learning a handwriting. Biological Cybernetics, 39, 139-156.
coordination skill: Interactive effects of instruction Hollerbach, J.M., & Flash, T. (1982). Dynamic
and feedback. Research Quarterly for Exercise and interactions between limb segments during planar
Sport, 72, 132-142. arm movement. Biological Cybernetics, 44, 67-77.
Hodges, N.J., & Franks, l.M. (2002a). Learning as a Hollingworth, H.L. (1909). The inaccuracy of movement.
function of coordination bias: Building upon pre- Archives of Psychology, 13, 1-87.
 References 525

Holmes, G. (1939). The cerebellum of man. Brain, 62, Huxley, H.E. (1969). Structural organization and the
1-30. contraction mechanism in striated muscle. In S.
Hommel, B., & Prinz, W. (Eds.). (1997). Theoretical issues Devons (Ed.), Biology and the physical sciences (pp.
in stimulus-response compatibility. Amsterdam: 114-138). New York: Columbia University Press.
Elsevier. Hyman, I.E. Jr., Boss, S.M., Wise, B.M., McKenzie, K.E.,
Horak, M. (1992). The utility of connectionism for motor & Caggiano, J.M. (2010). Did you see the unicycling
learning: A reinterpretation of contextual interference clown? Inattentional blindness while walking
in movement schemas. Journal of Motor Behavior, 24, and talking on a cell phone. Applied Cognitive
58-66. Psychology, 24, 597-607.
Horrey, W.J., & Wickens, C.D. (2006). The impact of cell Hyman, R. (1953). Stimulus information as a determinant
phone conversations on driving using meta-analytic of reaction time. Journal of Experimental Psychology,
techniques. Human Factors, 48, 196-205. 45, 188-196.
Houk, J.C. (1979). Regulation of stiffness by skeletomotor Imanaka, K., &Abernethy, B. (1991). The mediating effect
reflexes. Annual Review of Physiology, 41, 99-114. of learning on the interference between location and
distance recall from motor short-term memory. Acta
Houk, J.C., & Henneman, E. (1967). Responses of Golgi
Psychologica, 77, 153-165.
tendon organs to active contractions of the soleus
muscle of the cat. Journal of Neurophysiology, 30, Imanaka, K., & Abernethy, B. (1992). Interference
466-481. between location and distance information in motor
Houk, J.C., & Rymer, W.Z. (1981). Neural control of short-term memory: The respective roles of direct
muscle length and tension. In V.B. Brooks (Ed.), kinesthetic signals and abstract codes. Journal of
Handbook of physiology: Section 1: The nervous Motor Behavior, 24, 274-280.
system. Vol. 2. Motor control (pp. 257-323). Bethesda, Imanaka, K., Abernethy, B., & Quek, J.-J. (1998). The
MD: American Physiological Society. locus of distance-location interference in movement
Howard, L.A., & Tipper, S.P. (1997). Hand deviations reproduction: Do we know any more 25 years on? In
away from visual cues: Indirect evidence for J.P. Piek (Ed.), Motor behavior and human skill (pp.
inhibition. Experimental Brain Research, 113, 144- 29-55). Champaign, IL: Human Kinetics.
152. Immink, M.A., & Wright, D.L. (1998). Contextual
Howell, M.L. (1953). Influence of emotional tension on interference: A response planning account. Quarterly
speed of reaction and movement. Research Quarterly, Journal of Experimental Psychology, 51A, 735-754.
24, 22-32. Immink,M.A., & Wright, D.L. (2001). Motor programming
Howell, M.L. (1956). Use of force-time graphs for during practice conditions high and low in contextual
performance analysis in facilitating motor learning. interference. Journal of Experimental Psychology:
Research Quarterly, 27, 12-22. Human Perception and Performance, 27, 423-437.
Hoyt, D.F., & Taylor, C.R. (1981). Gait and the energetics Irion, AL. (1948). The relation of "set" to retention.
of locomotion in horses. Science, 292, 239-240. Psychological Review, 55, 336-341.
Huang, VS., Shadmehr, R., & Diedrichsen, J. (2008). Irion, AL. (1966). A brief history of research on the
Active learning: Learning a motor skill without a acquisition of skill. In E.A. Bilodeau (Ed.), Acquisition
coach. Journal of Neurophysiology, 100, 879-887. of skill (pp. 1-46). New York: Academic Press.
Hubbard, AW., & Seng, C.N. (1954). Visual movements Irwin, M., Fitzgerald, C., & Berg, W.P. (2000). Effect of
of batters. Research Quarterly, 25, 42-57. the intensity of wireless telephone conversations on
Huet, M., Camachon, C., Fernandez, L., Jacobs, D.M., reaction time in a braking response. Perceptual and
Motor Skills, 90, 1130-1134.
& Montagne, G. (2009). Self-controlled concurrent
feedback and the education of attention towards Ishigami, Y., & Klein, R.M. (2009). Is a hands-free phone
perceptual invariants. Human Movement Science, safer than a handheld phone? Journal of Safety
28, 450-467. Research, 40, 157-164.
Hughes, C.M.L., & Franz, E.A. (2008). Goal-related Ivry, R., & Corcos, D.M. (1993). Slicing the variability
planning constraints in bimanual grasping and pie: Component analysis of coordination and
placing of objects. Experimental Brain Research, 188, motor dysfunction. In K.M. Newell & D.M. Corcos
541-550. (Eds.), Variability and motor control (pp. 415-447).
Hull, CL. (1943). Principles of behavior. New York: Champaign, IL: Human Kinetics.
Appleton-Century-Crofts. Ivry, R.B., & Keele, S.W. (1989). Timing functions of the
Hurley, S.R., & Lee, T.D. (2006). The influence of cerebellum. Journal of Cognitive Neuroscience, 1,
augmented feedback and prior learning on the 136-152.
acquisition of a new bimanual coordination pattern. Ivry, R.B., Keele, S.W., & Diener, H.C. (1988). Dissociation
Human Movement Science, 25, 339-348. of the lateral and medial cerebellum in movement
526 mi References

timing and movement execution. Experimental Brain Jamieson, B.A., & Rogers, W.A. (2000). Age-related
Research, 73, 167-180. effects of blocked and random practice schedules on
Ivry, R.B., Richardson, T.C., & Helmuth, L.L. (2002). learning a new technology. Journal of Gerontology:
Improved temporal stability in multi-effector Psychological Sciences, 55B, P343-P353.
movements. Journal of Experimental Psychology: Janelle, C.M. (1999). Ironic mental processes in
Human Perception and Performance, 28, 72-92. sport: Implications for sport psychologists. Sport
Jabusch, H.-C., Alpers, H., Kopiez, R., Vauth, H., & Psychologist, 13, 210-220.
Altenmilller, E. (2009). The influence of practice Janelle, C.M., Barba, D.A., Frehlich, S.G., Tennant, L.K.,
on the development of motor skills in pianists: A & Cauraugh, J.H. (1997). Maximizing performance
longitudinal study in a selected motor task. Human feedback effectiveness through videotape replay and
Movement Science, 28, 74-84. a self-controlled learning environment. Research
Jackson, G.M., German, K., & Peacock, K. (2002). Quarterly for Exercise and Sport, 68, 269-279.
Functional coupling between the limbs during Janelle, C.M., Champenoy, J.D., Coombes, S.A., &
bimanual reach-to-grasp movements. Human Mousseau, M.B. (2003). Mechanisms of attentional
Movement Science, 21, 317-333. cueing during observational learning to facilitate skill
Jackson, G.M., Jackson, S.R., & Kritikos, A. (1999). acquisition. Journal of Sports Sciences, 21, 825-838.
Attention for action: Coordinating bimanual reach- Janelle, C.M., Kim, J., & Singer, R.N. (1995). Subject-
to-grasp movements. British Journal of Psychology, controlled performance feedback and learning of a
90, 247-270. closed motor skill. Perceptual and Motor Skills, 81,
Jackson, R.C., & Farrow, D. (2005). Implicit perceptual 627-634.
training: How, when, and why? Human Movement Janelle, C.M., Singer, R.N., & Williams, A.M. (1999).
Science, 24, 308-325. External distraction and attentional narrowing:
Jacoby, L.L. (1978). On interpreting the effects of Visual search evidence. Journal of Sport and Exercise
repetition: Solving a problem versus remembering Psychology, 21, 70-91.
a solution. Journal of Verbal Learning and Verbal Janis, I., Defares, P., & Grossman, P. (1983). Hypervigilant
Behavior, 17, 649-667. reactions to threat. In H. Selye (Ed.), Selye's guide to
Jacoby, L.L., & Dallas, M. (1981). On the relationship stress research (pp. 1-43). New York: Van Nostrand
between autobiographical memory and perceptual Reinhold.
learning. Journal of Experimental Psychology: Janssen, L., Beuting, M., Meulenbroek, R., & Steenbergen,
General, 110, 306-340. B. (2009). Combined effects of planning and
Jacoby, L.L., Ste-Marie, D., & Toth, J.P. (1993). Redefining execution constraints on bimanual task performance.
automaticity: Unconscious influences, awareness, Experimental Brain Research, 192, 61-73.
and control. In A. Baddeley & L. Weiskrantz (Eds.), Jaric, S., & Latash, M.L. (2000). The equilibrium-point
Attention: Selection, awareness, and control: A hypothesis is still doing fine. Human Movement
tribute to Donald Broadbent (pp. 261-282). Oxford: Science, 19, 933-938.
Clarendon Press. Jarus, T., & Goverover, Y. (1999). Effects of contextual
Jagacinski, R.J., & Flach, J.M. (2003). Control theory interference and age on acquisition, retention, and
for humans: Quantitative approaches to modeling transfer of motor skill. Perceptual and Motor Skills,
performance. Mahwah, NJ: Erlbaum. 88, 437-447.
Jagacinski, R.J., & Hah, S. (1988). Progression-regression Jasiewicz, J., & Simmons, R.W. (1996). Response timing
effects in tracking repeated patterns. Journal of accuracy as a function of movement velocity and
Experimental Psychology: Human Perception and distance. Journal of Motor Behavior, 28, 224-232.
Performance, 14, 77-88. Jastrow, J. (1906). The subconscious. Boston: Houghton-
Jagacinski, R.J., & Monk, D.L. (1985). Fitts' law in two Mifflin.
dimensions with hand and head movements. Journal Jeannerod, M. (1981). Intersegmental coordination
of Motor Behavior, 17, 77-95. during reaching at natural visual objects. In J. Long
Jakobson, L.S., & Goodale, M.A. (1991). Factors affecting & A. Baddeley (Eds.), Attention and performance IX
higher-order movement planning: A kinematic (pp. 153-169). Hillsdale, NJ: Erlbaum.
analysis of human prehension. Experimental Brain Jeannerod, M. (1984). The timing of natural prehension
Research, 86, 199-208. movements. Journal of Motor Behavior, 16, 235-254.
James, W. (1890). The principles of psychology (Vol. 1). Jeannerod, M. (1996). Reaching and grasping: Parallel
New York: Holt. specification of .visuomotor channels. In H. Heuer
Jami, L. (1992). Golgi tendon organs in mammalian & S.W. Keele (Eds.), Handbook of perception and
skeletal muscle: Functional properties and central action. Vol. 2: Motor skills (pp. 405-460). San Diego:
actions. Physiological Reviews, 72, 623-666. Academic Press.
 References mi 527

Jeannerod, M. (1997). The cognitive neuroscience of Jones, L.L., & French, K.E. (2007). Effects of contextual
action. Oxford: Blackwell. interference on acquisition and retention of three
Jeannerod, M. (2001). Neural simulation of action: A volleyball skills. Perceptual and Motor Skills, 105,
unifying mechanism for motor cognition. Neurolmage, 883-890.
14, 103-109. Jones, M.B. (1962). Practice as a process of simplification.
Jeannerod, M., & Frak, V. (1999). Mental imaging of Psychological Review, 69, 274-294.
motor activity in humans. Current Opinion in Jones, M.B. (1966). Individual differences. In E.A.
Neurobiology, 9, 735-739. Bilodeau (Ed.), Acquisition of skill (pp. 109-146).
Jeannerod, M., & Jacob, P. (2005). Visual cognition: New York: Academic Press.
A new look at the two-visual systems model. Jones, S.K. (1993). A modular approach to individual
Neuropsychologia, 43, 301-312. differences in skill and coordination. In J.L. Starkes &
Jeannerod, M., & Marteniuk, R.G. (1992). Functional F. Allard (Eds.), Cognitive issues in motor expertise
characteristics of prehension: From data to artificial (pp. 273-293). Amsterdam: Elsevier.
neural networks. In L. Proteau & D. Elliott (Eds.), Jonides, J., Naveh-Benjamin, M., & Palmer, J. (1985).
Vision and motor control (pp. 197-232). Amsterdam: Assessing automaticity. Acta Psychologica, 60, 157-
Elsevier. 171.
Jeeves, M.A. (1961). Changes in performance at a serial Jordan, M.I. (1995). The organization of action sequences:
reaction task under conditions of advance and delay Evidence from a relearning task. Journal of Motor
of information. Ergonomics, 4, 329-338. Behavior, 27, 179-192.
Jeka, J.J., & Kelso, J.A.S. (1989). The dynamic pattern Judd, C.H. (1908). The relation of special training to
approach to coordinated behavior: A tutorial review. general intelligence. Educational Review, 36, 28-42.
In S.A. Wallace (Ed.), Perspectives on the coordination Kahneman, D. (1973). Attention and effort. Englewood
of movement (pp. 3-45). Amsterdam: Elsevier. Cliffs, NJ: Prentice Hall.
Jenison, R.L. (1997). On acoustic information for motion. Kahol, K., Satava, R.M., Ferrara, J., & Smith, M.L.
Ecological Psychology, 9, 131-151. (2009). Effect of short-term pretrial practice on
Jervis, C., Bennett, K., Thomas, J., Lim, S., & Castiello, U. surgical proficiency in simulated environments: A
(1999). Semantic category interference effects upon randomized trial of the "preoperative warm-up"
the reach-to-grasp movement. Neuropsychologia, 37, effect. Journal of the American College of Surgeons,
857-868. 208, 255-268.
Johansson, R.S., & Flanagan, J.R. (2009). Coding and use of Kandel, E.R., Schwartz, J.H., & Jessell, J.M. (Eds.). (2000).
tactile signals from the fingertips in object manipulation Principles of neural science. New York, McGraw-Hill.
tasks. Nature Reviews: Neuroscience, 10, 345-359. Kao, J.C., Ringenbach, S.D., & Martin, P.E. (2003). Gait
Johansson, R.S., & Westling, G. (1984). Roles of transitions are not dependent on changes in intralimb
glabrous skin receptors and sensorimotor memory coordination variability. Journal of Motor Behavior,
in automatic control of precision grip when lifting 35, 211-214.
rougher or more slippery objects. Experimental Brain Karlin, L., & Kestenbaum, R. (1968). Effects of number of
Research, 56, 550-564. alternatives on the psychological refractory period.
Johansson, R.S., & Westling, G. (1988). Programmed Quarterly Journal of Experimental Psychology, 20,
and triggered actions to rapid load changes during 167-178.
precision grip. Experimental Brain Research, 71, 72-86. Karwowski, W. (Ed.). (2001). International encyclopedia
Johansson, R.S., & Westling, G. (1990). Tactile afferent of ergonomics and human factors. London: Taylor &
signals in the control of precision grip. In M. Francis.
Jeannerod (Ed.), Attention and performance XIII (pp. Keele, S.W. (1968). Movement control in skilled motor
677-713). Hillsdale, NJ: Erlbaum. performance. Psychological Bulletin, 70, 387-403.
Johansson, R.S., & Westling, G. (1991). Afferent signals Keele, S.W. (1973). Attention and human performance.
during manipulative tasks in man. In 0. Franzen & Pacific Palisades, CA: Goodyear.
J. Westman (Eds.), Somatosensory mechanisms (pp. Keele, S.W. (1981). Behavioral analysis of movement.
25-48). London: Macmillan. In V. Brooks (Ed.), Handbook of physiology: Section
Johnson, K.A., Bennett, J.E., Georgiou, N., Bradshaw, 1: The nervous system. Vol. 2. Motor control, part 2
J.L., Chiu, E., Cunnington, R., & Iansek, R. (2000). (pp. 1391-1414). Baltimore: American Physiological
Bimanual co-ordination in Huntington's disease. Society.
Experimental Brain Research, 134, 483-489. Keele, S.W. (1986). Motor control. In K.R. Boff, L.
Jones, C.M., & Miles, T.R. (1978). Use of advance cues in Kaufman, & J.P. Thomas (Eds.), Handbook of
predicting the flight of a lawn tennis ball. Journal of perception and performance (pp. 30.1-30.60). New
Human Movement Studies, 4, 231-235. York: Wiley.
528 References

Keele, S.W. (1998). Programming or planning Kelly, S.W., Burton, A.M., Riedel, B., & Lynch, E. (2003).
conceptions of motor control speak to different Sequence learning by action and observation:
phenomena than dynamical systems conceptions. In Evidence for separate mechanisms. British Journal of
J.P. Piek (Ed.), Motor behavior and human skill (pp. Psychology, 94, 355-372.
403-405). Champaign, IL: Human Kinetics. Kelso, B.A. (1984). The effects of extended practice
Keele, S.W., Cohen, A., & Ivry, R. (1990). Motor programs: on aiming movements in terms of Fitts' Law.
Concepts and issues. In M. Jeannerod (Ed.), Attention Unpublished master's thesis, York University.
and performance XIII (pp. 77-110). Hillsdale, NJ: Kelso, J.A.S. (1977). Motor control mechanisms
Erlbaum. underlying human movement reproduction. Journal
Keele, S.W., & Ells, J.C. (1972). Memory characteristics of of Experimental Psychology: Human Perception and
kinesthetic information. Journal of Motor Behavior, Performance, 3, 529-543.
4, 127-134. Kelso, J.A.S. (1982). Concepts and issues in human motor
Keele, S.W., & Hawkins, H.L. (1982). Explorations of behavior: Coming to grips with the jargon. In J.A.S.
individual differences relevant to high level skill. Kelso (Ed.), Human motor behavior: An introduction
Journal of Motor Behavior, 14, 3-23. (pp. 21-58). Hillsdale, NJ: Erlbaum.
Keele, S.W., & Ivry, R.B. (1987). Modular analysis of Kelso, J.A.S. (1984). Phase transitions and critical
timing in motor skill. In C.H. Bower (Ed.), The behavior in human bimanual coordination. American
psychology of learning and motivation (Vol. 21) (pp. Journal of Physiology: Regulatory, Integrative and
183-228). New York: Academic Press. Comparative Physiology, 15, R1000-Rl004.
Keele, S.W., Ivry, R., & Pokorny, R.A. (1987). Force Kelso, J.A.S. (1992). Theoretical concepts and strategies
control and its relation to timing. Journal of Motor for understanding perceptual-motor skill: From
Behavior, 19, 96-114. information capacity in closed systems to self-
Keele, S.W., Jennings, P., Jones, S., Caulton, D., & organization in open, nonequilibrium systems.
Cohen, A. (1995). On the modularity of sequence Journal of Experimental Psychology: General, 121,
representation. Journal of Motor Behavior, 27, 17-30. 260-261.
Keele, S.W., Pokorny, R.A., Corcos, D.M., & Ivry, R. Kelso, J.A.S. (1995). Dynamic patterns: The self-
(1985). Do perception and motor production share organization of brain and behavior. Cambridge, MA:
common timing mechanisms: A correlational MIT Press.
analysis. Acta Psychologica, 60, 173-191. Kelso, J.A.S. (2008). Haken-Kelso-Bunz model.
Keele, S.W., & Posner, M.I. (1968). Processing of Scholarpedia, 3(10), 1612.
visual feedback in rapid movements. Journal of Kelso, J.A.S., Buchanan, J.J., & Wallace, S.A. (1991). Order
Experimental Psychology, 77, 155-158. parameters for the neural organization of single,
Keetch, K.M., & Lee, T.D. (2007). The effect of self- multijoint limb movement patterns. Experimental
regulated and experimenter-imposed practice Brain Research, 85, 432-444.
schedules on motor learning for tasks of varying Kelso, J.A.S., & Engstrom, D.A. (2005). The
difficulty. Research Quarterly for Exercise and Sport, complementary nature. Cambridge, MA: MIT Press.
78, 476-486.
Kelso, J.A.S., Holt, K.G., & Flatt, A.E. (1980). The
Keetch, K.M., Lee, T.D., & Schmidt, R.A. (2008). Especial role of proprioception in the perception and
skills: Specificity embedded within generality. control of human movement: Toward a theoretical
Journal of Sport and Exercise Psychology, 30, 723-736. reassessment. Perception & Psychophysics, 28, 45-52.
Keetch, K.M., Schmidt, R.A., Lee, T.D., & Young, D.E. Kelso, J.A.S., & Jeka, J.J. (1992). Symmetry breaking
(2005). Especial skills: Their emergence with massive dynamics of human multilimb coordination. Journal
amounts of practice. Journal of Experimental of Experimental Psychology: Human Perception and
Psychology: Human Perception and Performance, Performance, 18, 645-668.
31, 970-978.
Kelso, J.A.S., Putnam, C.A., & Goodman, D. (1983).
Keisler, A., & Willingham, D.T. (2007). Non-declarative On the space-time structure of human interlimb
sequence learning does not show savings in coordination. Quarterly Journal of Experimental
relearning. Human Movement Science, 26, 247-256. Psychology, 35A, 347-375.
Keller, F.S. (1958). The phantom plateau. Journal of the Kelso, J.A.S., Scholz, J.P., & Schaner, G. (1986).
Experimental Analysis of Behavior, 1, 1-13. Nonequilibrium phase transitions in coordinated
Kellogg, R.T. (2003). Cognitive psychology (2nd ed.). biological motion: Critical fluctuations. Physics
Thousand Oaks, CA: Sage. Letters A, 118, 279-284.
Kelly, J.P. (1991). The sense of balance. In E.R. Kandel, J.H. Kelso, J.A.S., Scholz, J.P., & Schaner, G. (1988). Dynamics
Schwartz, & T.M. Jessell (Eds.), Principles of neural governs switching among patterns of coordination in
science (3rd ed.) (pp. 500-511). Amsterdam: Elsevier. biological movement. Physics Letters A, 134, 8-12.
 References 529

Kelso, J.A.S., Southard, D.L., & Goodman, D. (1979). On Khan, M.A., Franks, I.M., Elliott, D., Lawrence, G.P.,
the coordination of two-handed movements. Journal Chua, R., Bernier, P.M., Hansen, S., & Weeks, D.J.
of Experimental Psychology: Human Perception and (2006). Inferring online and offline processing of
Performance, 5, 229-238. visual feedback in target-directed movements from
Kelso, J.A.S., & Stelmach, G.E. (1976). Central and kinematic data. Neuroscience & Biobehavioral
peripheral mechanisms in motor control. In G.E. Reviews, 30, 1106-1121.
Stelmach (Ed.), Motor control: Issues and trends (pp. Klapp, S.T. (1977a). Reaction time analysis of
1-40). New York: Academic Press. programmed control. Exercise and Sport Sciences
Kelso, J.A.S., Stelmach, G.E., & Wannamaker, W.M. Reviews, 5, 231-253.
(1976). The continuing saga of the nerve compression Klapp, S.T. (1977b). Response programming, as assessed
block technique. Journal of Motor Behavior, 8, 155- by reaction time, does not establish commands for
160. particular muscles. Journal of Motor Behavior, 9, 301-
Kelso, J.A.S., Tuller, B., Vatikiotis-Bateson, E., & Fowler, 312.
C.A. (1984). Functionally specific articulatory Klapp, S.T. (1979). Doing two things at once: The role
cooperation following jaw perturbations during of temporal compatibility. Memory & Cognition, 7,
speech: Evidence for coordinative structures. Journal 375-381.
of Experimental Psychology: Human Perception and
Klapp, S.T. (1981). Temporal compatibility in dual
Performance, 10, 812-832.
motor tasks II: Simultaneous articulation and hand
Kelso, J.A.S., & Zanone, P.-G. (2002). Coordination movements. Memory & Cognition, 9, 398-401.
dynamics of learning and transfer across different
Klapp, S.T. (1995). Motor response programming during
effector systems.Journal of Experimental Psychology:
simple and choice reaction time: The role of practice.
Human Perception and Performance, 28, 776-797.
Journal of Experimental Psychology: Human
Kennerley, S.W., Diedrichsen, J., Hazeltine, E., Semjen, Perception and Performance, 21, 1015-1027.
A., & lvry, R.B. (2002). Callosotomy patients exhibit
Klapp, S.T. (1996). Reaction time analysis of central
temporal uncoupling during continuous bimanual
motor control. In H.N. Zelaznik (Ed.), Advances in
movements. Nature Neuroscience, 5, 376-381.
motor learning and control (pp. 13-35). Champaign,
Keogh, J., & Sugden, D. (1985). Movement skill IL: Human Kinetics.
development. New York: Macmillan.
Klapp, S.T., & Erwin, C.I. (1976). Relation between
Kerlinger, F.N. (1973). Foundations of behavioral programming time and duration of the response being
research (2nd ed.). New York: Holt, Rinehart & programmed. Journal of Experimental Psychology:
Winston. Human Perception and Performance, 2, 591-598.
Kernodle, M.W., & Carlton, L.G. (1992). Information Klapp, S.T., Hill, M.D., Tyler, J.G., Martin, Z.E.,
feedback and the learning of multiple-degree-of- Jagacinski, R.J., & Jones, M.R. (1985). On marching
freedom activities. Journal of Motor Behavior, 24, to two different drummers: Perceptual aspects of
187-196. the difficulties. Journal of Experimental Psychology:
Kerr, B. (1973). Processing demands during mental Human Perception and Performance, 11, 814-827.
operations. Memory & Cognition, 1, 401-412. Klapp, S.T., Martin, Z.E., McMillan, G.G., & Brock, D.T.
Kerr, B. (1975). Processing demands during movement. (1987). Whole-task and part-task training in dual
Journal of Motor Behavior, 7, 15-27. motor tasks. In L.S. Mark, J.S. Warm, & R.L. Huston
Kerr, B., Condon, S.M., & McDonald, L.A. (1985). (Eds.), Ergonomics and human factors (pp. 125-130).
Cognitive spatial processing and the regulation of New York: Springer-Verlag.
posture.Journal of Experimental Psychology: Human Klapp, S.T., Nelson, J.M., & Jagacinski, R.J. (1998). Can
Perception and Performance, 11, 617-622. people tap concurrent bimanual rhythms indepen-
Kerr, R. (1973). Movement time in an underwater dently? Journal of Motor Behavior, 30, 301-322.
environment. Journal of Motor Behavior, 5, 175-178. Klavora, P. (1977). An attempt to derive inverted-U
Kerr, R. (1978). Diving, adaptation, and Fitts Law. curves based on the relationship between anxiety
Journal of Motor Behavior, 10, 255-260. and athletic performance. In D.M. Landers & R.W.
Kerr, R., & Booth, B. (1977). Skill acquisition in Christina (Eds.), Psychology of motor behavior and
elementary school children and schema theory. In sport-1976 (pp. 369-377). Champaign, IL: Human
D.M. Landers & R.W. Christina (Eds.), Psychology of Kinetics.
motor behavior and sport-1976 (Vol. 2). Champaign, Klein, R.M. (2000). Inhibition of return. Trends in
IL: Human Kinetics. Cognitive Sciences, 4, 138-147.
Kerr, R., & Booth, B. (1978). Specific and varied practice Klein, R.M. (2004). Orienting and inhibition of return.
of motor skill. Perceptual and Motor Skills, 46, 395- In M.S. Gazzaniga (Ed.), The cognitive neurosciences
401. (3rd ed.) (pp. 545-560). Cambridge, MA: MIT Press.
530 111 References

Klein, R., Levy, S., & McCabe, J. (1984). The parameter Kovacs, A.J., Buchanan, J.J., & Shea, C.H. (2010).
preferences of acquired motor programs for rapid, Impossible is nothing: 5:3 and 4:3 multi-frequency
discrete movements: I. Transfer of training. Memory bimanual coordination. Experimental Brain Research,
& Cognition, 12, 374-379. 201, 249-259.
Kleine, D. (1982). Psychomotorik und Intelligenz. Krakauer, J., & Chez, C. (2000). Voluntary movement.
Theoretische und empirische Untersuchungen In E.R. Kandel, J.H. Schwartz, & T.M. Jessell (Eds.),
unter Beriicksichtigung von Ubungseffekten im Principles of neural science (pp. 756-799). New York:
Psychomotorikberich [Psychomotor skills and McGraw-Hill.
intelligence. Theoretical and empirical studies with Krakauer, J.W., & Shadmehr, R. (2006). Consolidation of
consideration of training effects in psychomotor motor memory. Trends in Neurosciences, 29, 58-64.
information]. Unpublished doctoral dissertation,
Free University of Berlin. Krist, H., Fieberg, E.L., & Wilkening, F. (1993). Intuitive
physics in action and judgment: The development
Kleine, D. (1985). Psychomotor performance and of knowledge about projectile motion. Journal of
intellectual abilities. An analysis of the changes in Experimental Psychology: Learning, Memory, and
the relationship during practice. Paper presented at Cognition, 19, 952-966.
the VIth International Congress for Sport Psychology,
Copenhagen. Kucera, H., & Francis, W.N. (1967). Computational an-
alysis of present-day American English. Providence,
Klemmer, E.T. (1956). Time uncertainty in simple RI: Brown University Press.
reaction time. Journal of Experimental Psychology,
51, 179-184. Kugler, P.N., Kelso, J.A.S., & Turvey, M.T. (1980). On
the concept of coordinative structures as dissipative
Knapp, B. (1963). Skill in sport: The attaiillnent of
structures: I. Theoretical lines of convergence. In
proficiency. London: Routledge & Kegan Paul.
G.E. Stelmach & J. Requin (Eds.), Tutorials in motor
Kohl, R.M., & Fisicaro, S.A. (1995). Imaging goal- behavior (pp. 3-47). Amsterdam: Elsevier.
directed movement. Research Quarterly for Exercise
Kugler, P.N., & Turvey, M.T. (1987). Information, natural
and Sport, 66, 17-31.
law, and the self-assembly of rhythmic movement.
Konczak, J., vander Velden, H., & Jaeger, L. (2009). Hillsdale, NJ: Erlbaum.
Learning to play the violin: Motor control by freezing,
Kuhn, T.S. (1962). The structure of scientific revolutions.
not freeing degrees of freedom. Journal of Motor
Chicago: University of Chicago Press.
Behavior, 41, 243-252.
Kunde, W., & Weigelt, M. (2005). Goal congruency in
Konzem, P.B. (1987). Extended practice and patterns
bimanual object manipulation. Journal of Experimen-
of bimanual interference. Unpublished doctoral
tal Psychology: Human Perception and Performance,
dissertation, University of Southern California.
31, 145-156.
Kornblum, S., Hasbroucq, T., & Osman, A. (1990).
Kurtz, S., & Lee, T.D. (2003). Part and whole perceptual-
Dimensional overlap: Cognitive basis for stimulus-
motor practice of a polyrhythm. Neuroscience
response compatibility-a model and taxonomy.
Letters, 338, 205-208.
Psychological Review, 97, 253-270.
Kvalseth, T.O. (1980). An alternative to Fitts' law. Bulletin
Kornell, N., & Bjork, R.A. (2008). Learning concepts
of the Psychonomic Society, 16, 371-373.
and categories: Is spacing the "enemy of induction"?
Psychological Science, 19, 585-592. Kveraga, K., Boucher, L., & Hughes, H.C. (2002). Saccades
Kostrubiec, V., & Zanone, P.-G. (2002). Memory operate in violation of Hick's law. Experimental Brain
dynamics: Distance between the new task and Research, 146, 307-314.
existing behavioural patterns affects learning and Kyllo, L.B., & Landers, D.M. (1995). Goal setting in
interference in bimanual coordination in humans. sport and exercise: A research synthesis to resolve
Neuroscience Letters, 331, 193-197. the controversy. Journal of Sport and Exercise
Kots, Y.M. (1977). The organization of voluntary Psychology, 17, 117-137.
movement: Neurophysiological mechanisms. New Laabs, G.J. (1973). Retention characteristics of different
York: Plenum Press. reproduction cues in motor short-term memory.
Koustanai:, A., Boloix, E., Van Elslande, P., & Bastien, Journal of Experimental Psychology, 100, 168-177.
C. (2008). Statistical analysis of "looked-but-failed- Laban, R. (1956). Principles of dance and movement
to-see" accidents: Highlighting the involvement of notation. London: MacDonald and Evans.
two distinct mechanisms. Accident Analysis and LaBerge, D. (1973). Identification of two components
Prevention, 40, 461-469. of the time to switch attention: A test of a serial and
Kovacs,A.J., Buchanan, J.J., & Shea, C.H. (2009). Bimanual a parallel model of attention. In S. Kornblum (Ed.),
1:1 with 90° continuous relative phase: Difficult or Attention and performance IV (pp. 71-85). New York:
easy! Experimental Brain Research, 193, 129-136. Academic Press.
 References 531

Lackner, J.R., & DiZio, P.A. (2000). Aspects of body self- in behavior: The Hixon symposium (pp. 112-136).
calibration. Trends in Cognitive Sciences, 4, 279-288. New York: Wiley.
Laguna, P.L. (2008). Task complexity and sources of task- Laszlo, J.I. (1967). Training of fast tapping with reduction
related information during the observational learning of kinaesthetic, tactile, visual and auditory sensations.
process. Journal of Sports Sciences, 26, 1097-1113. Quarterly Journal of Experimental Psychology, 19,
Lai, Q., & Shea, C.H. (1998). Generalized motor program 344-349.
(GMP) learning: Effects of reduced frequency of Laszlo, J.1., & Bairstow, P.J. (1979). The compression-
knowledge of results and practice variability. Journal block technique: A reply to Chambers and Schumsky
of Motor Behavior, 30, 51-59. (1978). Journal of Motor Behavior, 11, 283-284.
Lai, Q., & Shea, C.H. (1999). Bandwidth knowledge Latash, M.L. (1993). Control of human movement.
of results enhances generalized motor program Champaign, IL: Human Kinetics.
learning. Research Quarterly for Exercise and Sport, Latash, M.L. (1999). Mirror writing: Learning, transfer,
70, 79-83. and implications for internal inverse models. Journal
Lai, Q., Shea, C.H., Wulf, G., & Wright, D.L. (2000). of Motor Behavior, 31, 107-111.
Optimizing generalized motor program and
Latash, M.L. (2008a). Evolution of motor control: From
parameter learning. Research Quarterly for Exercise
reflexes and motor programs to the equilibrium-point
and Sport, 71, 10-24.
hypothesis. Journal of Human Kinetics, 19, 3-24.
Lam, W.K., Maxwell, J.P., & Masters, R.S.W. (2009).
Latash, M.L. (2008b). Neurophysiological basis of
Analogy versus explicit learning of a modified
movement (2nd ed.). Champaign, IL: Human
basketball shooting task. Performance and kinematic
Kinetics.
outcomes. Journal of Sports Sciences, 27, 179-191.
Latash, M.L., & Gutman, S.R. (1993). Variability of
Lamble, D., Kauranen, T., Laasko, M., & Summala, H.
fast single-joint movements and the equilibrium-
(1999). Cognitive load and detection thresholds in car
point hypothesis. In K.M. Newell & D.M. Coreas
following situations: Safety implications for using
(Eds.), Variability and motor control (pp. 157-182).
mobile (cellular) telephones while driving. Accident
Champaign, IL: Human Kinetics.
Analysis and Prevention, 31, 617-623.
Latash, M.L., Scholz, J.P., & Scheiner, G. (2007). Toward
Landin, D., & Hebert, E.P. (1997). A comparison of three
a new theory of motor synergies. Motor Control, 11,
practice schedules along the contextual interference
276-308.
continuum. Research Quarterly for Exercise and
Sport, 68, 357-361. Latash, M.L., & Zatsiorsky, V.M. (Eds.). (2001). Classics in
movement science. Champaign, IL: Human Kinetics.
Lang,P.J. (1974). Learned controlofhumanheartratein a
computer directed environment. In P.A. Obrist, A.H. Lavery, J.J. (1962). Retention of simple motor skills as a
Black, J. Brenner, & L.V. DiCara (Eds.), Cardiovascular function of type of knowledge of results. Canadian
psychophysiology: Current issues in response Journal of Psychology, 16, 300-311.
mechanisms, biofeedback, and methodology (pp. Lavery, J.J., & Suddon, F.H. (1962). Retention of simple
392-405). Chicago: Aldine. motor skills as a function of the number of trials by
Langer, E.J. (1989). Mindfulness. New York: Addison- which KR is delayed. Perceptual and Motor Skills, 15,
Wesley. 231-237.
Langfeld, H.S. (1915). Facilitation and inhibition of Lay, B.S., Sparrow, W.A., & O'Dwyer, N.J. (2005). The
motor impulses: A study in simultaneous and metabolic and cognitive energy costs of stabilising
alternating finger movements. Psychological Review, a high-energy interlimb coordination task. Human
22, 453-478. Movement Science, 24, 833-848.
Langolf, G.D., Chaffin, D.B., & Foulke, J.A. (1976). An Leavitt, J.L. (1979). Cognitive demands of skating and
investigation of Fitts' law using a wide range of stick-handling in ice hockey. Canadian Journal of
movement amplitudes. Journal of Motor Behavior, 8, Applied Sports Science, 4, 46-55.
113-128. Lederman, S.J., & Klatzky, R.L. (1997). Haptic aspects of
Lashley, K.S. (1917). The accuracy of movement in motor control. In M. Jeannerod (Ed.), Handbook of
the absence of excitation from the moving organ. neuropsychology (Vol. 11) (pp.131-148). Amsterdam:
American Journal of Physiology, 43, 169-194. Elsevier.
Lashley, K.S. (1942). The problem of cerebral organization Lederman, S.J., & Klatzky, R.L. (2009). Haptic perception:
in vision. In J. Cattell (Ed.), Biological symposia. Vol. A tutorial. Attention, Perception, & Psychophysics,
VII. Visual mechanisms (pp. 301-322). Lancaster, PA: 71, 1439-1459.
Jaques Cattell Press. Lee, D.N. (1976). A theory of visual control of braking
Lashley, K.S. (1951). The problem of serial order in based on information about time-to-collision.
behavior. In L.A. Jeffress (Ed.), Cerebral mechanisms Perception, 5, 437-459.
532 References

Lee, D.N. (1980). Visuo-motor coordination in space- Lee, T.D., Almeida, Q.J., & Chua, R. (2002). Spatial
time. In G.E. Stelmach & J. Requin (Eds.), Tutorials constraints in bimanual coordination: Influences of
in motor behavior (pp. 281-295). Amsterdam: North- effector orientation. Experimental Brain Research,
Holland. 146, 205-212.
Lee, D.N. (1990). Getting around with light or sound. In Lee, T.D., Blandin, Y., & Proteau, L. (1996). Effects of task
R. Warren & A.H. Wertheim (Eds.), Perception and instructions and oscillation frequency on bimanual
control of self-motion (pp. 487-505). Hillsdale, NJ: coordination. Psychological Research, 59, 100-106.
Erlbaum. Lee, T.D., & Carnahan, H. (1990a). Bandwidth
Lee, D.N. (1998). Guiding movement by coupling taus. knowledge of results and motor learning: More than
Ecological Psychology, 10, 221-250. just a relative frequency effect. Quarterly Journal of
Lee, D.N. (2009). General tau theory: Evolution to date. Experimental Psychology, 42A, 777-789.
Perception, 38, 837-850. Lee, T.D., & Carnahan, H. (1990b). When to provide
Lee, D.N., & Aronson, E. (1974). Visual proprioceptive knowledge of results during motor learning:
control of standing in human infants. Perception & Scheduling effects. Human Performance, 3, 87-105.
Psychophysics, 15, 529-532. Lee, T.D., & Genovese, E.D. (1988). Distribution of
Lee, D.N., Georgopolous,A.P., Clark, M.J.O., Craig, C.M., practice in motor skill acquisition: Learning and
& Port, N.L. (2001). Guiding contact by coupling the performance effects reconsidered. Research Quarterly
taus of gaps. Experimental Brain Research, 139, 151- for Exercise and Sport, 59, 277-287.
159. Lee, T.D., & Genovese, E.D. (1989). Distribution of
Lee, D.N., Lishman, J.R., & Thomson, J.A. (1982). practice in motor skill acquisition: Different effects
Regulation of gait in long jumping. Journal of for discrete and continuous tasks. Research Quarterly
Experimental Psychology: Human Perception and for Exercise and Sport, 60, 59-65.
Performance, 8, 448-459. Lee, T.D., Ishikura, T., Kegel, S., Gonzalez, D., &
Lee, D.N., & Young, D.S. (1985). Visual timing of Passmore, S. (2008a). Head-putter coordination
interceptive action. In D. Ingle, M. Jeannerod, & D.N. patterns in expert and less-skilled golfers. Journal of
Lee (Eds.), Brain mechanisms and spatial vision (pp. Motor Behavior, 40, 267-272.
1-30). Dordrecht: Martinus Nijhoff. Lee, T.D., Ishikura, T., Kegel, S., Gonzalez, D., &
Lee, D.N., Young, D.S., Reddish, P.E., Lough, S., & Passmore, S. (2008b). Do expert golfers really keep
Clayton, T.M.H. (1983). Visual timing in hitting an their heads still while putting? Annual Review of
accelerating ball. Quarterly Journal of Experimental Golf Coaching, 2, 135-143.
Psychology, 35A, 333-346. Lee, T.D., & Magill, R.A. (1983a). Activity during the
Lee, J.D. (2008). Fifty years of driving safety research. post-KR interval: Effects upon performance or
Human Factors, 50, 521-528. learning? Research Quarterly for Exercise and Sport,
54, 340-345.
Lee, J.D., Caven, B., Haake, S., & Brown, T.L. (2001).
Speech-based interaction with in-vehicle computers: Lee, T.D., & Magill, R.A. (1983b). The locus of contextual
The effect of speech-based e-mail on drivers' attention interference in motor-skill acquisition. Journal of
to the roadway. Human Factors, 43, 631-640. Experimental Psychology: Learning, Memory, and
Cognition, 9, 730-746.
Lee, R.G., Murphy, J.T., & Tatton, W.G. (1983). Long
latency myotatic reflexes in man: Mechanisms, Lee, T.D., & Magill, R.A. (1985). Can forgetting facilitate
functional significance, and changes in patients with skill acquisition? In D. Goodman, R.B. Wilberg, &
Parkinson's disease or hemiplegia. In J. Desmedt I.M. Franks (Eds.), Differing perspectives in motor
(Ed.), Advances in neurology (pp. 489-508). Basel: learning, memory, and control (pp. 3-22). Amsterdam:
Karger. Elsevier.
Lee, T.D. (1988). Transfer-appropriate processing: A Lee, T.D., & Magill, R.A. (1987). Effects of duration
framework for conceptualizing practice effects in and activity during the post-KR interval on motor
motor learning. In O.G. Meijer & K. Roth (Eds.), learning. Psychological Research, 49, 237-242.
Complex movement behaviour: "The" motor-action Lee, T.D., Magill, R.A., & Weeks, D.J. (1985). Influence
controversy (pp. 201-215). Amsterdam: Elsevier. of practice schedule on testing schema theory
Lee, T.D. (1998). On the dynamics of motor learning predictions in adults. Journal of Motor Behavior, 17,
research. Research Quarterly for Exercise and Sport, 283-299.
69, 334-337. Lee, T.D., & Maraj, B.K.V. (1994). Effects of bandwidth
Lee, T.D. (2004). Intention in bimanual coordination goals and bandwidth knowledge of results on motor
performance and learning. In V.K. Jirsa & J.A.S. Kelso learning. Research Quarterly for Exercise and Sport,
(Eds.), Coordination dynamics: Issues and trends 65, 244-249.
(pp. 41-56). Berlin: Springer. Lee, T.D., & Simon, D.A. (2004). Contextual interference.
 References w 533

In AM. Williams & N.J. Hodges (Eds.), Skill Leuba, J.H., & Chamberlain, E. (1909). The influence of
acquisition in sport: Research, theory and practice the duration and of the rate of arm movements upon
(pp. 29-44). London: Routledge. the judgment of their length. American Journal of
Lee, T.D., & Swinnen, S.P. (1993). Three legacies of Bryan Psychology, 20, 374-385.
and Harter: Automaticity, variability and change in Lewis, D. (1947). Positive and negative transfer in motor
skilled performance. InJ.L. Starkes & F. Allard (Eds.), learning. American Psychologist, 2, 423.
Cognitive issues in motor expertise (pp. 295-315). Lewis, D. (1953). Motor skills learning. In AB. Nadel
Amsterdam: Elsevier. (Ed.), Symposium on psychology of learning basic to
Lee, T.D., Swinnen, S.P., & Serrien, D.J. (1994). Cognitive military training problems (HR-HTD 201/l, pp. 45-
effort and motor learning. Quest, 46, 328-344. 84). Washington, DC: Department of Defense.
Lee, T.D., Swinnen, S.P., & Verschueren, S. (1995). Lewis, D., McAllister, D.E., & Adams, J.A. (1951).
Relative phase alterations during bimanual skill Facilitation and interference in performance on
acquisition. Journal of Motor Behavior, 27, 263-274. the modified Mashburn apparatus: I. The effects of
Lee, T.D., & Weeks, D.J. (1987). The beneficial influence varying the amount of original learning. Journal of
of forgetting on short-term retention of movement Experimental Psychology, 41, 247-260.
information. Human Movement Science, 6, 233-245. Lewis, R.E.F. (1956). Consistency and car driving skill.
Lee, T.D., White, M.A., & Carnahan, H. (1990). On British Journal oflndustrial Medicine, 13, 131-141.
the role of knowledge of results in motor learning: Lewthwaite, R., & Wulf, G. (2010). Social-comparative
Exploring the guidance hypothesis. Journal of Motor feedback affects motor skill learning. Quarterly
Behavior, 22, 191-208. Journal of Experimental Psychology, 63, 738-749.
Lee, T.D., Wishart, L.R., Cunningham, S., & Carnahan, H. Li, Y., & Wright, D.L. (2000). An assessment of the
(1997). Modeled timing information during random attention demands during random- and blocked-
practice eliminates the contextual interference effect. practice schedules. Quarterly Journal of Experimental
Research Quarterly for Exercise and Sport, 68, 100- Psychology, 53A, 591-606.
105. Liao, C.M., & Masters, R.S.W. (2001). Analogy learning:
Lee, T.D., Wulf, G., & Schmidt, R.A. (1992). Contextual A means to implicit motor learning. Journal of Sports
interference in motor learning: Dissociated effects Sciences, 19, 307-319.
due to the nature of task variations. Quarterly Journal Libet, B. (1985). Unconscious cerebral initiative and the
of Experimental Psychology, 44A, 627-644. role of conscious will in voluntary action. Behavioral
Lee, W.A. (1980). Anticipatory control of postural and and Brain Sciences, 8, 529-566.
task muscles during rapid arm flexion. Journal of Lieberman, D.A., Vogel, A.CM., & Nisbet, J. (2008). Why
Motor Behavior, 12, 185-196. do the effects of delaying reinforcement in animals
Lenoir, M., Musch, E., Thiery, E., & Savelsbergh, G.J.P. and delaying feedback in humans differ? A working
(2002). Rate of change of angular bearing as the memory analysis. Quarterly Journal of Experimental
relevant property in a horizontal interception task Psychology, 61, 194-202.
during locomotion. Journal of Motor Behavior, 34, Lin, C.-H., Fisher, B.E., Winstein, C.J., Wu, AD., &
385-401. Gordon, J. (2008). Contextual interference effect:
Leob, G.E., & Ghez, C. (2000). The motor unit and muscle Elaborative processing or forgetting-reconstruction?
action. In E.R. Kandel, J.H. Schwartz, & T.M. Jessell A post hoc analysis of transcranial magnetic
(Eds.), Principles of neural science (pp. 674-694). New stimulation-induced effects on motor learning.
York, McGraw-Hill. Journal of Motor Behavior, 40, 578-586.
Leonard, J.A. (1953). Advance information in sensori- Lin, C.-H., Sullivan, K.J., Wu, AD., Kantak, K., &
motor skills. Quarterly Journal of Experimental Winstein, C.J. (2007). Effect of task practice order on
Psychology, 5, 141-149. motor skill learning in adults with Parkinson disease:
Leonard, J.A. (1954). An experiment with occasional A pilot study. Physical Therapy, 87, 1120-1131.
false information. Quarterly Journal of Experimental Lin, C.-H., Winstein, C.J., Fisher, B.E., & Wu, AD. (2010).
Psychology, 6, 79-85. Neural correlates of the contextual interference effect
Leonard, J.A. (1959). Tactual choice reactions: I. Quarterly in motor learning: A transcranial magnetic stimulation
Journal of Experimental Psychology, 11, 76-83. investigation. Journal of Motor Behavior, 42, 223-232.
Lersten, K.C. (1968). Transfer of movement components Lindahl, L.G. (1945). Movement analysis as an industrial
in a motor learning task. Research Quarterly, 39, 575- training method. Journal of Applied Psychology, 29,
581. 420-436.
Lersten, K.C. (1969). Retention of skill on the Rho Lindeburg, F.A. (1949). A study of the degree of transfer
apparatus after one year. Research Quarterly, 40, 418- between quickening exercises and other coordinated
419. movements. Research Quarterly, 20, 180-195.
534 References

Lindemann, 0., Abolafia, J.M., Girardi, G., & Bekkering, Lotze, M., & Halsband, U. (2006). Motor imagery. Journal
H. (2007). Getting a grip on numbers: Numerical of Physiology- Paris, 99, 386-395.
magnitude priming in object grasping. Journal of Maas, E., Robin, D.A, Austermann Hula, S.N.,
Experimental Psychology: Human Perception and Freedman, S.E., Wulf, G., Ballard, K.J., & Schmidt,
Performance, 33, 1400-1409. R.A (2008). Principles of motor learning in treatment
Lintern, G. (1988). Distributed practice: Are there useful of motor speech disorders. American Journal of
insights for application or theory? Research Quarterly Speech-Language Pathology, 17, 277-298.
for Exercise and Sport, 59, 298-302. Mack, M.G. (2001). Effects of time and movements of the
Lintern, G., Sheppard, D.J., Parker, D.L., Yates, K.E., preshot routine on free throw shooting. Perceptual
& Nolan, M.D. (1989). Simulator design and and Motor Skills, 93, 567-573.
instructional features for air-to-ground attack: A MacKay-Lyons, M. (2002). Central pattern generation
transfer study. Human Factors, 31, 87-99. of locomotion: A review of the evidence. Physical
Liu, J., & Wrisberg, C.A (1997). The effects of knowledge Therapy, 82, 69-83.
of results delay and the subjective estimation of MacKenzie, LS. (1989). A note on the information-
movement form on the acquisition and retention of theoretic basis of Fitts' law. Journal of Motor Behavior,
a motor skill. Research Quarterly for Exercise and 21, 323-330.
Sport, 68, 145-151.
MacKenzie, LS., & Buxton, W. (1994). The prediction
Locke, E.A (1991). Problems with goal-setting research of pointing and dragging times in graphical user
in sports-and their solution. Journal of Sport and interfaces. Interacting with Computers, 6, 213-227.
Exercise Psychology, 13, 311-316.
MacLeod, C.M. (1991). Half a century of research on the
Locke, E.A, & Latham, G.P. (1985). The application of Stroop effect: An integrative review. Psychological
goal setting to sports. Sport Psychology Today, 7, 205- Bulletin, 109, 163-203.
222.
MacLeod, C.M., & Dunbar, K. (1988). Training and
Locke, E.A, & Latham, G.P. (2006). New directions Stroop-like interference: Evidence for a continuum
in goal-setting theory. Current Directions in of automaticity. Journal of Experimental Psychology:
Psychological Science, 15, 265-268. Learning, Memory, and Cognition, 14, 126-135.
Lococo, K., & Tucker, M.E. (in press). Pedal application MacLeod, C.M., & MacDonald, P.A. (2000).
errors. Washington, DC: NHTSA, U.S. Department of Interdimensional interference in the Stroop effect:
Transportation. Uncovering the cognitive and neural anatomy of
Logan, G.D. (1982). On the ability to inhibit complex attention. Trends in Cognitive Sciences, 4, 383-391.
movements: A stop-signal study of typewriting. MacMahon, C., & McPherson, S.L. (2009). Knowledge
Journal of Experimental Psychology: Human base as a mechanism for perceptual-cognitive tasks:
Perception and Performance, 8, 778-792. Skill is in the details! International Journal of Sport
Logan, G.D. (1985). Skill and automaticity: Relations, Psychology, 40, 565-579.
implications, and future directions. Canadian Journal MacMahon, C., & Starkes, J.L. (2008). Contextual
of Psychology, 39, 367-386. influences on baseball ball-strike decisions in
Logan, G.D. (1988). Toward an instance theory of umpires, players, and controls. Journal of Sports
automatization. Psychological Review, 95, 492-527. Sciences, 26, 751-760.
Logan, G.D. (1994). On the ability to inhibit thought and MacNeilage, P.F. (1970). Motor control of serial ordering
action: A user's guide to the stop signal paradigm. of speech. Psychological Review, 77, 182-196.
In D. Dagenbach & T.H. Carr (Eds.), Inhibitory MacPherson, S.J., Dees, V., & Grindley, G.C. (1948).
processes in attention, memory, and language (pp. The effect of knowledge of results on learning and
189-239). San Diego: Academic Press. performance: II. Some characteristics of very simple
Logan, G.D., & Cowan, W.B. (1984). On the ability skills. Quarterly Journal of Experimental Psychology,
to inhibit thought and action: A theory of an act of 1, 68-78.
control. Psychological Review, 91, 295-327. MacPherson, S.J., Dees, V., & Grindley, G.C. (1949).
Lordahl, D.S., & Archer, E.J. (1958). Transfer effects on a The effect of knowledge of results on learning and
rotary pursuit task as a function of first-task difficulty. performance: III. The influence of time intervals
Journal of Experimental Psychology, 56, 421-426. between trials. Quarterly Journal of Experimental
Lorge, L, & Thorndike, E.L. (1935). The influence of Psychology, 1, 167-174.
delay in the after-effect of a connection. Journal of Macrae, AW., & Holding, D.H. (1965). Method and task
Experimental Psychology, 18, 186-194. in motor guidance. Ergonomics, 8, 315-320.
Lotter, W.S. (1960). Interrelationships among reaction Macrae, AW., & Holding, D.H. (1966). Transfer of
times and speeds of movement in different limbs. training after guidance or practice. Quarterly Journal
Research Quarterly, 31, 147-155. of Experimental Psychology, 18, 327-333.
 References 535

Magill, R.A. (1973). The post-KR interval: Time and Marsden, C.D., Merton, P.A., & Morton, H.B. (1972).
activity effects and the relationship of motor short- Servo action in human voluntary movement. Nature,
term memory theory. Journal of Motor Behavior, 5, 238, 140-143.
49-56. Marteniuk, R.G. (1973). Retention characteristics of
Magill, R.A. (1988). Activity during the post-knowledge motor short-term memory cues. Journal of Motor
of results interval can benefit motor skill learning. In Behavior, 5, 249-259.
O.G. Meijer & K Roth (Eds.), Complex movement Marteniuk, R.G. (1974). Individual differences in motor
behaviour: "The" motor-action controversy (pp. 231- performance and learning. Exercise and Sport
246). Amsterdam: Elsevier. Sciences Reviews, 2, 103-130.
Magill, R.A. (1998). Knowledge is more than we can talk Marteniuk, R.G. (1976). Information processing in motor
about: Implicit learning in motor skill acquisition. skills. New York: Holt, Rinehart & Winston.
Research Quarterly for Exercise and Sport, 69, 104-110.
Marteniuk, R.G. (1986). Information processes in move-
Magill, R.A. (2001). Augmented feedback in motor skill ment learning: Capacity and structural interference
acquisition. In R.N. Singer, H.A. Hausenblas, & C.M. effects. Journal of Motor Behavior, 18, 55-75.
Janelle (Eds.), Handbook of sport psychology (2nd
Marteniuk, R.G., Leavitt, J.L., MacKenzie, C.L., &
ed.) (pp. 86-114). New York: Wiley.
Athenes, S. (1990). Functional relationships between
Magill, R.A., Chamberlin, C.J., & Hall, KG. (1991). Verbal grasp and transport components in a prehension
knowledge of results as redundant information task. Human Movement Science, 9, 149-176.
for learning an anticipation timing skill. Human
Marteniuk, R.G., MacKenzie, C.L., & Baba, D.M.
Movement Science, 10, 485-507.
(1984). Bimanual movement control: Information
Magill, R.A., & Hall, KG. (1990). A review of the processing and interaction effects. Quarterly Journal
contextual interference effect in motor skill of Experimental Psychology, 36A, 335-365.
acquisition. Human Movement Science, 9, 241-289.
Marteniuk, R.G., MacKenzie, C.L., Jeannerod, M.,
Magill, R.A., Schmidt, R.A., Young, D.E., & Shapiro, D.C. Athenes, S., & Dugas, C. (1987). Constraints on
(1987). Unpublished data, University of California, human arm movement trajectories. Canadian Journal
Los Angeles. of Psychology, 41, 365-378.
Magill, R.A., & Schoenfelder-Zohdi, B. (1996). A visual Marteniuk, R.G., & Romanow, S.KE. (1983). Human
model and knowledge of performance as sources of movement organization and learning as revealed
information for learning a rhythmic gymnastics skill. by variability of movement, use of kinematic
International Journal of Sport Psychology, 27, 7-22. information, and Fourier analysis. In R.A. Magill
Magill, R.A., & Wood, C.A. (1986). Knowledge of (Ed.), Memory and control of action (pp. 167-197).
results precision as a learning variable in motor skill Amsterdam: Elsevier.
acquisition. Research Quarterly for Exercise and Martens, R., Burwitz, L., & Zuckerman, J. (1976).
Sport, 57, 170-173. Modeling effects on motor performance. Research
Malina, R., Bouchard, C., & Bar-Or, 0. (2004). Growth, Quarterly, 47, 277-291.
maturation, and physical activity (2nd ed.). Martens, R., & Landers, D.M. (1970). Motor performance
Champaign, IL: Human Kinetics. under stress: A test of the inverted-U hypothesis.
Mane, A.M., Adams, J.A., & Donchin, E. (1989). Adaptive Journal of Personality and Social Psychology, 16, 29-
and part-whole training in the acquisition of a 37.
complex perceptual-motor skill. Acta Psychologica, Martin, H.A. (1970). Long-term retention of a discrete
71, 179-196. motor task. Unpublished master's thesis, University
Mann, D.T.Y., Williams, A.M., Ward, P., & Janelle, C.M. of Maryland, College Park.
(2007). Perceptual-cognitive expertise in sport: Martin, J.H., & Jessell, T.M. (1991). Modality coding
A meta-analysis. Journal of Sport and Exercise in the somatic sensory system. In E.R. Kandel, J.H.
Psychology, 29, 457-478. Schwartz, & T.M. Jessell (Eds.), Principles of neural
Maraj, B.K.V., Elliott, D., Lee, T.D., & Pollock, B.J. (1993). science (3rd ed.) (pp. 341-352). Amsterdam: Elsevier.
Variance and invariance in expert and novice triple Maslovat, D., Brunke, KM., Chua, R., & Franks, I.M.
jumpers. Research Quarterly for Exercise and Sport, (2009). Feedback effects on learning a novel bimanual
64, 404-412. coordination pattern: Support for the guidance
Marchal Crespo, L., & Reinkensmeyer, D.J. (2008). hypothesis. Journal of Motor Behavior, 41, 45-54.
Haptic guidance can enhance motor learning of a Maslovat, D., Hayes, S.J., Horn, R., & Hodges, N.J.
steering task. Journal of Motor Behavior, 40, 545-556. (2010). Motor learning through observation. In D.
Marder, E., Bucher, D., Schulz, D.J., & Taylor, A.L. (2005). Elliott & M.A. Khan (Eds.), Vision and goal-directed
Invertebrate central pattern generation moves along. action: Neurobehavioural perspectives (pp. 315-339).
Current Biology, 15, R685-R699. Champaign, IL: Human Kinetics.
536 References

Massaro, D.W. (1989). Experimental psychology: McEvoy, S.P., Stevenson, M.R., & Woodward, M. (2007).
An information processing approach. San Diego: The contribution of passengers versus mobile phone
Harcourt Brace Jovanovich. use to motor vehicle crashes resulting in hospital
Masson, M.E.J. (1990). Cognitive theories of skill attendance by the driver. Accident Analysis and
acquisition. Human Movement Science, 9, 221-239. Prevention, 39, 1170-1176.
McGarry, T., Chua, R., & Franks, I.M. (2003). Stopping
Masters, R.S.W. (1992). Knowledge, knerves and know-
and restarting an unfolding action at various times.
how: The role of explicit versus implicit knowledge
Quarterly Journal of Experimental Psychology, 56A,
in the breakdown of a complex motor skill under
601-620.
pressure. British Journal of Psychology, 83, 343-358.
McGarry, T., & Franks, I.M. (1997). A horse race between
Masters, R.S.W. (2008). Applying implicit (motor)
independent processes: Evidence for a phantom
learning. In D. Farrow, J. Baker, & C. MacMahon (Eds.),
point of no return in the preparation of a speeded
Developing elite sports performers: Lessons from
motor response. Journal of Experimental Psychology:
theory to practice (pp. 89-103). Routledge: London.
Human Perception and Performance, 23, 1533-1542.
Masters, R.S.W., & Maxwell, J. (2008). The theory of
McGarry, T., & Franks, I.M. (2000). Inhibitory motor
reinvestment. International Review of Sport and
control in stop paradigms: Comment on Band and
Exercise Psychology, 1, 160-183.
van Boxtel (1999). Acta Psychologica, 105, 83-88.
Mates, J., & Aschersleben, G. (2000). Sensorimotor McGarry, T., & Franks, I.M. (2003). On the nature of
synchronization: The impact of temporally displaced stopping an earlier intended voluntary action. Motor
auditory feedback. Acta Psychologica, 104, 29-44. Control, 7, 155-198.
Mattar, A.AG., & Gribble, P.L. (2005). Motor learning by McGarry, T., Inglis, T., & Franks, I.M. (2000). Against
observing. Neuron, 46, 153-160. a final ballistic process in the control of voluntary
McCarthy, G., & Donchin, E. (1981). A metric for action: Evidence using the Hoffman reflex. Motor
thought: A comparison of P300 latency and reaction Control, 4, 469-485.
time. Science, 211, 77-80. McGaugh, J.L. (2000). Memory: A century of
McCartt, A.T., Hellinga, L.A., & Bratiman, K.A. (2006). consolidation. Science, 287, 248-251.
Cell phones and driving: Review of research. Traffic McGeoch, J.A., & Irion, A.L. (1952). The psychology of
Injury Prevention, 7, 89-106. human learning. New York: Longmans.
McCloy, C.H. (1934). The measurement of general McGraw, M.B. (1935). Growth: A study of Johnny and
motor capacity and general motor ability. Research Jimmy. New York: Appleton-Century.
Quarterly, 5 (Suppl. 5), 45-61.
McGraw, M.B. (1939). Later development of children
McCloy, C.H. (1937). An analytical study of the stunt specially trained during infancy: Johnny and Jimmy
type test as a measure of motor educability. Research at school age. Child Development, 10, 1-19.
Quarterly, 8, 46-55. McGuigan, F.J. (1959). The effect of precision, delay, and
McCracken, H.D., & Stelmach, G.E. (1977). A test of the schedule of knowledge of results on performance.
schema theory of discrete motor learning. Journal of Journal of Experimental Psychology, 58, 79-84.
Motor Behavior, 9, 193-201. McKnight, A.J., & McKnight, A.S. (1993). The effect of
McCullagh, P., & Caird, J.K. (1990). Correct and learning cellular phone use upon driver attention. Accident
models and the use of model knowledge of results in Analysis and Prevention, 25, 259-265.
the acquisition and retention of a motor skill. Journal McLeod, P. (1977). A dual task response modality effect:
of Human Movement Studies, 18, 107-116. Support for multiprocessor models of attention.
McCullagh, P., & Little, W.S. (1989). A comparison of mo- Quarterly Journal of Experimental Psychology, 29,
dalities in modeling. Human Performance, 2, 107-116. 651-667.
McCullagh, P., & Meyer, K.N. (1997). Learning versus McLeod, P. (1980). What can probe RT tell us about the
correct models: Influence of model type on the attentional demands of movement? In G.E. Stelmach
learning of a free-weight squat lift. Research & J. Requin (Eds.), Tutorials in motor behavior (pp.
Quarterly for Exercise and Sport, 68, 56-61. 579-589). Amsterdam: Elsevier.
McCullagh, P., & Weiss, M.R. (2001). Modeling: McLeod, P. (1987). Visual reaction time and high-speed
Considerations for motor skill performance and ball games. Perception, 16, 49-59.
psychological responses. In R.N. Singer, H.A. McLeod, P., & Dienes, Z. (1993). Running to catch the
Hausenblas, & C.M. Janelle (Eds.), Handbook of sport ball. Nature, 362, 23.
psychology (2nd ed.) (pp. 205-238). New York: Wiley. McLeod, P., & Dienes, Z. (1996). Do fielders know where
McDonald, P.V., van Emmerik, R.E.A., & Newell, K.M. to go to catch the ball or only how to get there? Journal
(1989). The effects of practice on limb kinematics in a of Experimental Psychology: Human Perception and
throwing task. Journal of Motor Behavior, 21, 245-264. Performance, 22, 531-543.
 References 537

McLeod, P., Reed, N., & Dienes, Z. (2001). Toward a & V.M. Zatsiorsky (Eds.), Classics in movement
unified fielder theory: What we do not yet know science (pp. 1-57). Champaign, IL: Human Kinetics.
about how people run to catch a ball. Journal of Meira, C.M. Jr., & Tani, G. (2001). The contextual
Experimental Psychology: Human Perception and interference effect in acquisition of dart-throwing
Performance, 27, 1347-1355. skill tested on a transfer test with extended trials.
McLeod, P., Reed, N., & Dienes, Z. (2003). How fielders Perceptual and Motor Skills, 92, 910-918.
arrive in time to catch the ball. Nature, 426, 244-245. Melton, AW. (Ed.). (1947). Apparatus tests. Washington,
McLeod, P., Reed, N., & Dienes, Z. (2006). The DC: U.S. Government Printing Office.
generalised optic acceleration cancellation theory Memmert, D., Hagemann, N., Althoetmar, R., Geppert,
of catching. Journal of Experimental Psychology: S., & Seiler, D. (2009). Conditions of practice in
Human Perception and Performance, 32, 139-148. perceptual skill learning. Research Quarterly for
McNeil, M.R., Robin, D., & Schmidt, R.A. (1997). Apraxia Exercise and Sport, 80, 32-43.
of speech: Definition, differentiation, and treatment. Memmert, D., & Furley, P. (2007). "I spy with my little
In M.R. McNeil (Ed.), Clinical management of eye!": Breadth of attention, inattentional blindness,
sensorimotor speech disorders (pp. 311-344). New and tactical decision making in team sports. Journal
York: Thieme Medical. of Sport and Exercise Psychology, 29, 365-381.
McNevin, N., Magill, R.A., & Buekers, M.J. (1994). The
Merkel, J. (1885). Die zeitlichen Verhaltnisse det
effects of erroneous knowledge of results on transfer
Willensthaiitigkeit. Philosophische Studien, 2, 73-
of anticipation timing. Research Quarterly for 127. (Cited in Woodworth, R.S., 1938, Experimental
Exercise and Sport, 65, 324-329.
psychology. New York: Holt)
McPherson, S.L. (1993). Knowledge representation and
Merton, P.A. (1953). Speculations on the servo control of
decision-making in sport. In J.L. Starkes & F. Allard
movement. In G.E.W. Wolstenholme (Ed.), The spinal
(Eds.), Cognitive issues in motor expertise (pp. 159-
cord. London: Churchill.
188). Amsterdam: Elsevier.
Merton, P.A. (1972). How we control the contraction of
McPherson, S.L. (1994). The development of sport exper-
our muscles. Scientific American, 226, 30-37.
tise: Mapping the tactical domain. Quest, 46, 223-240.
Meyer, D.E., Abrams, R.A., Kornblum, S., Wright,
McPherson, S.L., & MacMahon, C. (2008). How baseball
C.E., & Smith, J.E.K. (1988). Optimality in human
players prepare to bat: Tactical knowledge as a
motor performance: Ideal control of rapid aimed
mediator of expert performance in baseball. Journal
movements. Psychological Review, 95, 340-370.
of Sport and Exercise Psychology, 30, 755-778.
Meyer, D.E., Smith, J.E.K., Kornblum, S., Abrams, R.A.,
Mechsner, F. (2004). A perceptual-cognitive approach to
& Wright, C.E. (1990). Speed-accuracy tradeoffs
bimanual coordination. In V.K. Jirsa & J.A.S. Kelso
in aimed movements: Toward a theory of rapid
(Eds.), Coordination dynamics: Issues and trends
voluntary action. In M. Jeannerod (Ed.), Attention
(pp. 177-195). Berlin: Springer.
and performance XIII (pp. 173-226). Hillsdale, NJ:
Mechsner, F., Kerzel, D., Knoblich, G., & Prinz, W. (2001). Erlbaum.
Perceptual basis of bimanual coordination. Nature,
Meyer, D.E., Smith, J.E.K., & Wright, C.E. (1982). Models
414, 69-73.
for the speed and accuracy of aimed movements.
Mechsner, F., & Knoblich, G. (2004). Do muscles matter Psychological Review, 89, 449-482.
for coordinated action? Journal of Experimental
Meyers, J.L. (1967). Retention of balance coordination
Psychology: Human Perception and Performance,
learning as influenced by extended lay-offs. Research
30, 490-503.
Quarterly, 38, 72-78.
Meegan, D.V., Aslin, R.N., & Jacobs, R.A. (2000). Motor
Michaels, C.F., & Oudejans, R.R.D. (1992). The optics
timing learned without motor training. Nature
and actions of catching fly balls: Zeroing out optic
Neuroscience, 3, 860-862.
acceleration. Ecological Psychology, 4, 199-222.
Meegan, D.V., Glazebrook, C.M., Dhillon, V.P., Tremblay,
Michon, J.A. (1966). Tapping regularity as a measure of
L., Welsh, T.N., & Elliott, D. (2004). The Miiller-Lyer
perceptual motor load. Ergonomics, 9, 401-412.
illusion affects the planning and control of manual
aiming movements. Experimental Brain Research, Michon, J.A. (1967). Timing in temporal tracking.
155, 37-47. Soesterberg, The Netherlands: Institute for
Meegan, D.V., & Tipper, S.P. (1998). Reaching into Perception, RNO-TNO.
cluttered visual environments: Spatial and temporal Miller, G.A. (1956). The magical number seven, plus or
influences of distracting objects. Quarterly Journal of minus two: Some limits on our capacity for processing
Experimental Psychology, 51, 225-249. information. Psychological Review, 63, 81-97.
Meijer, O.G. (2001). Making things happen: An introduc- Miller, G.A. (2003). The cognitive revolution: A historical
tion to the history of movement science. In M.L. Latash perspective. Trends in Cognitive Sciences, 7, 141-144.
538 References

Miller, G.A., Galanter, E., & Pribram, K.H. (1960). Plans Morris, C.D., Bransford, J.D., & Franks, J.J. (1977). Levels
and the structure of behavior. New York: Holt, of processing versus transfer appropriate processing.
Rinehart & Winston. Journal of Verbal Learning and Verbal Behavior, 16,
Milner, A.D., & Goodale, M.A. (1993). Visual pathways to 519-533.
perception and action. In T.P. Hicks, S. Molotchnikoff, Morris, M.E., Summers, J.J., Matyas, T.A., & Iansek, R.
& T. Ono (Eds.), Progress in brain research (Vol. 95). (1994). Current status of the motor program. Physical
Amsterdam: Elsevier. Therapy, 74, 738-748.
Milner, A.D., & Goodale, M.A. (1995). The visual brain Morrissey, M.C., Harman, E.A., & Johnson, M.J.
in action. Oxford: Oxford University Press. (1995). Resistance training modes: Specificity and
Milner, A.D., & Goodale, M.A. (2008). Two visual effectiveness. Medicine and Science in Sports and
systems re-viewed. Neuropsychologia, 46, 774-785. Exercise, 27, 648-660.

Milner, B., Corkin, S., & Teuber, H.L. (1968). Further Most, S.B., Scholl, B.J., Clifford, E.R., & Simons, D.J.
analysis of the hippocampal amnesic syndrome: 14 (2005). What you see is what you set: Sustained
year follow up study of H.M. Neuropsychologica, 6, inattentional blindness and the capture of awareness.
215-234. Psychological Review, 112, 217-242.
Movahedi, A., Sheikh, M., Bagherzadeh, F.,
Milone, F. (1971). Interference in motor short-term
Hemayattalab, R., & Ashayeri, H. (2007). A practice-
memory. Unpublished master's thesis, Pennsylvania
specificity-based model of arousal for achieving peak
State University, University Park.
performance. Journal of Motor Behavior, 39, 457-462.
Minas, S.C. (1978). Mental practice of a complex
Mowbray, C.H. (1960). Choice reaction times for skilled
perceptual-motor skill. Journal of Human Movement
responses. Quarterly Journal of Experimental
Studies, 4, 102-107.
Psychology, 12, 193-202.
Minas, S.C. (1980). Acquisition of a motor skill following
Mowbray, C.H., & Rhoades, M.V. (1959). On the
guided mental and physical practice. Journal of
reduction of choice reaction times with practice.
Human Movement Studies, 6, 127-141.
Quarterly Journal of Experimental Psychology, 11,
Mohler, B.J., Thompson, W.B., Creem-Regehr, S.H., Pick, 16-23.
H.L., & Warren, W.H. (2007). Visual flow influences
Mowrer, O.H. (1940). Preparatory set (expectancy):
gait transition speed and preferred walking speed.
Some methods of measurement. Psychological
Experimental Brain Research, 181, 221-228.
Monographs, 52, No. 233.
Monno, A., Temprado, J.J., Zanone, P.-G., & Laurent,
Mulder, T. (2007). Motor imagery and action observation:
M. (2002). The interplay of attention and bimanual
Cognitive tools for rehabilitation. Journal of Neural
coordination dynamics. Acta Psychologica, 110, 187-
Transmission, 114, 1265-1278.
211.
Mulder, T., & Hulstijn, W. (1985). Delayed sensory
Montagne, G., & Laurent, M. (1994). The effects of
feedback in the learning of a novel motor task.
environment changes on one-handed catching.
Psychological Research, 47, 203-209.
Journal of Motor Behavior, 26, 237-246.
Muller, S., Abernethy, B., Eid, M., McBean, R., &
Mon-Williams, M., Tresilian, J.R., Coppard, V.L., &
Rose, M. (2010). Expertise and the spatio-temporal
Carson, R.G. (2001). The effect of obstacle position
characteristics of anticipatory information pick-up
on reach-to-grasp movements. Experimental Brain
from complex movement patterns. Perception, 30,
Research, 137, 497-501.
745-760.
Moors, A., & De Houwer, J. (2006). Automaticity: A Muller, S., Abernethy, B., & Farrow, D. (2006). How do
theoretical and conceptual analysis. Psychological world-class cricket batsmen anticipate a bowler's
Bulletin, 132, 297-326.
intention? Quarterly Journal of Experimental
Moray, N. (1970). Attention: Selective processes in vision Psychology, 59, 2162-2186.
and hearing. New York: Academic Press. Munoz, D.P., & Everling, S. (2004). Look away: The
Moreno, F.J., Avila, F., Damas, J., Garcia, J.A., Luis, V., anti-saccade task and the voluntary control of eye
Reina, R., & Ruiz, A. (2003). Contextual interference movement. Nature Reviews: Neuroscience, 5, 218-
in learning precision skills. Perceptual and Motor 228.
Skills, 97, 121-128. Murata, A. (1999). Extending effective target width
Morgan, C.T., & King, RA. (1971). Introduction to in Fitts' law to a two-dimensional pointing task.
psychology (4th ed.). New York: McGraw-Hill. International Journal of Human-Computer
Morgan, M., Phillips, J.C., Bradshaw, J.L., Mattingley, Interaction, 8, 457-469.
J.B., Iansek, R., & Bradshaw, J.A. (1994). Age- Murata, A., & Iwase, H. (2001). Extending Fitts' law to a
related motor slowness: Simply strategic? Journal of three-dimensional pointing task. Human Movement
Gerontology: Medical Sciences, 49, M133-Ml39. Science, 20, 791-805.
 References 539

Murphy, H.H. (1916). Distribution of practice periods in Neumann, 0. (1996). Theories of attention. In 0.
learning. Journal of Educational Psychology, 7, 150- Neumann & A.F. Sanders (Eds.), Handbook of
162. perception and action. Vol. 3: Attention (pp. 389-446).
Musch, J., & Grondin, S. (2001). Unequal competition as San Diego: Academic Press.
an impediment to personal development. A review Neumann, 0., & Sanders, A.F. (Eds.). (1996). Handbook
of the relative age effect in sport. Developmental of perception and action. Vol. 3: Attention. San Diego:
Review, 21, 147-167. Academic Press.
Muybridge, E. (1887). Animal locomotion: An electro- Newell, A., & Rosenbloom, P.S. (1981). Mechanisms
photographic investigation of consecutive phases of of skill acquisition and the law of practice. In J.R.
animal movements. Philadelphia: Lippincott. Anderson (Ed.), Cognitive skills and their acquisition
Muybridge, E. (1979). Muybridge's complete human (pp. 1-55). Hillsdale, NJ: Erlbaum.
and animal locomotion: All 781 plates from the 1887 Newell, K.M. (1974). Knowledge of results and motor
Animal Locomotion. New York: Dover. learning. Journal of Motor Behavior, 6, 235-244.
Nacson, J., & Schmidt, R.A. (1971). The activity-set Newell, K.M. (1976a). More on absolute error, etc.
hypothesis for warm-up decrement. Journal of Motor Journal of Motor Behavior, 8, 139-142.
Behavior, 3, 1-15. Newell, K.M. (1976b). Motor learning without
Namikas, G., & Archer, E.J. (1960). Motor skill transfer knowledge of results through the development of a
as a function of intertask interval and pretransfer response recognition mechanism. Journal of Motor
task difficulty. Journal of Experimental Psychology, Behavior, 8, 209-217.
59, 109-112. Newell, K.M. (1980). The speed-accuracy paradox
Napier, J.R. (1956). The prehensile movements of the in movement control: Error of time and space. In
human hand. Journal of Bone and Joint Surgery, 38B, G.E. Stelmach (Ed.), Tutorials in motor behavior.
902-913. Amsterdam: Elsevier.
Nashner, L., & Berthoz, A. (1978). Visual contribution to Newell, K.M. (1981). Skill learning. In D.H. Holding
rapid motor responses during postural control. Brain (Ed.), Human skills. New York: Wiley.
Research, 150, 403-407. Newell, K.M. (1986). Constraints on the development
Nashner, L.M., & Woollacott, M. (1979). The organization of coordination. In M.G. Wade & H.T.A. Whiting
of rapid postural adjustments of standing humans: (Eds.), Motor development in children: Aspects
An experimental-conceptual model. In R.E. Talbott & of coordination and control (pp. 341-360). Boston:
D.R. Humphrey (Eds.), Posture and movement (pp. Nijhoff.
243-257). New York: Raven Press. Newell, K.M. (1991). Motor skill acquisition. Annual
Navon, D., & Gopher, D. (1979). On the economy of the Review of Psychology, 42, 213-237.
human processing system. Psychological Review, 86, Newell, K.M., & Carlton, L.G. (1985). On the relationship
214-255. between peak force and peak force variability in
Neda, Z., Ravasz, E., Brechet, Y., Vicsek, T., & Barabasi, isometric tasks. Journal of Motor Behavior, 17, 230-
A.-L. (2000). The sound of many hands clapping. 241.
Nature, 403, 849-850. Newell, K.M., & Carlton, L.G. (1988). Force variability
Neisser, U. (1967). Cognitive psychology. New York: in isometric responses. Journal of Experimental
Appleton-Century-Crofts. Psychology: Human Perception and Performance,
Neisser, U., & Becklen, R. (1975). Selective looking: 14, 37-44.
Attending to visually specified events. Cognitive Newell, K.M., Carlton, M.J., & Antoniou, A. (1990). The
Psychology, 7, 480-494. interaction of criterion and feedback information in
Nelson, T.O. (1985). Ebbinghaus's contribution to the learning a drawing task. Journal of Motor Behavior,
measurement of retention: Savings during relearning. 22, 536-552.
Journal of Experimental Psychology: Learning, Newell, K.M., Carlton, L.G., Carlton, M.J., & Halbert,
Memory, and Cognition, 11, 472-479. J.A. (1980). Velocity as a factor in movement timing
Neumann, E., & Ammons, R.B. (1957). Acquisition and accuracy. Journal of Motor Behavior, 12, 47-56.
long-term retention of a simple serial perceptual- Newell, K.M., Carlton, M.J., Fisher, A.T., & Rutter, B.G.
motor skill. Journal of Experimental Psychology, 53, (1989). Whole-part training strategies for learning
159-161. the response dynamics of microprocessor driven
Neumann, 0. (1987). Beyond capacity: A functional simulators. Acta Psychologica, 71, 197-216.
view of attention. In H. Heuer & A.F. Sanders (Eds.), Newell, K.M., Carlton, L.G., & Hancock, P.A. (1984).
Perspectives on perception and action (pp. 361-394). Kinetic analysis of response variability. Psychological
Hillsdale, NJ: Erlbaum. Bulletin, 96, 133-151.
540 111 References

Newell, K.M., Carlton, L.G., & Kim, S. (1994). Time actions of stretch reflex. Journal of Neurophysiology,
and space-time movement accuracy. Human 39, 119-142.
Performance, 7, 1-21. Nicholson, D.E., & Schmidt, R.A. (1991). Timing-task
Newell, K.M., Carlton, L.G., Kim, S., & Chung, C. (1993). duration determines post-response error-detection
Space-time accuracy of rapid movements. Journal of capabilities. Paper presented at NASPSPA annual
Motor Behavior, 25, 8-20. meeting, Monterey, CA.
Newell, K.M., Hoshizaki, L.E.F., Carlton, M.J., & Nickerson, R.S. (1973). Intersensory facilitation of
Halbert, J.A. (1979). Movement time and velocity as reaction time: Energy summation or preparation
determinants of movement timing accuracy. Journal enhancement? Psychological Review, 80, 489-509.
of Motor Behavior, 11, 49-58. Nickerson, R.S., Kalikow, D.N., & Stevens, K.N. (1976).
Newell, K.M., Liu, Y.-T., & Mayer-Kress, G. (2001). Computer-aided speech training for the deaf. Journal
Time scales in motor learning and development. of Speech and Hearing Disorders, 41, 120-132.
Psychological Review, 108, 57-82. Nideffer, R.M. (1976). The inner athlete: Mind plus
Newell, K.M., Liu, Y.-T., & Mayer-Kress, G. (2008). muscle for winning. New York: Crowell.
Landscapes beyond the HKB model. In A. Fuchs & Noble, C.E. (1968). The learning of psychomotor skills.
V.K. Jirsa (Eds.), Coordination: Neural, behavioral Annual Review of Psychology, 19, 203-250.
and social dynamics (pp. 27-44). Berlin: Springer.
Norman, D.A. (1969). Memory while shadowing. Quar-
Newell, K.M., Liu, Y.-T., & Mayer-Kress, G. (2009). Time terly Journal of Experimental Psychology, 21, 85-93.
scales, difficulty I skill duality, and the dynamics of
motor learning. In D. Sternad (Ed.), Progress in motor Norman, D.A. (1976). Memory and attention (2nd ed.).
control (pp. 457-476). Berlin: Springer. New York: Wiley.
Newell, K.M., & McDonald, P.V. (1992). Practice: A Norman, D.A. (1981). The categorization of action slips.
search for task solutions. In American Academy of Psychological Review, 88, 1-15.
Physical Education, Enhancing human performance Norman, D.A., & Bobrow, D.G. (1975). On data-
in sport: New concepts and developments (The limited and resource-limited processes. Cognitive
Academy Papers, No. 25, pp. 51-59). Champaign, IL: Psychology, 7, 44-64.
Human Kinetics. Norman, J. (2002). Two visual systems and two
Newell, K.M., Sparrow, W.A., & Quinn, J.T. Jr. (1985). theories of perception: An attempt to reconcile the
Kinetic information feedback for learning isometric constructivist and ecological approaches. Behavioral
tasks. Journal of Human Movement Studies, 11, 113- and Brain Sciences, 25, 73-144.
123. Obhi, S.S., Matkovich, S., & Gilbert, S.J. (2009).
Newell, K.M., & Vaillancourt, D.E. (2001a). Dimensional Modification of planned actions. Experimental Brain
change in motor learning. Human Movement Research, 192, 287-292.
Science, 20, 695-715. Ogden, G.D., Levine, J.M., & Eisner, E.J. (1979).
Newell, K.M., & Vaillancourt, D.E. (2001b ). Woodworth Measurement of workload by secondary tasks.
(1899): Movement variability and theories of Human Factors, 21, 529-548.
motor control. In M.L. Latash & V.M. Zatsiorsky Ojha, H.A., Kern, R.W., Lin, C.-H., & Winstein, C.J.
(Eds.), Classics in movement science (pp. 409-435). (2009). Age affects the attentional demands of stair
Champaign, IL: Human Kinetics. ambulation: Evidence from a dual-task approach.
Newell, K.M., & van Emmerik, R.E.A. (1989). The Physical Therapy, 89, 1080-1088.
acquisition of coordination: Preliminary analysis of Ollis, S., Button, C., & Fairweather, M. (2005). The
learning to write. Human Movement Science, 8, 17- influence of professional expertise and task
32. complexity upon the potency of the contextual
Newell, K.M., van Emmerik, R.E.A., & Sprague, R.L. interference effect. Acta Psychologica, 118, 229-244.
(1993). Stereotypy and variability. In K.M. Newell Onla-or, S., & Winstein, C.J. (2008). Determining the
& D.M. Corcos (Eds.), Variability and motor control optimal challenge point for motor skill learning in
(pp. 475-496). Champaign, IL: Human Kinetics. adults with moderately severe Parkinson's disease.
Newell, K.M., & Walter, C.B. (1981). Kinematic and Neurorehabilation and Neural Repair, 22, 385-395.
kinetic parameters as information feedback in motor Osgood, C.E. (1949). The similarity paradox in human
skill acquisition. Journal of Human Movement learning: A resolution. Psychological Review, 56, 132-
Studies, 7, 235-254. 143.
Nicklaus, J., & Bowden, K. (1974). Golf my way. New Osman, A., Kornblum, S., & Meyer, D.E. (1990). Does
York: Simon & Schuster. motor programming necessitate response execution?
Nichols, T.R., & Houk, J.C. (1976). Improvement of Journal of Experimental Psychology: Human Percep-
linearity and regulation of stiffness that results from tion and Performance, 16, 183-198.
 References 541

Osu, R., Hirai, S., Yoshioka, T., & Kawato, M. (2004). Partridge, L.D. (1979). Muscle properties: A problem
Random presentation enables subjects to adapt to two for the motor physiologist. In R.E. Talbott & D.R.
opposing forces on the hand. Nature Neuroscience, 7, Humphrey (Eds.), Posture and movement (pp. 189-
111-112. 229). New York: Raven Press.
Otto-de Haart, E.G., Carey, D.P., & Milne, A.B. (1999). Partridge, L.D. (1983). Neural control drives a muscle
More thoughts on perceiving and grasping the spring: A persisting yet limited motor theory.
Miiller-Lyer illusion. Neuropsychologia, 37, 1437- Experimental Brain Research Supplementum, 7, 280-
1444. 290.
Oudejans, R.R.D., Michaels, C.F., Bakker, F.C., & Partridge, L.D., & Benton, L.A. (1981). Muscle, the motor.
Davids, K. (1999). Shedding some light on catching In V.B. Brooks (Ed.), Handbook of physiology: Vol. 2.
in the dark: Perceptual mechanisms for catching fly Motor control (pp. 43-106). Bethesda, MD: American
balls. Journal of Experimental Psychology: Human Physiological Society.
Perception and Performance, 25, 531-542.
Pashler, H. (1993, January-February). Doing two things
Oudejans, R.R.D., & Pijpers, J.R. (2009). Training with at the same time. American Scientist, 81(1), 48-49.
anxiety has a positive effect on expert perceptual-
Pashler, H. (1994). Dual-task interference in simple tasks:
motor performance under pressure. Quarterly
Data and theory. Psychological Bulletin, 116, 220-244.
Journal of Experimental Psychology, 62, 1631-1647.
Pashler, H. (1999). The psychology of attention.
Oudejans, R.R.D., Verheijen, R., Bakker, F.C., Gerrits,
Cambridge, MA: MIT Press.
J.C., Steinbruckner, M., & Beek, P.J. (2000). Errors in
judging "offside" in football. Nature, 404, 33. Patla, A.E. (1989). In search of laws for the visual
control of locomotion: Some observations. Journal
Oude Nijhuis, L.B., Janssen, L., Bloem, B.R., vanDijk,
J.C., Gielen, S.C., Borm, G.F., & Overeem, S. (2007). of Experimental Psychology: Human Perception and
Performance, 15, 624-628.
Choice reaction times for human head rotations are
shortened by startling acoustic stimuli, irrespective Patla, A.E. (1997). Understanding the roles of vision in
of stimulus direction. Journal of Physiology, 584, 97- the control of human locomotion. Gait & Posture, 5,
109. 54-69.
Overduin, S.A., Richardson, A.G., Lane, C.E., Bizzi, E., Patla, A.E. (1998). How is human gait controlled by
& Press, D.Z. (2006). Intermittent practice facilitates vision? Ecological Psychology, 10, 287-302.
stable motor memories. Journal of Neuroscience, 26, Patla, A.E., Frank, J.S., Allard, F., & Thomas, E. (1985).
11888-11892. Speed-accuracy characteristics of saccadic eye
Paillard, J., & Bruchon, M. (1968). Active and passive movements. Journal of Motor Behavior, 17, 411-419.
movements in the calibration of position sense. In S.J. Patla, A.E., Rietdyk, S., Martin, C., & Prentice, S. (1996).
Freedman (Ed.), The neuropsychology of spatially Locomotor patterns of the leading and the trailing
oriented behavior. New York: Dorsey. limbs as solid and fragile obstacles are stepped
Papaxanthis, C., Schieppati, M., Gentili, R., & Pozzo, T. over: Some insights into the role of vision during
(2002). Imagined and actual arm movements have locomotion. Journal of Motor Behavior, 28, 35-47.
similar durations when performed under different
Patla, A.E., Robinson, C., Samways, M., & Armstrong,
conditions of direction and mass. Experimental Brain
C.J. (1989). Visual control of step length during
Research, 143, 447-452.
overground locomotion: Task-specific modulation
Park, J.-H., & Shea, C.H. (2002). Effector independence. of the locomotor synergy. Journal of Experimental
Journal of Motor Behavior, 34, 253-270. Psychology: Human Perception and Performance,
Park, J.-H., & Shea, C.H. (2003). Effect of practice on 15, 603-617.
effector independence. Journal of Motor Behavior, Patrick, J. (1971). The effect of interpolated motor
35, 33-40. activities in short-term motor memory. Journal of
Park, J.-H., & Shea, C.H. (2005). Sequence learning: Motor Behavior, 3, 39-48.
Response structure and effector transfer. Quarterly Patterson, J.T., & Wessel, J. (2002). Strategies for
Journal of Experimental Psychology, 58A, 387-419. retraining functional movement in persons with
Park, J.-H., Shea, C.H., &Wright, D.L. (2000). Reduced- Alzheimer disease: A review. Physiotherapy Canada,
frequency concurrent and terminal feedback: A test of 54, 274-280.
the guidance hypothesis. Journal of Motor Behavior, Paulignan, Y., & Jeannerod, M. (1996). Prehension
32, 287-296. movements: The visuomotor channels hypothesis
Parker, J.F. Jr., & Fleishman, E.A. (1960). Ability factors revisited. In A.M. Wing, P. Haggard, & J.R. Flanagan
and component performance measures as predictors (Eds.), Hand and brain: The neurophysiology of
of complex tracking behavior. Psychological psychology of hand movements (pp. 265-282). San
Monographs, 74 (Whole No. 503). Diego: Academic Press.
542 References

Paulignan, Y., Jeannerod, M., MacKenzie, C., & Marteniuk, Peters, M. (1985). Performance of a rubato-like task:
R. (1991). Selective perturbation of visual input during When two things cannot be done at the same time.
prehension movements. 2. The effects of changing Music Perception, 2, 471-482.
object size. Experimental Brain Research, 87, 407-420. Peterson, L.R., & Peterson, M.J. (1959). Short-term
Paulignan, Y., MacKenzie, C., Marteniuk, R., & Jeannerod, retention of individual verbal items. Journal of
M. (1991 ). Selective perturbation of visual input during Experimental Psychology, 58, 193-198.
prehension movements. 1. The effects of changing ob- Petrynski, W. (2003). R6ne rodzaje odpowiedzi
ject position. Experimental Brain Research, 83, 502- czucioworuchowych w schemacie Schmidta
512. [Various kinds of sensory-motor performances in the
Payne, V.G., & Isaacs, L.D. (2007). Human motor Schmidt's scheme]. Antropomotoryka, 13, 97-110.
development: A lifespan approach (7th ed.). Pew, R.W. (1966). Acquisition of hierarchical control
Columbus, OH: McGraw-Hill. over the temporal organization of a skill. Journal of
Pearson, K. (1976). The control of walking. Scientific Experimental Psychology, 71, 764-771.
American, 235, 72-86. Pew, R.W. (1970). Toward a process-oriented theory
Pearson, K., & Gordon, J. (2000a). Locomotion. In E.R. of human skilled performance. Journal of Motor
Kandel, J.H. Schwartz, & T.M. Jessell (Eds.), Principles Behavior, 2, 8-24.
of neural science (pp. 739-755). New York: McGraw- Pew, R.W. (1974a). Human perceptual-motor
Hill. performance. In B.H. Kantowitz (Ed.), Human
Pearson, K., & Gordon, J. (2000b). Spinal reflexes. In E.R. information processing: Tutorials in performance
Kandel, J.H. Schwartz, & T.M. Jessell (Eds.), Principles and cognition (pp. 1-39). Hillsdale, NJ: Erlbaum.
of neural science (pp. 713-736). New York: McGraw- Pew, R.W. (1974b). Levels of analysis in motor control.
Hill. Brain Research, 71, 393-400.
Pellecchia, G., Shockley, K., & Turvey, M.T. (2005). Phillips, J.R., & Berkhout, J. (1976). Uses of computer-
Concurrent cognitive task modulates coordination assisted instruction in developing psychomotor skills
dynamics. Cognitive Science, 29, 531-557. related to heavy machinery operation (Contract Rep.
Pelisson, D., Prablanc, C., Goodale, M.A., & Jeannerod, DAHC-19-75G0009). Alexandria, VA: U.S. Army
M. (1986). Visual control of reaching movements Research Institute.
without vision of the limb II. Evidence of fast Piek, J. (2006). Infant motor development. Champaign,
unconscious processes correcting the trajectory of the IL: Human Kinetics.
hand to the final position of double-step stimulus. Pigott, R.E., & Shapiro, D.C. (1984). Motor schema:
Experimental Brain Research, 62, 303-311. The structure of the variability session. Research
Peper, C.E., Ridderikhoff, A., Daffertshofer, A., & Beek, Quarterly for Exercise and Sport, 55, 41-45.
P.J. (2004). Explanatory limitations of the HKB model: Plamondon, R., & Alimi, A.M. (1997). Speed/accuracy
Incentives for a two-tiered model of rhythmic interlimb tradeoffs in target-directed movements. Behavioral
coordination. Human Movement Science, 23, 673-697. and Brain Sciences, 20, 279-349.
Pepper, R.L., & Herman, L.M. (1970). Decay and Playford, E.D., Siegert, R., Levack, W., & Freeman, J.
interference effects in the short-term retention (2009). Areas of consensus and controversy about
of a discrete motor act. Journal of Experimental goal setting in rehabilitation: A conference report.
Psychology, 83 (Monograph Suppl. 2). Clinical Rehabilitation, 23, 334-344.
Perel, M. (1976). Analyzing the role of driver/vehicle Pohl, P.S., Winstein, C.J., & Fisher, B.E. (1996). The
incompatibilities in accident causation using locus of age-related movement slowing: Sensory
police reports (Tech. Rep. No. DOT H5-806-509). processing in continuous goal-directed aiming.
Washington, DC: U.S. Department of Transportation. Journal of Gerontology: Psychological Sciences, 51B,
Perenin, M.-T., & Vighetto, A. (1988). Optic ataxia: A P94-Pl02.
specific disruption in visuomotor mechanisms I. Polanyi, M. (1958). Personal knowledge: Towards a
Different aspects of the deficit in reaching for objects. post-critical philosophy. London: Routledge and
Brain, 111, 643-674. Kegan Paul.
Perkins, G.D. (2007). Simulation in resuscitation training. Polit, A., & Bizzi, E. (1978). Processes controlling arm
Resuscitation, 73, 202-211. movements in monkeys. Science, 201, 1235-1237.
Perkins-Ceccato, N., Passmore, S.R., & Lee, T.D. (2003). Polit, A., & Bizzi, E. (1979). Characteristics of motor
Effects of focus of attention depend on golfers' skill. programs underlying arm movements in monkeys.
Journal of Sports Sciences, 21, 593-600. Journal of Neurophysiology, 42, 183-194.
Peters, M. (1977). Simultaneous performance of two Pollard, R. (1986). Home advantage in soccer: A
motor activities: The factor of timing. Neuropsy- retrospective analysis. Journal of Sports Sciences, 4,
chologica, 15, 461-464. 237-248.
 References 543

Pollock, B.J., & Lee, T.D. (1992). Effects of the model's feedback in rapid aimed limb movements. Journal of
skill level on observational motor learning. Research Motor Behavior, 28, 149-156.
Quarterly for Exercise and Sport, 63, 25-29. Pratt, J., Adam, J.J., & Fischer, M.H. (2007). Visual layout
Pollock, B.J., & Lee, T.D. (1997). Dissociated contextual modulates Fitts's law: The importance of first and
interference effects in children and adults. Perceptual last positions. Psychonomic Bulletin & Review, 14,
and Motor Skills, 84, 851-858. 350-355.
Pons, T.P., Garraghty, P.E., Ommaya, A.K., Kaas, J.H., Proctor, R.W., & Dutta, A. (1995). Skill acquisition and
Taub, E., & Mishkin, M. (1991). Massive cortical human performance. Thousand Oaks, CA: Sage.
reorganization after sensory deafferentation in adult Proctor, R.W., & Reeve, T.G. (Eds.). (1990). Stimulus-
macaques. Science, 252, 1857-1860.
response compatibility: An integrated perspective.
Populin, L., Rose, D.J., & Heath, K. (1990). The role of Amsterdam: Elsevier.
attention in one-handed catching. Journal of Motor Proctor, R.W., & Van Zandt, T. (1994). Human factors in
Behavior, 22, 149-158.
simple and complex systems. Boston: Allyn & Bacon.
Posner, M.I. (1969). Reduced attention and the
Prodoehl, J., Corcos, D.M., & Vaillancourt, D.E. (2009).
performance of "automated" movements. Journal of
Basal ganglia mechanisms underlying precision
Motor Behavior, 1, 245-258.
grip force control. Neuroscience & Biobehavioral
Posner, M.I. (1978). Chronometric explorations of mind. Reviews, 33, 900-908.
Hillsdale, NJ: Erlbaum.
Proteau, L. (1992). On the specificity of learning and the
Posner, M.I., & Cohen, Y. (1984). Components of visual role of visual information for movement control. In L.
orienting. In H. Bouma & D.G. Bouwhuis (Eds.), Proteau & D. Elliott (Eds.), Vision and motor control
Attention and performance X: Control of language (pp. 67-103). Amsterdam: Elsevier.
processes (pp. 531-556). Hillsdale, NJ: Erlbaum.
Proteau, L. (1995). Sensory integration in the learning
Posner, M.I., & Keele, S.W. (1969). Attentional demands of an aiming task. Canadian Journal of Experimental
of movement. Proceedings of the 16th Congress Psychology, 49, 113-120.
of Applied Psychology. Amsterdam: Swets and
Proteau, L. (2005). Visual afferent information dominates
Zeitlinger.
other sources of afferent information during mixed
Posner, M.I., & Konick, A.F. (1966). On the role of practice of a video-aiming task. Experimental Brain
interference in short-term retention. Journal of Research, 161, 441-456.
Experimental Psychology, 72, 221-231.
Proteau, L., Blandin, Y., Alain, C., & Dorion, A. (1994).
Posner, M.I., Nissen, M.J., & Ogden, W.C. (1978). The effects of the amount and variability of practice
Attended and unattended processing modes: The on the learning of a multi-segmented motor task.
role of set for spatial location. In H.L. Pick & I.J. Acta Psychologica, 85, 61-74.
Saltzman (Eds.), Modes of perceiving and processing
Proteau, L., & Isabelle, G. (2002). On the role of visual
information (pp. 137-157). Hillsdale, NJ: Erlbaum.
afferent information for the control of aiming
Posner, M.I., & Petersen, S.E. (1990). The attention system movements toward targets of different sizes. Journal
of the human brain. Annual Review of Neuroscience, of Motor Behavior, 34, 367-384.
13, 25-42.
Proteau, L., Marteniuk, R.G., & Levesque, L. (1992). A
Posner, M.I., & Snyder, C.R. (1975). Attention and sensorimotor basis for motor learning: Evidence
cognitive control. In R.L. Solso (Ed.), Information indicating specificity of practice. Quarterly Journal of
processing and cognition. Hillsdale, NJ: Erlbaum. Experimental Psychology, 44A, 557-575.
Poulton, E.C. (1950). Perceptual anticipation and reaction Provins, K.A. (1958). The effect of peripheral nerve
time. Quarterly Journal of Experimental Psychology, block on the appreciation and execution of finger
2, 99-112. movements. Journal of Physiology, 143, 55-67.
Poulton, E.C. (1957). On prediction in skilled movements. Purdy, W.C. (1958). The hypothesis of psychophysical
Psychological Bulletin, 54, 467-478. correspondence in space perception. PhD thesis,
Poulton, E.C. (1973). The effect of fatigue upon inspection Cornell University, Ithaca, NY.
work. Applied Ergonomics, 4, 73-83. Pyle, W.H. (1919). Transfer and interference in card-
Poulton, E.C. (1974). Tracking skill and manual control. distributing. Journal of Educational Psychology, 10,
New York: Academic Press. 107-110.
Pratt, J., & Abrams, R.A. (1994). Action-centered inhibi- Quesada, D.C., & Schmidt, R.A. (1970). A test of the
tion: Effects of distractors on movement planning and Adams-Creamer decay hypothesis for the timing of
execution. Human Movement Science, 13, 245-254. motor responses. Journal of Motor Behavior, 2, 273-283.
Pratt, J., & Abrams, R.A. (1996). Practice and component Quinn, J.T. Jr., Schmidt, R.A., Zelaznik, H.N., Hawkins,
submovements: The roles of programming and B., & McFarquhar, R. (1980). Target-size influences
544 References

on reaction time with movement time controlled. Rensink, R.A. (2002). Change detection. Annual Review
Journal of Motor Behavior, 12, 239-261. of Psychology, 53, 245-277.
Quinn, J.T. Jr., & Sherwood, D.E. (1983). Time Repp, B.H. (2001). Phase correction, phase resetting, and
requirements of changes in program and parameter phase shifts after subliminal timing perturbations
variables in rapid ongoing movements. Journal of in sensorimotor synchronization. Journal of
Motor Behavior, 15, 163-178. Experimental Psychology: Human Perception and
Rack, P.M.H., & Westbury, D.R. (1969). The effects of Performance, 27, 600-621.
length and stimulus rate on tension in the isometric Repp, B.H. (2005). Sensorimotor synchronization:
cat soleus muscle. Journal of Physiology, 204, 443-460. A review of the tapping literature. Psychonomic
Radlo, S.J., Janelle, C.M., Barba, D.A., & Frehlich, S.G. Bulletin & Review, 12, 969-992.
(2001). Perceptual decision-making for baseball pitch Reynolds, B., & Bilodeau, l.M. (1952). Acquisition and
recognition: Using P300 latency and amplitude to retention of three psychomotor tests as a function of
index attentional processing. Research Quarterly for distribution of practice during acquisition. Journal of
Exercise and Sport, 72, 22-31. Experimental Psychology, 44, 19-26.
Raibert, M.H. (1977). Motor control and learning by Richardson, M.J., Marsh, K.L., Isenhower, R.W.,
the state-space model (Tech. Rep. No. AI-TR-439). Goodman, J.R.L., & Schmidt, R.C. (2007). Rocking
Cambridge: Massachusetts Institute of Technology, together: Dynamics of intentional and unintentional
Artificial Intelligence Laboratory. interpersonal coordination. Human Movement
Ranganathan, R., & Newell, K.M. (2009). Influence of Science, 26, 867-891.
augmented feedback on coordination strategies. Richardson-Klavehn, A., & Bjork, R.A. (1988). Measures
Journal of Motor Behavior, 41, 317-330. of memory. Annual Review of Psychology, 39, 475-
Raynor, A.J., Yi, C.J., Abernethy, B., & Jong, Q.J. (2002). 543.
Are transitions in human gait determined by Richter-Heinrich, E., & Miller, N. (Eds.). (1982).
mechanical, kinetic or energetic factors? Human Biofeedback-basic problems and clinical
Movement Science, 21, 785-805. applications. Amsterdam: Elsevier.
Reason, J. (1990). Human error. Cambridge: Cambridge Riek, S., Tresilian, J.R., Mon-Williams, M., Coppard,
University Press. V.L., & Carson, R.G. (2003). Bimanual aiming and
Reason, J., & Mycielska, K. (1982). Absent-minded? The overt attention: One law for two hands. Experimental
psychology of mental lapses and everyday errors. Brain Research, 153, 59-75.
Englewood Cliffs, NJ: Prentice Hall. Rizzolatti, G., & Craighero, L. (2004). The mirror-neuron
Redelmeier, D.A., & Tibshirani, R.J. (1997). Association system. Annual Review of Neuroscience, 27, 169-192.
between cellular-telephone calls and motor vehicle colli- Rizzolatti, G., & Fabbri-Destro, M. (2010). Mirror
sions. New England Journal of Medicine, 336, 453-458. neurons: From discovery to autism. Experimental
Reed, E.S. (1988). Applying the theory of action systems Brain Research, 200, 223-237.
to the study of motor skills. In O.G. Meijer & K. Roth Roberts, J.R., Jones, R., Mansfield, N.J., & Rothberg, S.J.
(Eds.), Complex movement behaviour: "The" motor- (2005). Evaluation of impact sounds on the "feel" of
action controversy (pp. 45-86). Amsterdam: Elsevier. a golf shot. Journal of Sound and Vibration, 287, 651-
Reed, N., McLeod, P., & Dienes, Z. (2010). Implicit 666.
knowledge and motor skill: What people who know Robertson, S.D., Zelaznik, H.N., Lantern, D.A., Bojczyk,
how to catch don't know. Consciousness and Cogni- K.G., Spencer, R.M., Doffin, J.C., & Schneidt, T. (1999).
tion, 19, 63-76. Correlations for timing consistency among tapping
Reeve, T.G., Dornier, L.A., & Weeks, D.J. (1990). Precision and drawing tasks: Evidence against a single timing
of knowledge of results: Consideration of the process for motor control. Journal of Experimental
accuracy requirements imposed by the task. Research Psychology: Human Perception and Performance,
Quarterly for Exercise and Sport, 61, 284-290. 25, 1316-1330.
Reinhart, W. (1994). The effect of countermeasures to Robinson, C.H., & Kavinsky, R.C. (1976). On Fitts' law
reduce the incidence of unintended acceleration with two-handed movement. IEEE Transactions on
accidents (Paper No. 94 SS 0 07). Washington, DC: Systems, Man, and Cybernetics, 6, 504-505.
NHTSA, U.S. Department of Transportation. Rock, P.B., & Harris, M.G. (2006). Tau-dot as a potential
Reinkensmeyer, D.J., Emken, J., & Cramer, S. (2004). control variable for visually guided braking. Journal
Robotics, motor learning, and neurologic recovery. of Experimental Psychology: Human Perception and
Annual Review of Biomedical Engineering, 6, 497-525. Performance, 32, 251-267.
Reinkensmeyer, D.J., & Patton, J.L. (2009). Can robots Roediger, H.L. (1990). Implicit memory: Retention
help the learning of skilled actions? Exercise and without remembering. American Psychologist, 45,
Sport Sciences Reviews, 37, 43-51. 1043-1056.
 References 545

Roediger, H.L. (Ed.). (2008a). Cognitive psychology In M. Jeannerod (Ed.), Attention and performance
of memory (Vol. 2 of Learning and memory: A XIII (pp. 321-342). Hillsdale, NJ: Erlbaum.
comprehensive reference). Oxford: Elsevier. Rosenbaum, D.A., Meulenbroek, R.G.J., Vaughan, J., &
Roediger, H.L. (2008b). Relativity of remembering: Why Jansen, C. (2001). Posture-based motion planning:
the laws of memory vanished. Annual Review of Applications to grasping. Psychological Review, 108,
Psychology, 59, 225-254. 709-734.
Roediger, H.L., & McDermott, K.B. (1993). Implicit Rosenbaum, D.A., & Patashnik, 0. (1980). Time to
memory in normal human subjects. In H. Spinnler & time in the human motor system. In R.S. Nickerson
F. Boller (Eds.), Handbook of neuropsychology (Vol. (Ed.), Attention and performance VIII (pp. 93-106).
8) (pp. 63-131). Amsterdam: Elsevier. Hillsdale, NJ: Erlbaum.
Rogers, C.A. Jr. (1974). Feedback precision and Rosenbaum, D.A., van Heugten, C.M., & Caldwell, G.E.
postfeedback interval duration. Journal of (1996). From cognition to biomechanics and back:
Experimental Psychology, 102, 604-608. The end-state comfort effect and the middle-is-faster
Rogers, D.K., Bendrups, A.P., & Lewis, M.M. (1985). effect. Acta Psychologica, 94, 59-85.
Disturbed proprioception following a period of muscle Rosenbaum, D.A., Vaughan, J., Barnes, H.J., & Jorgensen,
vibration in humans. Neuroscience Letters, 57, 147-152. M.J. (1992). Time course of movement planning:
Roller, C.A., Cohen, H.S., Kimball, K.T., & Bloomberg, J.J. Selection of handgrips for object manipulation.
(2001). Variable practice with lenses improves visuo- Journal of Experimental Psychology: Learning,
motor plasticity. Cognitive Brain Research, 12, 341-352. Memory, and Cognition, 18, 1058-1073.
Romaiguere, P., Hasbroucq, T., Possamai:, C., & Seal, J. Rosenbaum, D.A., Weber, R.J., Hazelett, W.M., &
(1993). Intensity to force translation: A new effect of Hindorff, V. (1986). The parameter remapping effect
stimulus-response compatibility revealed by analysis in human performance: Evidence from tongue
of response time and electromyographic activity of a twisters and finger fumblers. Journal of Memory and
prime mover. Cognitive Brain Research, 1, 197-201. Language, 25, 710-725.
Ronsse, R., Miall, R.C., & Swinnen, S.P. (2009). Ross, D., Bird, A.M., Doody, S.G., & Zoeller, M. (1985).
Multisensory integration in dynamical behaviors: Effects of modeling and videotape feedback with
Maximum likelihood estimation across bimanual knowledge of results on motor performance. Human
skill learning. Journal of Neuroscience, 29, 8419-8428. Movement Science, 4, 149-157.
Rose, D.J., & Christina, R.W. (1990). Attention demands Rossignol, S., Dubuc, R., & Gossard, J.P. (2006).
of precision pistol-shooting as a function of skill Dynamic sensorimotor interactions in locomotion.
level. Research Quarterly for Exercise and Sport, 61, Physiological Review, 86, 89-154.
111-113. Roth, K. (1988). Investigations on the basis of the
Rosenbaum, D.A. (1980). Human movement initiation: generalized motor programme hypothesis. In O.G.
Specification of arm, direction, and extent. Journal of Meijer & K. Roth (Eds.), Complex motor behavior:
Experimental Psychology: General, 109, 444-474. "The" motor-action controversy (pp. 261-288).
Amsterdam: Elsevier.
Rosenbaum, D.A. (1983). The movement precuing
technique: Assumptions, applications, and Rothstein, A.L. (1973). Effect of temporal expectancy of
extensions. In R.A. Magill (Ed.), Memory and control the position of a selected foreperiod within a range.
of action (pp. 231-274). Amsterdam: Elsevier. Research Quarterly, 44, 132-139.
Rosenbaum, D.A. (2009). Human motor control (2nd Rothstein, A.L., & Arnold, R.K. (1976). Bridging the gap:
ed.). San Diego: Academic Press. Application of research on videotape feedback and
Rosenbaum, D.A. (2005). The Cinderella of psychology: bowling. Motor Skills: Theory Into Practice, 1, 35-62.
The neglect of motor control in the science of mental Rothwell, J.C., Traub, M.M., Day, B.L., Obeso, J.A.,
life and behavior. American Psychologist, 60, 308-317. Thomas, P.K., & Marsden, C.D. (1982). Manual motor
Rosenbaum, D.A., Cohen, R.G., Jax, S.A., Weiss, D.J., & performance in a deafferented man. Brain, 105, 515-
van der Wel, R. (2007). The problem of serial order 542.
in behavior: Lashley's legacy. Human Movement Rubin, W.M. (1978). Application of signal detection
Science, 26, 525-554. theory to error detection in ballistic motor skills.
Rosenbaum, D.A., Loukopoulos, L.D., Meulenbroek, Journal of Experimental Psychology: Human
R.G.J., Vaughan, J., & Engelbrecht, S.E. (1995). Perception and Performance, 4, 311-320.
Planning reaches by evaluating stored postures. Ryan, E.D. (1962). Retention of stabilometer and pursuit
Psychological Review, 102, 28-67. rotor skills. Research Quarterly, 33, 593-598.
Rosenbaum, D.A., Marchak, F., Barnes, H.J., Vaughan, J., Ryan, E.D. (1965). Retention of stabilometer performance
Slotta, J.D., & Jorgensen, M.J. (1990). Constraints for over extended periods of time. Research Quarterly,
action selection: Overhand versus underhand grips. 36, 46-51.
546 References

Ryan, L.J., & Fritz, M.S. (2007). Erroneous knowledge Savelsbergh, G., & Davids, K. (2002). "Keeping the eye
of results affects decision and memory processes on on the ball": The legacy of John Whiting (1929-2001)
timing tasks. Journal of Experimental Psychology: in sport science. Journal of Sports Sciences, 20, 79-82.
Human Perception and Performance, 33, 1468-1482. Savelsbergh, G.J.P., van der Kamp, J., Williams, A.M.,
Ryan, L.J., & Robey, T.B. (2002). Learning and & Ward, P. (2005). Anticipation and visual search
performance effects of accurate and erroneous behaviour in expert soccer goalkeepers. Ergonomics,
knowledge of results on time perception. Acta 48, 1686-1697.
Psychologica, 111, 83-100. Savelsbergh, G.J.P., & Whiting, H.T.A. (1996). Catching:
Sack, A.T., & Linden, D.E.J. (2003). Combining A motor learning and developmental perspective.
transcranial magnetic stimulation and functional In H. Heuer & S.W. Keele (Eds.), Handbook of
imaging in cognitive brain research: Possibilities and perception and action. Vol. 2: Motor skills (pp. 461-
limitations. Brain Research Reviews, 43, 41-56. 501). San Diego: Academic Press.
Sainburg, R.L. (2005). Handedness: Differential Savelsbergh, G.J.P., Whiting, H.T.A., & Bootsma, R.J.
specializations for control of trajectory and position. (1991). Grasping tau. Journal of Experimental
Exercise and Sport Sciences Reviews, 33, 206-213. Psychology: Human Perception and Performance,
Sale, D., & MacDougall, D. (1981). Specificity in strength 17, 315-322.
training: A review for the coach and athlete. Canadian Savelsbergh, G.J.P., Whiting, H.T.A., & Pijpers, J.R.
Journal of Applied Sport Sciences, 6, 87-92. (1992). The control of catching. In J.J. Summers
Salesse, R., Oullier, 0., & Temprado, J.J. (2005). Plane (Ed.), Approaches to the study of motor control and
of motion mediates the coalition of constraints in learning (pp. 313-342). Amsterdam: Elsevier.
rhythmic bimanual coordination. Journal of Motor Savelsbergh, G.J.P., Whiting, H.T.A., Pijpers, J.R., & van
Behavior, 37, 454-464. Santvoord, A.A.M. (1993). The visual guidance of
Salmela, J.H., & Fiorito, P. (1979). Visual cues in ice catching. Experimental Brain Research, 93, 148-156.
hockey goaltending. Canadian Journal of Applied Schaal, S., Sternad, D., Osu, R., & Kawato, M. (2004).
Sport Science, 4, 56-59. Rhythmic arm movement is not discrete. Nature
Salmoni, A.W., Ross, D., Dill, S., & Zoeller, M. (1983). Neuroscience, 1, 1137-1144.
Knowledge of results and perceptual-motor learning. Schacter, D.L. (1983). Amnesia observed: Remembering
Human Movement Science, 2, 77-89. and forgetting in a natural environment. Journal of
Salmoni, A.W., Schmidt, R.A., & Walter, C.B. (1984). Abnormal Psychology, 92, 236-242.
Knowledge of results and motor learning: A review Schacter, D.L. (1987). Implicit memory: History and
and critical reappraisal. Psychological Bulletin, 95, current status. Journal of Experimental Psychology:
355-386. Learning, Memory, and Cognition, 13, 501-518.
Salmoni, AW., Sullivan, S.J., & Starkes, J.L. (1976). The Schellekens, J.M.H., Kalverboer, A.F., & Scholten, C.A.
attention demands of movements: A critique of the (1984). The micro-structure of tapping movements in
probe technique. Journal of Motor Behavior, 8, 161-169. children. Journal of Motor Behavior, 16, 20-39.
Salthouse, T.A. (1984). Effects of age and skill in typing. Schendel, J.D., & Hagman, J.D. (1982). On sustaining
Journal of Experimental Psychology: General, 113, procedural skills over a prolonged retention interval.
345-371. Journal of Applied Psychology, 67, 605-610.
Salthouse, T.A. (1985). A theory of cognitive aging. Schendel, J.D., Heller, F.H., Finley, D.L., & Hawley, J.K.
Amsterdam: Elsevier. (1985). Use of the Weaponeer Marksmanship Trainer
Salthouse, T.A. (1986). Effects of practice on a typing-like in predicting M16Al rifle qualification performance.
keying task. Acta Psychologica, 62, 189-198. Human Factors, 27, 313-325.
Sanders, A.F. (1980). Stage analysis of reaction processes. Schendel, J.D., & Newell, K.M. (1976). On processing the
In G.E. Stelmach & J. Requin (Eds.), Tutorials in motor information from knowledge of results. Journal of
behavior (pp. 331-354). Amsterdam: Elsevier. Motor Behavior, 8, 251-255.
Sanders, M.S., & McCormick, E.J. (1993). Human factors Schlager, N. (Ed.). (1994). When technology fails:
in engineering and design (7th ed.). New York: Significant technological disasters, accidents and
McGraw-Hill. failures of the twentieth century. Detroit: Gale
Sanes, J.N. (1990). Motor representations in deafferented Research.
humans: A mechanism for disordered movement Schmidt, R.A. (1967). Motor factors in coincident timing.
performance. In M. Jeannerod (Ed.), Attention and Unpublished doctoral dissertation, University of
performance XIII (pp. 714-735). Hillsdale, NJ: Erlbaum. Illinois, Urbana.
Sanes, J.N., Mauritz, K.-H., Dalakas, M.C., & Evarts, Schmidt, R.A. (1968). Anticipation and timing in human
E.V. (1985). Motor control in humans with large-fiber motor performance. Psychological Bulletin, 70, 631-
sensory neuropathy. Human Neurobiology, 4, 101-114. 646.
 References 547

Schmidt, R.A. (1969a). Intra-limb specificity of motor contributions of the task approach. In J.S. Skinner, C.B.
response consistency. Journal of Motor Behavior, 1, Corbin, D.M. Landers, P.E. Martin, & C.L. Wells (Eds.),
89-99. Future directions in exercise and sport science research
Schmidt, R.A. (1969b ). Movementtime as a determiner of (pp. 395-410). Champaign, IL: Human Kinetics.
timing accuracy. Journal of Experimental Psychology, Schmidt, R.A. (1989b). Unintended acceleration: A
79, 43-47. review of human factors contributions. Human
Schmidt, R.A. (1969c). Consistency of response Factors, 31, 345-364.
components as a function of selected motor variables. Schmidt, R.A. (1990). Distinguished scholar award to
Research Quarterly, 40, 561-566. Jack Ashton Adams. NASPSPA Newsletter, 15, 4-5.
Schmidt, R.A. (1971a). Proprioception and the timing of Schmidt, R.A. (1991a). Frequent augmented feedback
motor responses. Psychological Bulletin, 76, 383-393. can degrade learning: Evidence and interpretations.
Schmidt, R.A. (1971b). Retroactive interference and In J. Requin & G.E. Stelmach (Eds.), Tutorials in
amount of original learning in verbal and motor motor neuroscience (pp. 59-75). Dordrecht: Kluwer.
tasks. Research Quarterly, 42, 314-326. Schmidt, R.A. (1991b). Motor learning principles for
Schmidt, R.A. (1972a). The case against learning and physical therapy. In M.J. Lister (Ed.), Contemporary
forgetting scores. Journal of Motor Behavior, 4, 79-88. management of motor control problems: Proceedings
Schmidt, R.A. (1972b). The index of preprogramming of the II step conference (pp. 49-63). Alexandria:
(IP): A statistical method for evaluating the role Foundation for Physical Therapy.
of feedback in simple movements. Psychonomic Schmidt, R.A. (1993). Unintended acceleration: Human
Science, 27, 83-85. performance considerations. In B. Peacock & W.
Schmidt, R.A. (1975a). Motor skills. New York: Harper Karwowski (Eds.), Automotive ergonomics (pp. 431-
&Row. 451). London: Taylor & Francis.
Schmidt, R.A. (l 975b). A schema theory of discrete motor Schmidt, R.A. (1994). Movement time, movement
skill learning. Psychological Review, 82, 225-260. distance, and movement accuracy: A reply to Newell,
Carlton, and Kim. Human Performance, 7, 23-28.
Schmidt, R.A. (1976a). Control processes in motor skills.
Exercise and Sport Sciences Reviews, 4, 229-261. Schmidt, R.A. (2003). Motor schema theory after 27
years: Reflections and implications for a new theory.
Schmidt, R.A. (1976b). Movement education and the
Research Quarterly for Exercise and Sport, 74, 366-375.
schema theory. In E. Crawford (Ed.), Report of the
1976 conference June 3-8. Cedar Falls, IA: National Schmidt, R.A., & Bjork, R.A. (1992). New concep-
Association for Physical Education of College Women. tualizations of practice: Common principles in
Schmidt, R.A. (1977). Schema theory: Implications for three paradigms suggest new concepts for training.
movement education. Motor Skills: Theory Into Psychological Science, 3, 207-217.
Practice, 2, 36-38. Schmidt, R.A., Christenson, R., & Rogers, P. (1975).
Schmidt, R.A. (1980). Past and future issues in motor Some evidence for the independence of recall and
programming. Research Quarterly for Exercise and recognition in motor behavior. In D.M. Landers, D.V.
Sport, 51, 122-140. Harris, & R.W. Christina (Eds.), Psychology of motor
behavior and sport II. State College, PA: Penn State
Schmidt, R.A. (1983). On the underlying structure HPER Series.
of well-learned motor responses: A discussion of
Namikas and Schneider and Fisk. In R.A. Magill Schmidt, R.A., Gielen, C.C.A.M., & van den Heuvel,
(Ed.), Memory and control of action (pp. 145-165). P.J.M. (1984). The locus of intersensory facilitation of
Amsterdam: Elsevier. reaction time. Acta Psychologica, 57, 145-164.
Schmidt, R.A. (1985). The search for invariance in skilled Schmidt, R.A., & Gordon, G.B. (1977). Errors in
movement behavior. Research Quarterly for Exercise motor responding, "rapid" corrections, and false
and Sport, 56, 188-200. anticipations. Journal of Motor Behavior, 9, 101-111.
Schmidt, R.A. (1987). The acquisition of skill: Some Schmidt, R.A., Heuer, H., Ghodsian, D., & Young, D.E.
modifications to the perception-action relationship (1998). Generalized motor programs and units of
through practice. In H. Heuer & A.F. Sanders (Eds.), action in bimanual coordination. In M. Latash (Ed.),
Perspectives on perception and action (pp. 77-103). Progress in motor control, Vol. 1: Bernstein's traditions
Hillsdale, NJ: Erlbaum. in movement studies (pp. 329-360). Champaign, IL:
Human Kinetics.
Schmidt, R.A. (1988). Motor and action perspectives on
motor behaviour. In O.G Meijer & K. Roth (Eds.), Schmidt, R.A., Lange, C., & Young, D.E. (1990).
Complex movement behaviour: "The" motor-action Optimizing summary knowledge of results for skill
controversy (pp. 3-44). Amsterdam: Elsevier. learning. Human Movement Science, 9, 325-348.
Schmidt, R.A. (1989a). Toward a better understanding Schmidt, R.A., & McGown, C. (1980). Terminal accuracy
of the acquisition of skill: Theoretical and practical of unexpectedly loaded rapid movements: Evidence
548 References

for a mass-spring mechanism in programming. & J.D. Hagman (Eds.), Transfer of learning (pp. 47-
Journal of Motor Behavior, 12, 149-161. 79). Orlando, FL: Academic Press.
Schmidt, R.A., McGown, C., Quinn, J.T., & Hawkins, B. Schmidt, R.A., & Young, D.E. (1991). Methodology for
(1986). Unexpected inertial loading in rapid reversal motor learning: A paradigm for kinematic feedback.
movements: Violations of equifinality. Human Journal of Motor Behavior, 23, 13-24.
Movement Science, 5, 263-273. Schmidt, R.A., Young, D.E., Swinnen, S., & Shapiro,
Schmidt, R.A., & Nacson, J. (1971). Further tests of the D.C. (1989). Summary knowledge of results for skill
activity-set hypothesis for warm-up decrement. acquisition: Support for the guidance hypothesis.
Journal of Experimental Psychology, 90, 56-64. Journal of Experimental Psychology: Learning,
Schmidt, R.A., & Pew, R.W. (1974). Predicting motor- Memory, and Cognition, 15, 352-359.
manipulative performances in the manufacture Schmidt, R.A., Zelaznik, H.N., & Frank, J.S. (1978). Sources
of dental appliances (Tech. Rep. to Heritage of inaccuracy in rapid movement. In G.E. Stelmach
Laboratories). Romulus, MI: University of Michigan. (Ed.), Information processing in motor control and
Schmidt, R.A., & Russell, D.G. (1972). Movement learning (pp. 183-203). New York: Academic Press.
velocity and movement time as determiners of degree Schmidt, R.A., Zelaznik, H.N., Hawkins, B., Frank, J.S.,
of preprogramming in simple movements. Journal of & Quinn, J.T. Jr. (1979). Motor-output variability:
Experimental Psychology, 96, 315-320. A theory for the accuracy of rapid motor acts.
Schmidt, R.A., & Russell, D.G. (1974). Error detection Psychological Review, 86, 415-451.
in positioning responses. Unpublished manuscript, Berlin: Springer.
University of Michigan, Ann Arbor. Schmidt, R.C., Carello, C., & Turvey, M.T. (1990). Phase
Schmidt, R.A., & Shapiro, D.C. (1986). Optimizing transitions and critical fluctuations in the visual
feedback utilization in motor skill training (Tech. coordination of rhythmic movements between
Rep. Contract No. MDA903-85-K-0225). Alexandria, people. Journal of Experimental Psychology: Human
CA: U.S. Army Research Institute. Perception and Performance, 16, 227-247.
Schmidt, R.A., & Shea, J.B. (1976). A note on delay Schmidt, R.C., Christianson, N., Carello, C., & Baron,
of knowledge of results in positioning responses. R. (1994). Effects of social and physical variables
Journal of Motor Behavior, 8, 129-131. on between-person visual coordination. Ecological
Psychology, 6, 159-183.
Schmidt, R.A., & Sherwood, D.E. (1982). An inverted-U
relation between spatial error and force requirements Schmidt, R.C., & O'Brien, B. (1997). Evaluating the
in rapid limb movements: Further evidence for the dynamics of unintended interpersonal coordination.
impulse-variability model. Journal of Experimental Ecological Psychology, 9, 189-206.
Psychology: Human Perception and Performance, 8, Schmidt, R.C., & Richardson, M.J. (2008). Dynamics of
158-170. interpersonal coordination. In A. Fuchs & V.K. Jirsa
Schmidt, R.A., Sherwood, D.E., Zelaznik, H.N., & (Eds.), Coordination: Neural, behavioral and social
Leikind, B.J. (1985). Speed-accuracy trade-offs in dynamics (pp. 281-308).
motor behavior: Theories of impulse variability. In H. Schmidt, R.C., Shaw, B.K., & Turvey, M.T. (1993).
Heuer, U. Kleinbeck, & K.-H. Schmidt (Eds.), Motor Coupling dynamics in interlimb coordination. Journal
behavior: Programming, control, and acquisition (pp. of Experimental Psychology: Human Perception and
79-123). Berlin: Springer-Verlag. Performance, 19, 397-415.
Schmidt, R.A., & White, J.L. (1972). Evidence for an error Schneider, D.M., & Schmidt, R.A. (1995). Units of action
detection mechanism in motor skills: A test of Adams' in motor control: Role of response complexity and
closed-loop theory. Journal of Motor Behavior, 4, 143- target speed. Human Performance, 8, 27-49.
153. Schneider, G.E. (1969). Two visual systems. Science, 163,
Schmidt, R.A., Wood, C.T., Young, D.E., & Kelkar, 895-902.
R. (1996). Evaluation of the BIC J26 child guard Schneider, K., Zernicke, R.F., Schmidt, R.A., & Hart, T.J.
lighter (Tech. Rep.). Los Angeles: Failure Analysis (1989). Changes in limb dynamics during the practice
Associates, Inc. of rapid arm movements. Journal of Biomechanics,
Schmidt, R.A., & Wrisberg, C.A. (1971). The activity-set 22, 805-817.
hypothesis for warm-up decrement in a movement- Schneider, VI., Healy, A.F., & Bourne, L.E. Jr. (2002).
speed task. Journal of Motor Behavior, 3, 318-325. What is learned under difficult conditions is hard
Schmidt, R.A., & Wulf, G. (1997). Continuous concurrent to forget: Contextual interference effects in foreign
feedback degrades skill learning: Implications for vocabulary acquisition, retention, and transfer.
training and simulation. Human Factors, 39, 509-525. Journal of Memory and Language, 46, 419-440.
Schmidt, R.A., & Young, D.E. (1987). Transfer of Schneider, W. (1985). Training high-performance skills:
movement control in motor learning. In S.M. Cormier Fallacies and guidelines. Human Factors, 27, 285-300.
 References 549

Schneider, W., Dumais, S.T., & Shiffrin, R.M. (1984). Seidler, R.D., & Noll, D.C. (2008). Neuroanatomical
Automatic processing and attention. In R. correlates of motor acquisition and motor transfer.
Parasuraman & R. Davies (Eds.), Varieties of attention Journal of Neurophysiology, 99, 1836-1845.
(pp. 1-27). New York: Academic Press. Semjen, A., & Ivry, R.B. (2001). The coupled oscillator
Schneider, W., & Fisk, AD. (1983). Attention theory and model of between-hand coordination in alternate-
mechanisms for skilled performance. In R.A. Magill hand tapping: A reappraisal. Journal of Experimental
(Ed.), Memory and control of action (pp. 119-143). Psychology: Human Perception and Performance, 27,
Amsterdam: Elsevier. 251-265.
Schneider, W., & Shiffrin, R.M. (1977). Controlled Senders, J.W. (1998). Analysis of an intersection.
and automatic human information processing: I. Ergonomics in Design, 6, 4-6.
Detection, search, and attention. Psychological Seow, S.C. (2005). Information theoretic models of HCI:
Review, 84, 1-66. A comparison of the Hick-Hyman Law and Fitts'
Scholz, J.P., & Kelso, J.A.S. (1990). Intentional switching Law. Human-Computer Interaction, 20, 315-352.
between patterns of bimanual coordination depends Sergent, V., Hellige, J.B., & Cherry, B. (1993). Effects of
on the intrinsic dynamics of the patterns. Journal of responding hand and concurrent verbal processing
Motor Behavior, 22, 98-124. on time-keeping and motor-implementation pro-
Scholz, J.P., Kelso, J.A.S., & Schi:iner, G. (1987). cesses. Brain and Cognition, 23, 243-262.
Nonequilibrium phase transitions in coordinated Serrien, D.J., Bogaerts, H., Suy, E., & Swinnen, S.P. (1999).
biological motion: Critical slowing down and The identification of coordination constraints across
switching time. Physics Letters A, 123, 390-394. planes of motion. Experimental Brain Research, 128,
Schi:iner, G., Zanone, P.G., & Kelso, J.A.S. (1992). Learning 250-255.
as change of coordination dynamics: Theory and Serrien, D.J., Burgunder, J.-M., & Wiesendanger, M. (2002).
experiment. Journal of Motor Behavior, 24, 29-48. Control of manipulative forces during unimanual and
Schutz, R. (1977). Absolute, constant, and variable bimanual tasks in patients with Huntington's disease.
error: Problems and solutions. In D. Mood (Ed.), The Experimental Brain Research, 143, 328-334.
measurement of change in physical education (pp. Serrien, D.J., Steyvers, M., Debaere, F., Stelmach, G.E.,
82-100). Boulder, CO: University of Colorado Press. & Swinnen, S.P. (2000). Bimanual coordination and
Schutz, R.W., & Roy, E.A. (1973). Absolute error: The devil limb-specific parameterization in patients with
in disguise. Journal of Motor Behavior, 5, 141-153. Parkinson's disease. Neuropsychologia, 38, 1714-
Schweickert, R. (1993). Information, time, and the 1722.
structure of mental events: A twenty-five-year review. Serrien, D.J., & Swinnen, S.P. (1997a). Coordination
In D.E. Meyer & S. Kornblum (Eds.), Attention and constraints induced by effector combination under
performance XIV (pp. 324-341). Cambridge, MA: isofrequency and multifrequency conditions. Journal
Bradford. of Experimental Psychology: Human Perception and
Schwoebel, J., Boronat, C.B., & Branch Coslett, H. (2002). Performance, 23, 1493-1510.
The man who executed "imagined" movements: Serrien, D.J., & Swinnen, S.P. (1997b). Isofrequency
Evidence for dissociable components of the body and multifrequency coordination patterns as a
schema. Brain and Cognition, 50, 1-16. function of the planes of motion. Quarterly Journal of
Scripture, E.W. (1894). Tests of mental ability as exhibited Experimental Psychology, 50A, 386-404.
in fencing. Studies From the Yale Psychological Serrien, D.J., & Swinnen, S.P. (1999). Intentional switching
Laboratory, 2, 122-124. between behavioral patterns of homologous and
Scully, D.M., & Newell, K.M. (1985). Observational nonhomologous effector combinations. Journal of
learning and the acquisition of motor skills: Toward Experimental Psychology: Human Perception and
a visual perception perspective. Journal of Human Performance, 25, 1253-1267.
Movement Studies, 11, 169-186. Servos, P., Carnahan, H., & Fedwick, J. (2000). The
Sears, T.A., & Newsom-Davis, J. (1968). The control of visuomotor system resists the horizontal-vertical
respiratory muscles during voluntary breathing. illusion. Journal of Motor Behavior, 32, 400-404.
Annals of the New York Academy of Sciences, 155, Seymour, W.D. (1954). Experiments on the acquisition of
183-190. industrial skills. Occupational Psychology, 28, 77-89.
Seibel, R. (1963). Discrimination reaction time for Shadmehr, R., & Krakauer, J.W. (2008). A computational
a 1023-alternative task. Journal of Experimental neuroanatomy for motor control. Experimental Brain
Psychology, 66, 215-226. Research, 185, 359-381.
Seidler, R.D. (2007). Aging affects motor learning but not Shaffer, L.H. (1971). Attention in transcription skill.
savings at transfer of learning. Learning & Memory, Quarterly Journal of Experimental Psychology, 23,
14, 17-21. 107-112.
550 Iii References

Shaffer, L.H. (1980). Analyzing piano performance: ences on performance and learning. Journal of Motor
A study of concert pianists. In G.E. Stelmach & J. Behavior, 23, 51-62.
Requin (Eds.), Tutorials in motor behavior (pp. 443- Shea, C.H., Wright, D.L., Wulf, G., & Whitacre, C. (2000).
455). Amsterdam: Elsevier. Physical and observational practice afford unique
Shaffer, L.H. (1984). Timing in solo and duet piano learning opportunities. Journal of Motor Behavior,
performances. Quarterly Journal of Experimental 32, 27-36.
Psychology, 36A, 577-595. Shea, C.H., & Wulf, G. (1999). Enhancing motor learning
Shannon, C.E., & Weaver, W. (1949). The mathematical through external-focus instruction and feedback.
theory of communication. Urbana, IL: University of Human Movement Science, 18, 553-571.
Illinois Press. Shea, C.H., & Wulf, G. (2005). Schema theory: A critical
Shapiro, D.C. (1977). A preliminary attempt to determine appraisal and reevaluation. Journal of Motor
the duration of a motor program. In D.M. Landers & Behavior, 37, 85-101.
R.W. Christina (Eds.), Psychology of motor behavior Shea, C.H., Wulf, G., & Whitacre, C. (1999). Enhancing
and sport-1976 (pp. 17-24). Champaign, IL: Human training efficiency and effectiveness through the use of
Kinetics. dyad training. Journal of Motor Behavior, 31, 119-125.
Shapiro, D.C. (1978). The learning of generalized motor Shea, J.B., & Graf, R.C. (1994). A model for contextual
programs. Unpublished doctoral dissertation, interference effects in motor learning. In C.R. Reynolds
University of Southern California, Los Angeles. (Ed.), Cognitive assessment: A multidisciplinary
Shapiro, D.C.,&Schmidt,R.A. (1982). The schema theory: perspective (pp. 73-87). New York: Plenum Press.
Recent evidence and developmental implications. In Shea, J.B., & Morgan, R.L. (1979). Contextual interference
J.A.S. Kelso & J.E. Clark (Eds.), The development of effects on the acquisition, retention, and transfer of
movement control and co-ordination (pp. 113-150). a motor skill. Journal of Experimental Psychology:
New York: Wiley. Human Leaming and Memory, 5, 179-187.
Shapiro, D.C., & Walter, C.B. (1982). Control of rapid bi- Shea, J.B., & Titzer, R.C. (1993). The influence of reminder
manual aiming movements: The effect of a mechanical trials on contextual interference effects. Journal of
block. Society for Neuroscience Abstracts, 8(2), 733. Motor Behavior, 25, 264-274.
Shapiro, D.C., Zernicke, R.F., Gregor, R.J., & Diestel, J.D. Shea, J.B., & Upton, G. (1976). The effects on skill
(1981). Evidence for generalized motor programs acquisition of an interpolated motor short-term
using gait pattern analysis. Journal of Motor memory task during the KR-delay interval. Journal
Behavior, 13, 33-47. of Motor Behavior, 8, 277-281.
Shapiro, K. (2001). The limits of attention: Temporal Shea, J.B.,& Wright,D.L. (1991). When forgetting benefits
constraints on human information processing. New motor retention. Research Quarterly for Exercise and
York: Oxford University Press. Sport, 62, 293-301.
Shea, C.H., Guadagnoli, M.A., & Dean, M. (1995). Shea, J.B., & Zimny, S.T. (1983). Context effects in
Response biases: Tonic neck response and memory and learning movement information. In
aftercontraction phenomenon. Journal of Motor R.A. Magill (Ed.), Memory and control of action (pp.
Behavior, 27, 41-51. 345-366). Amsterdam: Elsevier.
Shea, C.H., & Kohl, R.M. (1990). Specificity and Shea, J.B., & Zimny, S.T. (1988). Knowledge incorporation
variability of practice. Research Quarterly for in motor representation. In 0.G. Meijer & K. Roth
Exercise and Sport, 61, 169-177. (Eds.), Complex movement behaviour: "The"
Shea, C.H., & Kohl, R.M. (1991). Composition of practice: motor-action controversy (pp. 289-314). Amsterdam:
Influence on the retention of motor skills. Research Elsevier.
Quarterly for Exercise and Sport, 62, 187-195. Sheppard, D.J. (1984). Visual and part-task manipulations
Shea, C.H., Kohl, R., & Inderrnill, C. (1990). Contextual for teaching simulated carrier landings (Rep. No. 81-
interference: Contributions of practice. Acta C-0105-9). Orlando, FL: Naval Training Equipment
Psychologica, 73, 145-157. Center.
Shea, C.H., Lai, Q., Black, C., & Park, J.-H. (2000). Spacing Sherrington, C.S. (1906). The integrative action of the
practice sessions across days benefits the learning of nervous system. New Haven, CT: Yale University
motor skills. Human Movement Science, 19, 737-760. Press.
Shea, C.H., Lai, Q., Wright, D.L., Immink, M., & Black, C. Sherwood, D.E. (1988). Effect of bandwidth knowledge
(2001). Consistent and variable practice conditions: of results on movement consistency. Perceptual and
Effects on relative timing and absolute timing. Motor Skills, 66, 535-542.
Journal of Motor Behavior, 33, 139-152. Sherwood, D.E. (1990). Practice and assimilation effects
Shea, C.H., Shebilske, W.L., Kohl, R.M., & Guadagnoli, in a multilimb aiming task. Journal of Motor Behavior,
M.A. (1991). After-contraction phenomenon: Influ- 22, 267-291.
 References 551

Sherwood, D.E. (1991 ). Distance and location assimilation Shockley, K., Santana, M.V., & Fowler, C.A. (2003).
effects in rapid bimanual movement. Research Mutual interpersonal postural constraints are
Quarterly for Exercise and Sport, 62, 302-308. involved in cooperative conversation. Journal of
Sherwood, D.E. (1994). Interlimb amplitude differences, Experimental Psychology: Human Perception and
spatial assimilations, and the temporal structure Performance, 29, 326-332.
of rapid bimanual movements. Human Movement Shockley, K., & Turvey, M.T. (2006). Dual-task influences
Science, 13, 841-860. on retrieval from semantic memory and coordination
Sherwood, D.E. (1996). The benefits of random variable dynamics. Psychonomic Bulletin & Review, 13, 985-
practice for spatial accuracy and error detection in a 990.
rapid aiming task. Research Quarterly for Exercise Short, M.W., & Cauraugh, J.H. (1999). Precision
and Sport, 67, 35-43. hypothesis and the end-state comfort effect. Acta
Sherwood, D.E., & Canabal, M.Y. (1988). The effect of Psychologica, 100, 243-252.
practice on the control of sequential and simultaneous Shumway-Cook, A., & Woollacott, M.H. (2007). Motor
actions. Human Performance, 1, 237-260. control: Translating research into clinical practice
Sherwood, D.E., & Lee, T.D. (2003). Schema theory: (3rd ed.). Baltimore: Lippincott Williams & Wilkins.
Critical review and implications for the role of Sidaway, B., Ahn, S., Boldeau, P., Griffin, S., Noyes, B., &
cognition in a new theory of motor learning. Research Pelletier, K. (2008). A comparison of manual guidance
Quarterly for Exercise and Sport, 74, 376-382. and knowledge of results in the learning of a weight-
Sherwood, D.E., & Schmidt, R.A. (1980). The relationship bearing skill. Journal of N eurologic Physical Therapy,
between force and force variability in minimal and 32, 32-38.
near-maximal static and dynamic contractions. Sidaway, B., Moore, B., & Schoenfelder-Zohdi, B. (1991).
Journal of Motor Behavior, 12, 75-89. Summary and frequency of KR presentation effects
Sherwood, D.E., Schmidt, R.A., & Walter, C.B. (1988). The on retention of a motor skill. Research Quarterly for
force/ force-variability relationship under controlled Exercise and Sport, 62, 27-32.
temporal conditions. Journal of Motor Behavior, 20, Sidaway, B., Sekiya, H., & Fairweather, M. (1995).
106-116. Movement variability as a function of accuracy
Shewokis, P.A., & Klopfer, D. (2000). Some blocked demand in programmed serial aiming responses.
practice schedules yield better learning than random Journal of Motor Behavior, 27, 67-76.
practice schedules with anticipation timing tasks. Sidman, M. (1952). A note on functional relations
Journal of Human Movement Studies, 38, 57-73. obtained from group data. Psychological Bulletin, 49,
Shiffrin, R.M., & Schneider, W. (1977). Controlled 263-269.
and automatic human information processing: II. Siegmund, G.P., Inglis, J.T., & Sanderson, D.J. (2001).
Perceptual learning, automatic attending, and a Startle response of human neck muscles sculpted
general theory. Psychological Review, 84, 127-190. by readiness to perform ballistic head movements.
Shik, M.L., & Orlovskii, G.N. (1976). Neurophysiology Journal of Physiology, 535, 289-300.
of a locomotor automatism. Physiological Reviews, Simon, D.A., & Bjork, R.A. (2001). Metacognition in
56, 465-501. motor learning. Journal of Experimental Psychology:
Shik, M.L., Orlovskii, G.N., & Severin, F.V. (1968). Learning, Memory, and Cognition, 27, 907-912.
Locomotion of the mesencephalic cat elicited by Simon, D.A., & Bjork, R.A. (2002). Models of
stimulation of the pyramids. Biofizika, 13, 143-152. performance in learning multisegment movement
Shim, J., Carlton, L.G., Chow, J.W., Chae, W-S. (2005). tasks: Consequences for acquisition, retention, and
The use of anticipatory visual cues by highly skilled judgments of learning. Journal of Experimental
tennis players. Journal of Motor Behavior, 37, 164- Psychology: Applied, 8, 222-232.
175. Simon, D.A., Lee, T.D., & Cullen, J.D. (2008). Win-shift,
Shinar, D., Tractinsky, N., & Compton, R. (2005). Effects lose-stay: Contingent switching and contextual
of practice, age, and task demands, on interference interference in motor learning. Perceptual and Motor
from a phone task while driving. Accident Analysis Skills, 107, 407-418.
and Prevention, 37, 315-326. Simon, J.R. (1969a). Reactions toward the source of
Shirley, M.M. (1931). The first two years (Vol. 1). stimulation. Journal of Experimental Psychology, 81,
Minneapolis: University of Minnesota Press. 174-176.
Shockley, K., Baker, A.A., Richardson, M.J., & Fowler, Simon, J.R. (1969b). Stereotypic reaction information
C.A. (2007). Articulatory constraints on interpersonal processing. In L.E. Smith (Ed.), Psychology of motor
postural coordination. Journal of Experimental learning (pp. 27-50). Chicago: Athletic Institute.
Psychology: Human Perception and Performance, Simon, J.R. (1990). The effects of an irrelevant directional
33, 201-208. cue on human information processing. In R.W.
552 References

Proctor & T.G. Reeve (Eds.), Stimulus-response Small, A.M. (1990). Foreword. In R.W. Proctor & T.G.
compatibility: An integrated perspective (pp. 31-86). Reeve (Eds.), Stimulus-response compatibility: An in-
Amsterdam: Elsevier. tegrated perspective (pp. v-vi). Amsterdam: Elsevier.
Simon, J.R., & Rudell, A.P. (1967). Auditory S-R Smeeton, N.J., Ward, P., & Williams, A.M. (2004). Do
compatibility: The effect of an irrelevant cue pattern recognition skills transfer across sports? A
on information processing. Journal of Applied preliminary analysis. Journal of Sports Sciences, 22,
Psychology, 51, 300-304. 205-213.
Simons, D.J. (2000). Attentional capture and inattentional Smeets, J.B.J., & Brenner, E. (1999). A new view on
blindness. Trends in Cognitive Sciences, 4, 147-155. grasping. Motor Control, 3, 237-271.
Simons, D.J., & Chabris, C.F. (1999). Gorillas in our Smeets, J.B.J., & Brenner, E. (2001). Action beyond our
midst: Sustained inattentional blindness for dynamic grasp. Trends in Cognitive Sciences, 5, 287.
events. Perception, 28, 1059-1074. Smeets, J.B.J., Brenner, E., & Biegstraaten, M. (2002).
Simons, D.J., & Levin, D.T. (1998). Failure to detect Independent control of the digits predicts an apparent
changes to people in a real-world interaction. hierarchy of visuomotor channels in grasping.
Psychonomic Bulletin & Review, 5, 644-649. Behavioural Brain Research, 136, 427-432.
Simons, D.J., & Rensink, R.A. (2005). Change blindness: Smeets, J.B.J., Brenner, E., de Grave, D.D.J., & Cuijpers,
Past, present, and future. Trends in Cognitive R.H. (2002). Illusions in action: Consequences
Sciences, 9, 16-20. of inconsistent processing of spatial attributes.
Simons, J.P., Wilson, J.M., Wilson, G.J., & Theall, S. (2009). Experimental Brain Research, 147, 135-144.
Challenges to cognitive bases for an especial motor Smeets, J.B.J., Brenner, E., & Martin, J. (2009). Grasping
skill at the regulation baseball pitching distance. Occam's razor. In D. Sternad (Ed.), Progress in motor
Research Quarterly for Exercise and Sport, 80, 469-479. control (pp. 499-522). Berlin: Springer.
Singer, R.N. (1975). Motor learning and human Smeets, J.B.J., Brenner, E., Trebuchet, S., & Mestre, D.R.
performance (2nd ed.). New York: Macmillan. (1996). Is judging time-to-contact based on "tau"?
Singer, R.N. (1980). Motor learning and human Perception, 25, 583-590.
performance (3rd ed.). New York: Macmillan. Smethurst, C.J., & Carson, R.G. (2001). The acquisition
Singer, R.N. (2000). Performance and human factors: of movement skills: Practice enhances the dynamic
Considerations about cognition and attention for stability of bimanual coordination. Human
self-paced and externally-paced events. Ergonomics, Movement Science, 20, 499-529.
43, 1661-1680. Smethurst, C.J., & Carson, R.G. (2003). The effect of
Singer, R.N., & Gaines, L. (1975). Effects of prompted and volition on the stability of bimanual coordination.
trial-and-error learning on transfer performance of a Journal of Motor Behavior, 35, 309-319.
serial motor task. American Educational Research Smith, J.L. (1969). Fusimotor neuron block and voluntary
Journal, 12, 395-403. arm movement in man. Unpublished doctoral
Singer, R.N., & Pease, D. (1976). A comparison of dis- dissertation, University of Wisconsin, Madison.
covery learning and guided instructional strategies Smith, J.L. (1977). Mechanisms of neuromuscular
on motor skill learning, retention, and transfer. Re- control. Los Angeles: UCLA Printing and Production.
search Quarterly, 47, 788-796.
Smith, J.L. (1978). Sensorimotor integration during motor
Sittig, A.C. (1986). Kinesthesis and motor control. programming. In G.E. Stelmach (Ed.), Information
Unpublished doctoral dissertation, University of processing in motor control and learning (pp. 173-
Utrecht, Holland. 182). New York: Academic Press.
Sittig, A.C., Denier van der Con, J.J., & Gielen, C.C.A.M. Smith, J.L., Bradley, N.S., Carter, M.C., Giuliani, C.A.,
(1985a). Separate control of arm position and velocity Hoy, M.G., Koshland, G.F., & Zernicke, R.F. (1986).
demonstrated by vibration of muscle tendon in man. Rhythmical movements of the hindlimbs in spinal
Experimental Brain Research, 60, 445-453. cat: Considerations for a controlling network. In
Sittig, A.C., Denier van der Con, J.J., & Gielen, C.C.A.M. M.E. Goldberger, A. Corio, & M. Murray (Eds.),
(1985b). Different control mechanisms for slow and Development and plasticity of the mammalian spinal
fast human arm movements. Neuroscience Letters, cord (pp. 362-374). Padova: Liviana Press.
22, S128. Smith, J.L., Roberts, E.M., & Atkins, E. (1972). Fusimotor
Skoglund, S. (1956). Anatomical and physiological neuron block and voluntary arm movement in man.
studies of the knee joint innervation in the cat. Acta American Journal of Physical Medicine, 5, 225-239.
Physiologica Scandinavica, 36 (Suppl. 124). Smith, M.C. (1969). The effect of varying information on
Slater-Hammel, A.T. (1960). Reliability, accuracy the psychological refractory period. In W.G. Koster
and refractoriness of a transit reaction. Research (Ed.), Attention and performance II (pp. 220-231).
Quarterly, 31, 217-228. Amsterdam: North-Holland.
 References 11 553

Smith, P.J.K. (2002). Applying contextual interference to Sperling, G. (1960). The information available in brief
snowboarding skills. Perceptual and Motor Skills, 95, visual presentations. Psychological Monographs, 74
999-1005. (11, Whole No. 498).
Smith, P.J.K., & Davies, M. (1995). Applying contextual Sperry, R.W. (1950). Neural basis of the spontaneous
interference to the Pawlata roll. Journal of Sports optokinetic response produced by visual inversion.
Sciences, 13, 455-462. Journal of Comparative and Physiological Psycho-
Smith, P.J.K., Gregory, S.K., & Davies, M. (2003). logy, 43, 482-489.
Alternating versus blocked practice in learning a Spijkers, W., & Heuer, H. (1995). Structural constraints
cartwheel. Perceptual and Motor Skills, 96, 1255-1264. on the performance of symmetrical bimanual
Smith, P.J.K., Taylor, S.J., & Withers, K. (1997). Applying movements with different amplitudes. Quarterly
bandwidth feedback scheduling to a golf shot. Journal of Experimental Psychology, 48A, 716-740.
Research Quarterly for Exercise and Sport, 68, 215- Spijkers, W.A.C., & Lochner, P. (1994). Partial visual
221. feedback and spatial end-point accuracy of visual aim-
Smith, S.M., & Vela, E. (2001). Environmental context- ing movements. Journal of Motor Behavior, 26, 283-295.
dependent memory: A review and meta-analysis. Spirduso, W.W., Francis, K.L., & MacRae, P.S. (2005).
Psychonomic Bulletin & Review, 8, 203-220. Physical dimensions of aging (2nd ed.). Champaign,
Smith, W.M., & Bowen, K.F. (1980). The effects of delayed IL: Human Kinetics.
and displaced visual feedback on motor control. Stahl, J.S. (2001). Eye-head coordination and the variation
Journal of Motor Behavior, 12, 91-101. of eye-movement accuracy with orbital eccentricity.
Smits-Engelsman, B.C.M., Van Galen, G.P., & Duysens, Experimental Brain Research, 136, 200-210.
J. (2002). The breakdown of Fitts' law in rapid, Starkes, J.L. (1987). Attention demands of spatially
reciprocal aiming movements. Experimental Brain locating position of a ball in flight. Perceptual and
Research, 145, 222-230. Motor Skills, 64, 127-135.
Snoddy, G.S. (1926). Learning and stability: A Starkes, J.L., & Allard, F. (Eds.). (1993). Cognitive issues
psychophysical analysis of a case of motor learning in motor expertise. Amsterdam: Elsevier.
with clinical applications. Journal of Applied Starkes, J.L., Cullen, J.D., & MacMahon, C. (2004). A life-
Psychology, 10, 1-36. span model of the acquisition and retention of expert
Snoddy, G.S. (1935). Evidence for two opposed processes perceptual-motor performance. In A.M. Williams &
in mental growth. Lancaster, PA: Science Press. N.J. Hodges (Eds.), Skill acquisition in sport: Research,
theory, and practice (pp. 259-281). New York:
Solley, W.H. (1952). The effects of verbal instruction of
Routledge.
speed and accuracy upon the learning of a motor
skill. Research Quarterly, 23, 231-240. Starkes, J.L., & Ericsson, K.A. (2003). Expert performance
in sport: Advances in research on sport expertise.
Sonstroem, R.J., & Bernardo, P. (1982). Intraindividual
Champaign, IL: Human Kinetics.
pregame state anxiety and basketball performance:
A reexamination of the inverted-CT curve. Journal of Starkes, J.L., Helsen, W., & Elliott, D. (2002). A menage
Sport Psychology, 4, 235-245. a trois: The eye, the hand and on-line processing.
Journal of Sports Sciences, 20, 217-224.
Soucy, M.-C., & Proteau, L. (2001). Development of
multiple movement representations with practice: Starkes, J.L., Helsen, W., & Jack, R. (2001). Expert
Specificity versus flexibility. Journal of Motor performance in sport and dance. In R.N. Singer, H.A.
Behavior, 33, 243-254. Hausenblas, & C.M. Janelle (Eds.), Handbook of sport
psychology (2nd ed.) (pp. 174-201). New York: Wiley.
Southard, D., & Higgins, T. (1987). Changing movement
patterns: Effects of demonstration and practice. Steffens, M.C. (2007). Memory for goal-directed
Research Quarterly for Exercise and Sport, 60, 348- sequences of actions: Is doing better than seeing?
356. Psychonomic Bulletin & Review, 14, 1194-1198.
Sparrow, W.A., & Summers, J.J. (1992). Performance on Stelmach, G.E. (1969). Prior positioning responses as a
trials without knowledge of results (KR) in reduced factor in short-term retention of a simple motor task.
relative frequency presentations of KR. Journal of Journal of Experimental Psychology, 81, 523-526.
Motor Behavior, 24, 197-209. Stelmach, G.E., Kelso, J.A.S., & Wallace, S.A. (1975).
Spencer, J.P., & Thelen, E. (1999). A multimuscle state Preselection in short-term motor memory. Journal
analysis of adult motor learning. Experimental Brain of Experimental Psychology: Human Learning and
Research, 128, 505-516. Memory, 1, 745-755.
Spencer, R.M., Zelaznik, H.N., Diedrichsen, J., & Ivry, Stelmach, G.E., & Requin, J. (Eds.). (1980). Tutorials in
R.B. (2003). Disrupted timing of discontinuous but motor behavior. Amsterdam: North-Holland.
not continuous movements by cerebellar lesions. Stelmach, G.E., & Requin, J. (Eds.). (1992). Tutorials in
Science, 300, 1437-1439. motor behavior IL Amsterdam: Elsevier.
554 References

Ste-Marie, D.M., Clark, S.E., Findlay, L.C., & Latimer, Stroop, J.R. (1935). Studies of interference in serial verbal
A.E. (2004). High levels of contextual interference reactions. Journal of Experimental Psychology, 18,
enhance handwriting skill acquisition. Journal of 643-662.
Motor Behavior, 36, 115-126. Stuart, D.G., Mosher, C.G., Gerlack, R.L., & Reinking,
Sternad, D. (2001). Kurt Wachholder: Pioneering R.M. (1972). Mechanical arrangement and trans-
electrophysiological studies of voluntary movements. ducing properties of Golgi tendon organs. Experi-
In M.L. Latash & V.M. Zatsiorsky (Eds.), Classics mental Brain Research, 14, 274-292.
in movement science (pp. 375-407). Champaign, IL: Stuart, D.G., Pierce, P.A., Callister, R.J., Brichta, A.M., &
Human Kinetics. McDonagh, J.C. (2001). Sir Charles S. Sherrington: Hu-
Sternad, D. (2002). Wachholder and Altenburger 1927: manist, mentor, and movement neuroscientist. In M.L.
Foundational experiments for current hypotheses on Latash & V.M. Zatsiorsky (Eds.), Classics in movement
equilibrium-point control in voluntary movements. science (pp. 317-374). Champaign, IL: Hum an Kinetics.
Motor Control, 6, 299-318. Studenka, B.E., & Zelaznik, H.N. (2008). The influence of
Sternad, D., & Corcos, D. (2001). Effect of task and dominant versus non-dominant hand on event and
instruction on patterns of muscle activation: emergent motor timing. Human Movement Science,
Wachholder and beyond. Motor Control, 5, 307-336. 27, 29-52.
Sternberg, S. (1969). The discovery of processing stages: Sturm, L.P., Windsor, J.A., Cosman, P.H., Cregan, P.,
Extensions of Danders' method. In W.G. Koster Hewett, P.J., & Maddern, G.J. (2008). A systematic
(Ed.), Attention and performance II (pp. 276-315). review of skills transfer after surgical simulation
Amsterdam: North-Holland. training. Annals of Surgery, 248, 166-179.
Sternberg, S., Mansell, S., Knoll, R.L., & Wright, C.E. Sullivan, K.J., Kantak, S.S., & Burtner, P.A. (2008).
(1978). The latency and duration of rapid movement Motor learning in children: Feedback effects on skill
sequences: Comparisons of speech and typewriting. acquisition. Physical Therapy, 88, 720-732.
In G.E. Stelmach (Ed.), Information processing in Summers, J.J. (1975). The role of timing in motor program
motor control and learning (pp. 117-152). New York: representation. Journal of Motor Behavior, 7, 229-241.
Academic Press.
Summers, J.J. (Ed.). (1992). Approaches to the study of
Stevens, J.A. (2005). Interference effects demonstrate motor control and learning. Amsterdam: Elsevier.
distinct roles for visual and motor imagery during the
Summers, J.J. (2002). Practice and training in bimanual
mental representation of human action. Cognition,
coordination tasks: Strategies and constraints. Brain
95, 329-350.
and Cognition, 48, 166-178.
Stevens, L.T. (1886). On the time-sense. Mind, 11, 393-
Summers, J.J. (2004). A historical perspective on skill
404.
acquisition. In A.M. Williams & N.J. Hodges (Eds.),
Stewart, D., Cudworth, C.J., & Lishman, J.R. (1993). Skill Acquisition in sport: Research, theory and
Misperception of time-to-collision by drivers in practice (pp. 1-26). London: Routledge.
pedestrian accidents. Perception, 22, 1227-1244.
Summers, J.J., & Anson, J.G. (2009). Current status of
Stickgold, R., & Walker, M.P. (2006). Memory the motor program: Revisited. Human Movement
consolidation and reconsolidation: What is the role of Science, 28, 566-577.
sleep? Trends in Neurosciences, 28, 408-415. Summers, J.J., & Kennedy, T. (1992). Strategies in the
Stimpel, E. (1933). Der Wurk [The throw]. Neue production of a 5:3 polyrhythm. Human Movement
Psychologische Studien, 9, 105-138. Science, 11, 101-112.
Stratton, S.M., Liu, Y.T., Hong, S.L., Mayer-Kress, G., & Summers, J.J., & Pressing, J. (1994). Coordinating the
Newell, K.M. (2007). Snoddy (1926) revisited: Time two hands in polyrhythmic tapping. In S.P. Swinnen,
scales of motor learning. Journal of Motor Behavior, H. Heuer, J. Massion, & P. Casaer (Eds.), Interlimb
39, 503-515. coordination: Neural, dynamical, and cognitive
Strayer, D.L., Drews, F.A., & Crouch, D.J. (2006). A constraints (pp. 571-593). San Diego: Academic Press.
comparison of the cell phone driver and the drunk Summers, J.J., Rosenbaum, D.A., Burns, B.D., & Ford,
driver. Human Factors, 48, 381-391. S.K. (1993). Production of polyrhythms. Journal of
Strayer, D.L., Drews, F.A., & Johnston, W.A. (2003). Experimental Psychology: Human Perception and
Cell phone-induced failures of visual attention Performance, 19, 416-428.
during simulated driving. Journal of Experimental Summers, J.J., Todd, J.A., & Kim, Y.H. (1993). The
Psychology: Applied, 9, 23-32. influence of perceptual and motor factors on
Strayer, D.L., & Johnston, W.A. (2001). Driven to bimanual coordination in a polyrhythmic tapping
distraction: Dual-task studies of simulated driving task. Psychological Research, 55, 107-115.
and conversing on a cellular telephone. Psychological Swanson, L.R., & Lee, T.D. (1992). Effects of aging and
Science, 12, 462-466. schedules of knowledge of results on motor learning.
 References 555

Journal of Gerontology: Psychological Sciences, 47, Exploring interlimb constraints during bimanual
P406-P411. graphic performance: Effects of muscle grouping and
Sweezy, R.W., & Andrews, D.H. (Eds.). (2001). Readings direction. Behavioural Brain Research, 79-87.
in simulation and training: A 30-year perspective. Swinnen, S.P., Schmidt, R.A., Nicholson, D.E., & Shapiro,
Santa Monica, CA: Human Factors and Ergonomics D.C. (1990). Information feedback for skill acquisition:
Society. Instantaneous knowledge of results degrades
Swets, J.A. (1964). Signal detection and recognition by learning. Journal of Experimental Psychology:
human observers. New York: Wiley. Learning, Memory, and Cognition, 16, 706-716.
Swets, J.A., Dawes, R.M., & Monahan, J. (2000). Swinnen, S.P., Walter, C.B., Lee, T.D., & Serrien, D.J.
Psychological science can improve diagnostic (1993). Acquiring bimanual skills: Contrasting forms
decisions. Psychological Science in the Public of information feedback for interlimb decoupling.
Interest, 1, 1-26. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 19, 1328-1344.
Swift, E.J., & Schuyler, W. (1907). The learning process.
Psychological Bulletin, 4, 307-310. Swinnen, S.P., Walter, C.B., Pauwels, J.M., Meugens, P.F.,
& Beirinckx, M.B. (1990). The dissociation of interlimb
Swinnen, S.P. (1988). Learning and long-term retention of constraints. Human Performance, 3, 187-215.
absolute and relative time. Unpublished manuscript,
Catholic University of Leuven, Belgium. Swinnen, S.P., Walter, C.B., & Shapiro, D.C. (1988).
The coordination of limb movements with different
Swinnen, S.P. (1990). Interpolated activities during the kinematic patterns. Brain and Cognition, 8, 326-347.
knowledge-of-results delay and post-knowledge-of-
results interval: Effects on performance and learning. Swinnen, S.P., & Wenderoth, N. (2004). Two hands, one
Journal of Experimental Psychology: Learning, brain: Cognitive neuroscience of bimanual skill.
Memory, and Cognition, 16, 692-705. Trends in Cognitive Sciences, 8, 18-25.
Swinnen, S.P. (1992). Coordination of upper-limb Szalma, S.L., Hancock, P.A., Warm, J.S., Dember, W.N.,
movement: A neuro-dynamics account. In G.E. & Parsons, KS. (2006). Training for vigilance: Using
Stelmach & J. Requin (Eds.), Tutorials in motor predictive power to evaluate feedback effectiveness.
behavior II (pp. 695-711). Amsterdam: Elsevier. Human Factors, 48, 682-692.
Swinnen, S.P. (1996). Information feedback for motor Tallet, J., Kostrubiec, V., & Zanone, P.-G. (2008). The role
skill learning: A review. In H.N. Zelaznik (Ed.), of stability in the dynamics of learning, memorizing,
Advances in motor learning and control (pp. 37-66). and forgetting new coordination patterns. Journal of
Champaign, IL: Human Kinetics. Motor Behavior, 40, 103-116.
Swinnen, S.P. (2002). Intermanual coordination: Tarpy, R.M., & Sawabini, F.L. (1974). Reinforcement
From behavioural principles to neural-network delay: A selective review of the last decade.
interactions. Nature Reviews: Neuroscience, 3, 350- Psychological Bulletin, 81, 984-997.
361. Taub, E. (1976). Movement in nonhuman primates
Swinnen, S.P., Dounskaia, N., Verschueren, S., deprived of somatosensory feedback. Exercise and
Serrien, D.J., & Daelman, A. (1995). Relative phase Sport Sciences Reviews, 4, 335-374.
destabilization during interlimb coordination: The Taub, E., & Berman, A.J. (1968). Movement and learning
disruptive role of kinesthetic afferences induced by in the absence of sensory feedback. In S.J. Freedman
passive movement. Experimental Brain Research, (Ed.), The neuropsychology of spatially oriented
105, 439-454. behavior. Homewood, IL: Dorsey Press.
Swinnen, S.P., Heuer, H., Massion, J., & Casaer, P. (Eds.). Taub, E., Uswatte, G., & Pidikiti, R. (1999). Constraint-
(1994). Interlimb coordination: Neural, dynamical, induced movement therapy: A new family of techni-
and cognitive constraints. San Diego: Academic ques with broad application to physical rehabilitation.
Press. A clinical review. Journal of Rehabilitation Research
Swinnen, S.P., Jardin, K., & Meulenbroek, R. (1996). Development, 36, 237-251.
Between-limb asynchronies during bimanual Taylor, T.L., & Klein, R.M. (1998). On the causes and
coordination: Effects of manual dominance and effects of inhibition of return. Psychonomic Bulletin
attentional cueing. Neuropsychologia, 34, 1203-1213. & Review, 5, 625-643.
Swinnen, S.P., Jardin, K., Meulenbroek, R., Dounskaia, Teasdale, N., Forget, R., Bard, C., Paillard, J., Fleury,
N., & Hofkens-van den Brandt, M. (1997). Egocentric M., & Lamarre, Y. (1993). The role of proprioceptive
and allocentric constraints in the expression of information for the production of isometric forces and
patterns of interlimb coordination. Journal of for handwriting tasks. Acta Psychologica, 82, 179-191.
Cognitive Neuroscience, 9, 348-377. Telford, C.W. (1931). The refractory phase of voluntary
Swinnen, S.P., Jardin, K., Verschueren, S., Meulenbroek, and associative responses. Journal of Experimental
R., Franz, L., Dounskaia, N., & Walter, C.B. (1998). Psychology, 14, 1-36.
556 References

Temprado, J.J. (2004). A dynamical approach to the Todd, J.T. (1981). Visual information about moving
interplay of attention and bimanual coordination. objects. Journal of Experimental Psychology: Human
In V.K. Jirsa & J.A.S. Kelso (Eds.), Coordination Perception and Performance, 7, 795-810.
dynamics: Issues and trends (pp. 21-39). Berlin: TOrnros, J.E.B., & Bolling, A.K. (2005). Mobile phone
Springer. use - effects of handheld and hands-free phones
Temprado, J.J., & Laurent, M. (2004). Attentional on driving performance. Accident Analysis and
load associated with performing and stabilizing Prevention, 37, 902-909.
a between-persons coordination of rhythmic limb Torre, K., & Balasubramaniam, R. (2009). Two different
movements. Acta Psychologica, 115, 1-16. processes for sensorimotor synchronization in
Terzuolo, C.A., & Viviani, P. (1979). The central continuous and discontinuous rhythmic movements.
representation of learning motor programs. In Experimental Brain Research, 199, 157-166.
R.E. Talbott & D.R. Humphrey (Eds.), Posture and Trachtman, D. (2003). Contextual interference and
movement (pp. 113-121). New York: Raven Press. the role of response retrieval during learning.
Terzuolo, C.A., & Viviani, P. (1980). Determinants and Unpublished doctoral dissertation, University of
characteristics of motor patterns used for typing. California, Los Angeles.
Neuroscience, 5, 1085-1103. Trankle, U., & Deutschmann, D. (1991). Factors
Thelen, E., & Ulrich, B.D. (1991). Hidden skills: A influencing speed and precision of cursor positioning
dynamic systems analysis of treadmill stepping using a mouse. Ergonomics, 34, 161-174.
during the first year. Monographs of the Society for Trbovich, P., & Harbluk, J.L. (2003). Cell phone comm-
Research in Child Development, Serial No. 223, 56(1). unication and driver visual behaviour: The impact of
Thomas, J.R. (1980). Acquisition of motor skills: cognitive distraction. Conference on Computer-Human
Information processing differences between children Interactions, April 8, 2003, Fort Lauderdale, FL.
and adults. Research Quarterly for Exercise and Treisman, A.M. (1969). Strategies and models of selective
Sport, 51, 158-173. attention. Psychological Review, 76, 282-299.
Thomas, J.R. (1997). Motor behavior. In J.D. Massengale Tremblay, L., & Proteau, L. (1998). Specificity of practice:
& R.A. Swanson (Eds.), The history of exercise and The case of powerlifting. Research Quarterly for
sport science (pp. 203-292). Champaign, IL: Human Exercise and Sport, 69, 284-289.
Kinetics.
Tresilian, J.R. (1995). Perceptual and cognitive processes
Thomas, J.R. (2000). Children's control, learning, and in time-to-contact estimation: Analysis of prediction-
performance of motor skills. Research Quarterly for motion and relative judgment tasks. Perception &
Exercise and Sport, 71, 1-9. Psychophysics, 57, 231-245.
Thomas, J.R., Nelson, J.K., & Silverman, S.J. (2005). Tresilian, J.R. (1997). Correcting some misperceptions of
Research methods in physical activity (5th ed.). time-to-collision: A critical note. Perception, 26, 229-
Champaign, IL: Human Kinetics. 236.
Thomas, J.R., Thomas, K.T., & Gallagher, J.D. (1993). Tresilian, J.R. (1999). Visually timed action: Time-out for
Developmental considerations in skill acquisition. "tau"? Trends in Cognitive Sciences, 3, 301-310.
In R.N. Singer, M. Murphey, & L.K. Tennant (Eds.),
Trevarthen, C.B. (1968). Two mechanisms of vision in
Handbook of research on sport psychology (pp. 73-
primates. Psychologische Forschung, 31, 299-337.
105). New York: Macmillan.
Trowbridge, M.H., & Cason, H. (1932). An experimental
Thorndike, E.L. (1906). The principles of teaching, based study of Thorndike's theory of learning. Journal of
on psychology. New York: Seiler. General Psychology, 7, 245-260.
Thorndike, E.L. (1914). Educational psychology. New Trumbo, D., Ulrich, L., & Noble, M.E. (1965). Verbal
York: Columbia University. coding and display coding in the acquisition and re-
Thorndike, E.L. (1927). The law of effect. American tention of tracking skill. Journal of Applied Psycho-
Journal of Psychology, 39, 212-222. logy, 49, 368-375.
Thorndike, E.L., & Woodworth, R.S. (1901). The Tsutsui, S., & Imanaka, K. (2003). Effect of manual guid-
influence of improvement in one mental function ance on acquiring a new bimanual coordination
upon the efficiency of other functions. Psychological pattern. Research Quarterly for Exercise and Sport, 74,
Review, 8, 247-261. 104-109.
Tiffin, L & Rogers, H.B. (1943). The selection and Tsutsui, S., Lee, T.D., & Hodges, N.J. (1998). Contextual
training of inspectors. Personnel, 22, 3-20. interference in learning new patterns of bimanual
Tipper, S.P., Lortie, C., & Baylis, G.C. (1992). Selective coordination. Journal of Motor Behavior, 30, 151-157.
reaching: Evidence for action-centered attention. Tubbs, M.E. (1986). Goal setting: A meta-analytic exami-
Journal of Experimental Psychology: Human nation of empirical evidence. Journal of Applied
Perception and Performance, 18, 891-905. Psychology, 71, 474-483.
 References 557

Tuffiash, M., Roring, R.W., & Ericsson, K.A. (2007). Underwood, G., & Everatt, J. (1996). Automatic and
Expert performance in SCRABBLE: Implications for controlled information processing: The role of
the study of the structure and acquisition of complex attention in the processing of novelty. In 0. Neumann
skills. Journal of Experimental Psychology: Applied, & AF. Sanders (Eds.), Handbook of perception and
13, 124-134. action. Vol. 3: Attention (pp. 185-227). San Diego:
Tuller, B., & Kelso, J.A.S. (1989). Environmentally- Academic Press.
specified patterns of movement coordination in Ungerleider, LG., & Mishkin, M. (1982). Two cortical
normal and split-brain subjects. Experimental Brain visual systems. In D.J. Ingle, M.A. Goodale, & R.J.W.
Research, 75, 306-316. Mansfield (Eds.), Analysis of visual behavior (pp.
Turvey, M.T. (1977). Preliminaries to a theory of action 549-586). Cambridge, MA: MIT Press.
with reference to vision. In R. Shaw & J. Bransford Vallbo, AB. (1974). Human muscle spindle discharge
(Eds.), Perceiving, acting, and knowing (pp. 211-265). during isometric voluntary contractions: Amplitude
Hillsdale, NJ: Erlbaum. relations between spindle frequency and torque. Acta
Physiologica Scandinavica, 90, 319-336.
Turvey, M.T. (1990). Coordination. American Psycho-
logist, 45, 938-953. Valls-Sole, J., Kumru, H., & Kofler, M. (2008). Interaction
between startle and voluntary reactions in humans.
Turvey, M.T., & Carello, C. (1996). Dynamics of
Experimental Brain Research, 187, 497-507.
Bernstein's level of synergies. In M.L. Latash & M.T.
Turvey (Eds.), Dexterity and its development (pp. Valls-Sole, J., Rothwell, J.C., Goulart, F., Cossu, G., &
339-376). Mahwah, NJ: Erlbaum. Munoz, E. (1999). Patterned ballistic movements
triggered by a startle in healthy humans. Journal of
Turvey, M.T., & Fonseca, S. (2009). Nature of motor
Physiology, 516, 931-938.
control: Perspectives and issues. In D. Sternad
(Ed.), Progress in motor control (pp. 93-123). Berlin: van Boxtel, G.J.M., & Band, G.P.H. (2000). Inhibitory
Springer. motor control in stop paradigms: Reply to McGarry
and Franks (2000). Acta Psychologica, 105, 79-82.
Turvey, M.T., Rosenblum, L.D., Schmidt, R.C., & Kugler,
P.N. (1986). Fluctuations and phase symmetry Van Caekenberghe, I., Segers, V., De Smet, K., Aerts,
in coordinated rhythmic movements. Journal of P., & De Clercq, D. (2010). Influence of treadmill
Experimental Psychology: Human Perception and acceleration on actual walk-to-run transition. Gait &
Performance, 12, 564-583. Posture, 31, 52-56.
Turvey, M.T., Schmidt, R.C., & Rosenblum, L.D. (1989). Van de Crommert, H.W.A.A., Mulder, T., & Duysens, J.
"Clock" and "motor" components in absolute coordi- (1998). Neural control of locomotion: Sensory control
nation of rhythmic movements. Neuroscience, 33, 1-10. of the central pattern generator and its relation to
treadmill training. Gait & Posture, 7, 251-263.
Tyldesley, D.A., & Whiting, H.T.A. (1975). Operational
Vander Linden, D.W., Cauraugh, J.H., & Greene, T.A.
timing. Journal of Human Movement Studies, 1, 172-
(1993). The effect of frequency of kinetic feedback
177.
on learning an isometric force production task in
Ulrich, B.D., & Reeve, T.G. (2005). Studies in motor nondisabled subjects. Physical Therapy, 73, 79-87.
behavior: 75 years of research in motor development,
van der Meijden, O.A.J., & Schijven, M.P. (2009). The
learning, and control. Research Quarterly for Exercise
value of haptic feedback in conventional and robot-
and Sport, 76 (Suppl. 2), S62-S70.
assisted minimal invasive surgery and virtual reality
Ulrich, R., & Wing, AM. (1991). A recruitment theory of training: A current review. Surgical Endoscopy, 23,
force-time relations in the production of brief force 1180-1190.
pulses: The parallel force unit model. Psychological
van Dijk, H., Mulder, T., & Hermens, H.J. (2007). Effects
Review, 98, 268-294.
of age and content of augmented feedback on learning
Umilta, C., & Nicoletti, R. (1990). Spatial stimulus- an isometric force-production task. Experimental
response compatibility. In R.W. Proctor & T.G. Reeve Aging Research, 33, 341-353.
(Eds.), Stimulus-response compatibility (pp. 89-116).
van Donkelaar, P. (1999). Pointing movements are
Amsterdam: Elsevier.
affected by size-contrast illusions. Experimental
Umilta, C., Priftis, K., & Zorzi, M. (2009). The spatial Brain Research, 125, 517-520.
representation of numbers: Evidence from neglect Van Rossum, J.H.A. (1990). Schmidt's schema theory:
and pseudoneglect. Experimental Brain Research, The empirical base of the variability of practice
192, 561-569. hypothesis. A critical analysis. Human Movement
Underwood, B.J. (1957). Interference and forgetting. Science, 9, 387-435.
Psychological Review, 64, 49-60. Vanvenckenray, J., Buekers, M.J., Mendes, R.S., & Helsen,
Underwood, B.J. (1975). Individual differences as a W.F. (1999). Relearning movements: Modifications of
crucible in theory construction. American Psycholo- an incorrectly timed reversal movement. Perceptual
gist, 30, 128-134. and Motor Skills, 89, 195-203.
558 References

Vaughan, C.L., & O'Malley, M.J. (2005). Froude physiology of animals and man (Vol. 1) (pp. 3-32).
and the contribution of naval architecture to our London: Methuen. (Original work published in 1937)
understanding of bipedal locomotion. Gait & Posture, von Holst, E., & Mittelstaedt, H. (1950). Das
21, 350-362. reafferenzprinzip [The reafference principle].
Verbruggen, F., & Logan, G.D. (2008). Response Naturwissenschaften, 37, 464-476.
inhibition in the stop-signal paradigm. Trends in Vorberg, D., & Wing, A. (1996). Modeling variability
Cognitive Sciences, 12, 418-424. and dependence in timing. In H. Heuer & S.W. Keele
Vercher, J.L., Magenes, G., Prablanc, C., & Gauthier, (Eds.), Handbook of perception and action. Vol. 2:
G.M. (1994). Eye-head-hand coordination in pointing Motor skills (pp. 181-262). San Diego: Academic
at visual targets: Spatial and temporal analysis. Press.
Experimental Brain Research, 99, 507-523. Wadman, W.J., Denier van der Gon, J.J., Geuze, R.H., &
Verdolini-Marston, K., & Balota, D.A. (1994). Role of Mol, C.R. (1979). Control of fast goal-directed arm
elaborative and perceptual integrative processes movements. Journal of Human Movement Studies,
in perceptual-motor performance. Journal of 5, 3-17.
Experimental Psychology: Learning, Memory, and Walker, M.P., Brakefield, T., Hobson, J.A., & Stickgold,
Cognition, 20, 739-749. R. (2003). Dissociable stages of human memory
Vereijken, B., van Emmerik, RE.A., Whiting, HT.A., & consolidation and reconsolidation. Nature, 425, 616-
Newell, K.M. (1992). Free(z)ing degrees of freedom in 620.
skill acquisition. Journal of Motor Behavior, 24, 133- Walker, N., Philbin, D.A., & Fisk, AD. (1997). Age-
142. related differences in movement control: Adjusting
Vereijken, B., Whiting, H.T.A., & Beek, W.J. (1992). A submovement structure to optimize performance.
dynamical systems approach to skill acquisition. Journal of Gerontology: Psychological Sciences, 52B,
Quarterly Journal of Experimental Psychology, 45A, P40-P52.
323-344. Wallace, R.J. (1971). S-R compatibility and the idea of a
Verwey, W.B., & Dronkert, Y. (1996). Practicing a response code. Journal of Experimental Psychology,
structured continuous key-pressing task: Motor 88, 354-360.
chunking or rhythm consolidation? Journal of Motor Wallace, S.A. (1996). Dynamic pattern perspective
Behavior, 28, 71-79. of rhythmic movement: An introduction. In H.N.
Vicente, K.J., & Wang, J.H. (1998). An ecological theory Zelaznik (Ed.), Advances in motor learning and
of expertise effects in memory recall. Psychological control (pp. 155-194). Champaign, IL: Human
Review, 105, 33-57. Kinetics.
Vickers, J. (2007). Perception cognition, and decision Wallace, S.A., & Hagler, R.W. (1979). Knowledge of
training. Champaign, IL: Human Kinetics. performance and the learning of a closed motor skill.
Vickers, J.N., Livingston, L.F., Umeris-Bohnert, S., & Research Quarterly, 50, 265-271.
Holden, D. (1999). Decision training: The effects of Wallace, S.A., & Weeks, D.L. (1988). Temporal constraints
complex instruction, variable practice and reduced in the control of prehensile movements. Journal of
delayed feedback on the acquisition and transfer of Motor Behavior, 20, 81-105.
a motor skill. Journal of Sports Sciences, 17, 357-367.
Wallace, S.A., Weeks, D.L., & Kelso, J.A.S. (1990).
Vince, M.A., & Welford, A.T. (1967). Time taken to Temporal constraints in reaching and grasping
change the speed of a response. Nature, 213, 532-533. behavior. Human Movement Science, 9, 69-93.
Vincent, W.J. (2005). Statistics in kinesiology (3rd ed.). Walsh, W.D., Russell, D.G., Imanaka, K., & James, B.
Champaign, IL: Human Kinetics. (1979). Memory for constrained and preselected
Vogt, S. (1995). On relations between perceiving, movement location and distance: Effects of starting
imaging, and performing in the learning of cyclical position and length. Journal of Motor Behavior, 11,
movement sequences. British Journal of Psychology, 201-214.
86, 191-216. Walter, C.B., & Swinnen, S.P. (1990). Asymmetric
von Hofsten, C. (1983). Catching skills in infancy. Journal interlimb interference during the performance of
of Experimental Psychology: Human Perception and a dynamic bimanual task. Brain and Cognition, 14,
Performance, 9, 75-85. 185-200.
von Holst, E. (1954). Relations between the central Walter, C.B., & Swinnen, S.P. (1992). Adaptive tuning of
nervous system and the peripheral organs. British interlimb attraction to facilitate bimanual decoupling.
Journal of Animal Behavior, 2, 89-94. Journal of Motor Behavior, 24, 95-104.
von Holst, E. (1973). On the nature of order in the Walter, C.B., & Swinnen, S.P. (1994). The formation and
central nervous system. In R. Martin (Trans.), The dissolution of "bad habits" during the acquisition
selected papers of Erich von Holst. The behavioural of coordination skills. In S.P. Swinnen, H. Heuer, J.
 References 559

Massion, & P. Casaer (Eds.), Interlimb coordination: lus and response dimensions. Journal of Experimental
Neural, dynamical, and cognitive constraints (pp. Psychology: General, 119, 355-366.
491-513). San Diego: Academic Press. Weeks, D.J., Reeve, T.G., Dornier, L.A., & Fober, G.W.
Walter, C.B., Swinnen, S.P., & Franz, E.A. (1993). (1991). Inter-criterion interval activity and the
Stability of symmetric and asymmetric discrete retention of movement information: A test of the
bimanual actions. In K.M. Newell & D.M. Corcos forgetting hypothesis for contextual interference
(Eds.), Variability and motor control (pp. 359-380). effects. Journal of Human Movement Studies, 20,
Champaign, IL: Human Kinetics. 101-110.
Wann, J.P. (1996). Anticipating arrival: Is the tau Weeks, D.L., & Sherwood, D.E. (1994). A comparison
margin a specious theory? Journal of Experimental of knowledge of results scheduling methods for
Psychology: Human Perception and Performance, promoting motor skill acquisition and retention.
22, 1031-1048. Research Quarterly for Exercise and Sport, 65, 136-
Wann, J.P., & Land, M. (2000). Steering with or without 142.
the flow: Is the retrieval of heading necessary? Trends Weeks, D.J., Zelaznik, H., & Beyak, B. (1993). An
in Cognitive Sciences, 4, 319-324. empirical note on reduced frequency of knowledge
Wann, J.P., & Nimmo-Smith, I. (1990). Evidence against of results. Journal of Human Movement Studies, 25,
the relative invariance of timing in handwriting. 193-201.
Quarterly Journal of Experimental Psychology, 42A, Wegner, D.M. (1994). Ironic processes of mental control.
105-119. Psychological Review, 101, 34-52.
Warren, W.H. Jr. (1998). Visually controlled locomotion: Wegner, D.M. (2009). How to think, say, or do precisely
40 years later. Ecological Psychology, 10, 177-219. the worst thing for any occasion. Science, 325, 48-50.
Warren, W.H. Jr., & Whang, S. (1987). Visual guidance of Wegner, D.M., Ansfield, M.E., & Pilloff, D. (1998). The
walking through apertures: Body-scaled information putt and the pendulum: Ironic effects of the mental
for affordances. Journal of Experimental Psychology: control of action. Psychological Science, 9, 196-199.
Human Perception and Performance, 13, 371-383. Weigelt, M., Kunde, W., & Prinz, W. (2006). End-
Warren, W.H. Jr., & Yaffe, D.M. (1989). Dynamics of state comfort in bimanual object manipulation.
step length adjustment during running: A comment Experimental Psychology, 53, 143-148.
on Patla, Robinson, Samways, and Armstrong Weinberg, D.R., Guy, D.E., & Tupper, R.W. (1964).
(1989). Journal of Experimental Psychology: Human Variation of postfeedback interval in simple motor
Perception and Performance, 15, 618-623. learning. Journal of Experimental Psychology, 67, 98-
Warren, W.H. Jr., Young, D.S., & Lee, D.N. (1986). Visual 99.
control of step length during running over irregular Weinberg, R.S. (1994). Goal setting and performance in
terrain. Journal of Experimental Psychology: Human sport and exercise settings: A synthesis and critique.
Perception and Performance, 12, 259-266. Medicine and Science in Sports and Exercise, 26, 469-
Waterson, P., & Sell, R. (2006). Recurrent themes and 477.
developments in the history of the Ergonomics Weinberg, R.S., & Gould, D. (2011). Foundations of sport
Society. Ergonomics, 49, 743-799. and exercise psychology (S'h ed). Champaign, IL:
Weber, E. (1846). Muskelbewegung [Muscle movement]. Human Kinetics.
In R. Wagner (Ed.), Handworterbuch der Physiologie Weinberg, R.S., & Ragan, J. (1978). Motor performance
(Vol. 3, Pt. 2, pp. 1-122). Braunschweig: Bieweg. under three levels of trait anxiety and stress. Journal
Weeks, D.L., & Anderson, L.P. (2000). The interaction of of Motor Behavior, 10, 169-176.
observational learning with overt practice: Effects on Weir, P.L., Weeks, D.J., Welsh, T.N., Elliott, D., Chua, R.,
motor skill learning. Acta Psychologica, 104, 259-271. Roy, E.A., & Lyons, J. (2003). Influence of terminal
Weeks, D.J., Chua, R., & Elliott, D. (2000). Perceptual- action requirements on action-centered distractor
motor behavior in Down syndrome. Champaign, IL: effects. Experimental Brain Research, 149, 207-213.
Human Kinetics. Weiskrantz, L., Elliott, J., & Darlington, C. (1971).
Weeks, D.L., Hall, AK., & Anderson, L.P. (1996). A Preliminary observations on tickling oneself. Nature,
comparison of imitation strategies in observational 230, 598-599.
learning of action patterns. Journal of Motor Behavior, Weiss, A.D. (1965). The locus of reaction time change
28, 348-358. with set, motivation, and age. Journal of Gerontology,
Weeks, D.L., & Kordus, R.N. (1998). Relative frequency of 20, 60-64.
knowledge of performance and motor skill learning. Welch, J.C. (1898). On the measurement of mental activity
Research Quarterly for Exercise and Sport, 69, 224-230. through muscular activity and the determination
Weeks, D.J., & Proctor, R.W. (1990). Salient-features of a constant of attention. American Journal of
coding in the translation between orthogonal stimu- Physiology, 1, 283-306.
560 mi References

Welford, A.T. (1952). The "psychological refractory of Experimental Psychology: Human Perception and
period" and the timing of high-speed performance-a Performance, 2, 1-13.
review and a theory. British Journal of Psychology, Wickens, C.D. (1980). The structure of attentional
43, 2-19. resources. In R.S. Nickerson (Ed.), Attention and
Welford, A.T. (1968). Fundamentals of skill. London: performance VIII (pp. 239-257). Hillsdale, NJ:
Methuen. Erlbaum.
Welford, A.T. (1984). Psychomotor performance. Annual Wickens, C.D., & Hollands, J.G. (2000). Engineering
Review of Gerontology and Geriatrics, 4, 237-273. psychology and human performance (3rd ed.). Upper
Welford, A.T., Norris, A.H., & Shock, N.W. (1969). Speed Saddle River, NJ: Prentice Hall.
and accuracy of movement and their changes with Wickens, C.D., & McCarley, J.S. (2008). Applied attention
age. Acta Psychologica, 30, 3-15. theory. Boca Raton, FL: CRC Press.
Welsh, T., Almeida, Q., & Lee, T.D. (2005). The effects Wiener, N. (1948). Cybernetics. New York: Wiley.
of spatial orientation between the upper limbs Wightman, D.C., & Lintem, G. (1985). Part-task training
and postural stability on· bimanual coordination. for tracking and manual control. Human Factors, 27,
Experimental Brain Research, 161, 265-275. 267-283.
Welsh, T.N., & Pratt, J. (2008). Actions modulate Wightman, D.C., & Sistrunk, F. (1987). Part-task training
attentional capture. Quarterly Journal ofExperimental strategies in simulated carrier landing final-approach
Psychology, 61, 968-976. training. Human Factors, 29, 245-254.
Weltman, G., & Egstrom, G.H. (1966). Perceptual
Wilberg, R.B. (1990). The retention and free recall of
narrowing in novice divers. Human Factors, 8, 499-
multiple movements. Human Movement Science, 9,
505.
437-479.
Wenderoth, N., & Bock, 0. (2001). Learning of a new
Wilde, H., & Shea, C.H. (2006). Proportional and
bimanual coordination pattern is governed by three
nonproportional transfer of movement sequences.
distinct processes. Motor Control, 5, 23-35.
Quarterly Journal of Experimental Psychology, 59,
Wenderoth, N., Bock, 0., & Krohn, R. (2002). Leaming a 1626-1647.
new bimanual coordination pattern is influenced by
Williams, A.M. (2000). Perceptual skill in soccer:
existing attractors. Motor Control, 6, 166-182.
Implications for talent identification and
Westling, G., & Johansson, R.S. (1984). Factors development. Journal of Sports Sciences, 18, 737-750.
influencing the force control during precision grip.
Williams, A.M., Davids, K., & Williams, J.G. (1999).
Experimental Brain Research, 53, 277-284.
Visual perception and action in sport. London: E &
Westwood, D.A., Dubrowski, A., Carnahan, H., & FNSpon.
Roy, E.A. (2000). The effect of illusory size on force
Williams, A.M., & Grant, A. (1999). Training perceptual
production when grasping objects. Experimental
skill in sport. International Journal of Sport
Brain Research, 135, 535-543.
Psychology, 30, 194-220.
Wetzel, M.C., & Stuart, D.G. (1976). Ensemble
Williams, A.M., & Ward, P. (2003). Perceptual expertise:
characteristics of cat locomotion and its neural
Development in sport. In J.L. Starkes & K.A. Ericsson
control. Progress in Neurobiology, 7, 1-98.
(Eds.), Expert performance in sport: Advances in
White, R.M. Jr., & Schmidt, S.W. (1972). Preresponse research on sport expertise (219-247). Champaign, IL:
intervals versus postinformative feedback intervals Human Kinetics.
in concept identification. Journal of Experimental
Williams, A.M., Ward, P., & Smeeton, N.J. (2004).
Psychology, 94, 350-352.
Perceptual and cognitive expertise in sport:
Whiting, H.T.A. (1969). Acquiring ball skill: A Implications for skill acquisition and performance
psychological interpretation. London: Bell. enhancement. In A.M. Williams & N.J. Hodges
Whiting, HT.A. (Ed.). (1984). Human motor actions: (Eds.), Skill acquisition in sport: Research, theory and
Bernstein reassessed. Amsterdam: Elsevier. practice (pp. 328-347). London: Routledge.
Whiting, HT.A., Bijlard, M.J., & den Brinker, B.P.L.M. Williams, H.G., Woollacott, M.H., & Ivry, R. (1992).
(1987). The effect of the availability of a dynamic Timing and motor control in clumsy children. Journal
model on the acquisition of a complex cyclical action. of Motor Behavior, 24, 165-172.
Quarterly Journal of Experimental Psychology, 39A, Wilson, A.D., & Bingham, G.P. (2008). Identifying the
43-59. information for the visual perception of relative
Whiting, H.T.A., Gill, E.B., & Stephenson, J.M. (1970). phase. Perception & Psychophysics, 70, 465-476.
Critical time intervals for taking in flight information Wilson, A.D., Bingham, G.P., & Craig, J.C. (2003). Propri-
in a ball-catching task. Ergonomics, 13, 265-272. oceptive perception of phase variability. Journal of
Wickens, C.D. (1976). The effects of divided attention on Experimental Psychology: Human Perception and
information processing in manual tracking. Journal Performance, 29, 1179-1190.
 References 561

Wilson, D.M. (1961). The central nervous control of trial. Journal of the American Medical Association,
flight in a locust. Journal of Experimental Biology, 38, 296, 2095-2104.
471-490. Wolf, S.L., Winstein, C.J., Miller, J.P., Thompson, P.A.,
Wimmers, R.H., Beek, P.J., & van Wieringen, P.C.W. Taub, E., Uswatte, G., Morris, D., Blanton, S., Nichols-
(1992). Phase transitions in rhythmic tracking Larsen, D., & Clark, P.C. (2008). Retention of upper
movements: A case of unilateral coupling. Human limb function in stroke survivors who have received
Movement Science, 11, 217-226. constraint-induced movement therapy: The EXCITE
Wing, A.M. (1980). The long and short of timing in randomised trial. Lancet Neurology, 7, 33-40.
response sequences. In G.E. Stelmach & J. Requin Wolpert, D.M., & Ghahramani, Z. (2000). Computational
(Eds.), Tutorials in motor behaviour (pp. 469-486). principles of movement neuroscience. Nature
Amsterdam: Elsevier. Neuroscience, 3, 1212-1217.
Wing, A.M. (2002). Voluntary timing and brain function: Wolpert, D.M., Ghahramani, Z., & Flanagan, J.R. (2001).
An information processing approach. Brain and Perspectives and problems in motor learning. Trends
Cognition, 48, 7-30. in Cognitive Sciences, 5, 487-494.
Wing, A.M., & Fraser, C. (1983). The contribution of the Wolpert, D.M., Miall, R.C., & Kawato, M. (1998). Internal
thumb to reaching movements. Quarterly Journal of models in the cerebellum. Trends in Cognitive
Experimental Psychology, 35A, 297-309. Sciences, 2, 338-347.
Wing, A.M., & Kristofferson, A.B. (1973a). The timing of Woodman, T., & Hardy, L. (2001). Stress and anxiety. In
interresponse intervals. Perception & Psychophysics, R.N. Singer, H.A. Hausenblas, & C.M. Janelle (Eds.),
13, 455-460. Handbook of sport psychology (2nd ed.) (pp. 290-
318). New York: Wiley.
Wing, A.M., & Kristofferson, A.B. (1973b ). Response
delays and the timing of discrete motor responses. Woodrow, H. (1932). The effect of rate of sequence
Perception & Psychophysics, 14, 5-12. upon the accuracy of synchronization. Journal of
Experimental Psychology, 15, 357-379.
Wing, A.M., Turton, A., & Fraser, C. (1986). Grasp size
and accuracy of approach in reaching. Journal of Woods, T. (2001). How I play golf. New York: Warner
Motor Behavior, 18, 245-260. Books.
Winstein, C.J. (1991). Designing practice for motor Woodworth, R.S. (1899). The accuracy of voluntary
learning: Clinical implications. In M.J. Lister (Ed.), movement. Psychological Review Monographs, 3
Contemporary management of motor control (Whole No. 13).
problems: Proceedings of the II step conference (pp. Woodworth, R.S. (1903). Le mouvement. Paris: Doin.
65-76). Alexandria: Foundation for Physical Therapy. Woodworth, R.S. (1938). Experimental psychology. New
Winstein, C.J., & Garfinkel, A. (1989). Qualitative York: Holt.
dynamics of disordered human locomotion: A Woodworth, R.S., & Schlosberg, H. (1954). Experimental
preliminary investigation. Journal of Motor Behavior, psychology (rev. ed.). New York: Holt, Rinehart &
21, 373-391. Winston.
Winstein, C.J., Pohl, P.S., & Lewthwaite, R. (1994). Effects Woollacott, M.H., & Jensen, J.L. (1996). Posture and
of physical guidance and knowledge of results on locomotion. In H. Heuer & S.W. Keele (Eds.),
motor learning: Support for the guidance hypothesis. Handbook of perception and action. Vol. 2: Motor
Research Quarterly for Exercise and Sport, 65, 316-323. skills (pp. 333-403). San Diego: Academic Press.
Winstein, C.J., & Schmidt, R.A. (1990). Reduced Woollacott, M.H., & Shumway-Cook, A. (2002). Attention
frequency of knowledge of results enhances motor and the control of posture and gait: A review of an
skill learning. Journal of Experimental Psychology: emerging area of research. Gait & Posture, 16, 1-14.
Learning, Memory and Cognition, 16, 677-691. Woollacott, M.H., & Sveistrup, H. (1994). The develop-
Wishart, L.R., Lee, T.D., Murdoch, J.E., & Hodges, N.J. ment of sensorimotor integration underlying pos-
(2000). Effects of aging on automatic and effortful ture control in infants during the transition to
processes in bimanual coordination. Journal of independent stance. In S.P. Swinnen, H. Heuer, J.
Gerontology: Psychological Sciences, 55B, P85-P94. Massion, & P. Casaer (Eds.), Interlimb coordination:
Wolf, S.L., Blanton, S., Baer, H., Breshears, J., & Butler, Neural, dynamical, and cognitive constraints (pp.
A.J. (2002). Repetitive task practice: A critical review 371-389). San Diego: Academic Press.
of constraint-induced movement therapy in stroke. Worringham, C.J. (1992). Some historical roots of
Neurologist, 8, 325-338. phenomena and methods in motor behavior research.
Wolf, S.L., Winstein, C.J., Miller, J.P., Taub, E., Uswatte, In G.E. Stelmach & J. Requin (Eds.), Tutorials in motor
G., Morris, D., Guiliani, C., Light., K.E., & Nichols- behavior II (pp. 807-825). Amsterdam: Elsevier.
Larsen, D. (2006). Effect of constraint-induced Wraga, M., Creem, S.H., & Proffitt, D.R. (2000).
movement therapy on upper extremity function 3 to 9 Perception-action dissociations of a walkable Muller-
months after stroke: The EXCITE randomized clinical Lyer configuration. Psychological Science, 11, 239-243.
562 References

Wright, C.E. (1990). Generalized motor programs: Wrisberg, C.A., & Schmidt, R.A. (1975). A note on motor
Reexamining claims of effector independence learning without post-response knowledge of results.
in writing. In M. Jeannerod (Ed.), Attention and Journal of Motor Behavior, 7, 221-225.
performance XIII (pp. 294-320). Hillsdale, NJ: Wu, W.F.W. (2007). Self-control of learning multiple
Erlbaum. motor skills. Unpublished doctoral thesis, Louisiana
Wright, C.E., & Meyer, D.E. (1983). Conditions for a State University.
linear speed-accuracy trade-off in aimed movements. Wulf, G. (1991). The effect of type of practice on motor
Quarterly Journal of Experimental Psychology, 35A, learning in children. Applied Cognitive Psychology,
279-296. 5, 123-134.
Wright, D.L. (1991). The role of intertask and intratask Wulf, G. (2007). Attention and motor skill learning.
processing in acquisition and retention of motor Champaign, IL: Human Kinetics.
skills. Journal of Motor Behavior, 23, 139-145.
Wulf, G., Horger, M., & Shea, C.H. (1999). Benefits of
Wright, D.L., Black, C., Park, J.-H., & Shea, C.H.
blocked over serial feedback on complex motor skill
(2001). Planning and executing simple movements:
learning. Journal of Motor Behavior, 31, 95-103.
Contributions of relative-time and overall-duration
specification. Journal of Motor Behavior, 33, 273-285. Wulf, G., & Lee, T.D. (1993). Contextual interference in
movements of the same class: Differential effects on
Wright, D.L., Li, Y., & Whitacre, C. (1992). The
program and parameter learning. Journal of Motor
contribution of elaborative processing to the
Behavior, 25, 254-263.
contextual interference effect. Research Quarterly for
Exercise and Sport, 63, 30-37. Wulf, G., Lee, T.D., & Schmidt, R.A. (1994). Reducing
knowledge of results about relative versus absolute
Wright, D.L., & Shea, C.H. (1991). Contextual
timing: Differential effects on learning. Journal of
dependencies in motor skills. Memory & Cognition,
Motor Behavior, 26, 362-369.
19, 361-370.
Wright, D.L., & Shea, C.H. (1994). Cognition and Wulf, G., McConnel, N., Gartner, M., & Schwarz, A.
motor skill acquisition: Contextual dependencies. (2002). Feedback and attentional focus: Enhancing
In C.R. Reynolds (Eds.), Cognitive assessment: A the learning of sport skills through external-focus
multidisciplinary perspective (pp. 89-106). New feedback. Journal of Motor Behavior, 34, 171-182.
York: Plenum Press. Wulf, G., McNevin, N.H., & Shea, C.H. (2001). The
Wright, D.L., & Shea, C.H. (2001). Manipulating automaticity of complex motor skill learning as a
generalized motor program difficulty during blocked function of attentional focus. Quarterly Journal of
and random practice does not affect parameter Experimental Psychology, 54A, 1143-1154.
learning. Research Quarterly for Exercise and Sport, Wulf, G., & Prinz, W. (2001). Directing attention to
72, 32-38. movement effects enhances learning: A review.
Wright, D.L., Smith-Munyon, V.L., & Sidaway, B. (1997). Psychonomic Bulletin & Review, 8, 648-660.
How close is too close for precise knowledge of Wulf, G., & Schmidt, R.A. (1989). The learning of
results? Research Quarterly for Exercise and Sport, generalized motor programs: Reducing the relative
68, 172-176. frequency of knowledge of results enhances memory.
Wright, D.L., Snowden, S., & Willoughby, D. (1990). Journal of Experimental Psychology: Learning,
Summary KR: How much information is used from Memory and Cognition, 15, 748-757.
the summary? Journal of Human Movement Studies, Wulf, G., & Schmidt, R.A. (1994a). Contextual-
19, 119-128. interference effects in motor learning: Evaluating
Wrisberg, C.A. (1991). A field test of the effect of a KR-usefulness hypothesis. In J.R. Nitsch & R.
contextual variety during skill acquisition. Journal of Seiler (Eds.), Movement and sport: Psychological
Teaching in Physical Education, 11, 21-30. foundations and effects. Vol. 2. Motor control and
motor learning (pp. 304-309). Sankt Augustin,
Wrisberg, C.A., & Anshel, M.H. (1993). A field test of the
Germany: Academia Verlag.
activity-set hypothesis for warm-up decrement in an
open skill. Research Quarterly for Exercise and Sport, Wulf, G., & Schmidt, R.A. (1994b). Feedback-induced
64, 39-45. variability and the learning of generalized motor
Wrisberg, C.A., & Liu, Z. (1991). The effect of contextual programs. Journal of Motor Behavior, 26, 348-361.
variety on the practice, retention, and transfer of an Wulf, G., & Schmidt, R.A. (1996). Average KR degrades
applied motor skill. Research Quarterly for Exercise parameter learning. Journal of Motor Behavior, 28,
and Sport, 62, 406-412. 371-381.
Wrisberg, C.A., & Ragsdale, M.R. (1979). Further tests of Wulf, G., & Schmidt, R.A. (1997). Variability of practice
Schmidt's schema theory: Development of a schema and implicit motor learning. Journal of Experimental
rule for a coincident timing task. Journal of Motor Psychology: Learning, Memory, and Cognition, 23,
Behavior, 11, 159-166. 987-1006.
 References 111 563

Wulf, G., Schmidt, R.A., & Deubel, H. (1993). Reduced Young, D.E., & Schmidt, RA. (1992). Augmented
feedback frequency enhances generalized motor kinematic feedback for motor learning. Journal of
program learning but not parameterization learning. Motor Behavior, 24, 261-273.
Journal of Experimental Psychology: Leaming, Young, M.S., & Stanton, N.A. (2007). Back to the future:
Memory, and Cognition, 19, 1134-1150. Brake reaction times for manual and automated
Wulf, G., & Shea, C.H. (2002). Principles derived from the vehicles. Ergonomics, 50, 46-58.
study of simple skills do not generalize to complex Zaal, F.T.J.M., Bingham, G.P., & Schmidt, RC. (2000).
skill learning. Psychonomic Bulletin & Review, 9, Visual perception of mean relative phase and phase
185-211. variability. Journal of Experimental Psychology:
Wulf, G., & Shea, C.H. (2004). Understanding the role Human Perception and Performance, 26, 1209-1220.
of augmented feedback: The good, the bad and the Zaichkowsky, L.D., & Baltzell, A. (2001). Arousal and
ugly. In AM. Williams & N.J. Hodges (Eds.), Skill performance. In RN. Singer, H.A. Hausenblas, &
Acquisition in sport: Research, theory and practice C.M. Janelle (Eds.), Handbook of sport psychology
(pp. 121-144). London: Routledge. (2nd ed.) (pp. 319-339). New York: Wiley.
Wulf, G., & Toole, T. (1999). Physical assistance devices Zanone, P.-G., & Kelso, J.A.S. (1992). Evolution of
in complex motor skill learning: Benefits of a self- behavioral attractors with learning: Nonequilibrium
controlled practice schedule. Research Quarterly for phase transitions. Journal of Experimental
Exercise and Sport, 70, 265-272. Psychology: Human Perception and Performance,
Wulf, G., Weigelt, M., Poulter, D.R., & McNevin, N.H. 18, 403-421.
(2003). Attentional focus on supra-postural tasks Zanone, P.-G., & Kelso, J.A.S. (1994). The coordination
affects balance learning. Quarterly Journal of dynamics of learning: Theoretical structure and
Experimental Psychology, 56A, 1191-1211. experimental agenda. In S.P. Swinnen, H. Heuer, J.
Yamanishi, J., Kawato, M., & Suzuki, R. (1980). Two Massion, & P. Casaer (Eds.), Interlimb coordination:
coupled oscillators as a model for the coordinated Neural, dynamical, and cognitive constraints (pp.
finger tapping by both hands. Biological Cybernetics, 461-490). San Diego: Academic Press.
37, 219-225. Zanone, P.-G., & Kelso, J.A.S. (1997). The coordination
Yantis, S. (1993). Stimulus-driven attention capture. dynamics of learning and transfer: Collective
Current Directions in Psychological Science, 2, 156-161. and component levels. Journal of Experimental
Yao, W., Fischman, M.G., & Wang, Y.T. (1994). Motor skill Psychology: Human Perception and Performance,
acquisition and retention as a function of average 23, 1454-1480.
feedback, summary feedback, and performance Zavala, A., Locke, E.A., Van Cott, H.P., & Fleishman, E.A.
variability. Journal of Motor Behavior, 26, 273-282. (1965). The analysis of helicopter pilot performance
Yardley, L., Gardner, M., Leadbetter, A., & Lavie, N. (Tech. Rep. AIR-E-29-6/65-TR). Washington, DC:
(1999). Effect of articulatory and mental tasks on American Institutes for Research.
postural control. Neuroreport, 10, 215-219. Zehr, E.P. (2005). Neural control of rhythmic human
Yerkes, RM., & Dodson, J.D. (1908). The relation of movement: The common core hypothesis. Exercise
strength of stimulus to rapidity of habit-formation. and Sport Sciences Reviews, 33, 54-60.
Journal of Comparative Neurology and Psychology, Zehr, E.P., & Stein, R.B. (1999). What functions do
18, 459-482. reflexes serve during human locomotion? Progress in
Yiu, E. M-L., Verdolini, K., & Chow, L.P.Y. (2005). Neurobiology, 58, 185-205.
Electromyographic study of motor learning for a Zelaznik, H.N., & Hahn, R. (1985). Reaction time
voice production task. Journal of Speech, Language, methods in the study of motor programming: The
and Hearing Research, 48, 1254-1268. precuing of hand, digit, and duration. Journal of
Young, D.E., Cohen, M.J., & Husak, W.S. (1993). Motor Behavior, 17, 190-218.
Contextual interference and motor skill acquisition: Zelaznik, H.N., Hawkins, B., & Kisselburgh, L. (1983).
On the processes that influence retention. Human Rapid visual feedback processing in single-aiming
Movement Science, 12, 577-600. movements. Journal of Motor Behavior, 15, 217-236.
Young, D.E., & Schmidt, RA. (1990). Units of motor Zelaznik, H.N., Mone, S., McCabe, G.P., & Thaman,
behavior: Modifications with practice and feedback. C. (1988). Role of temporal and spatial precision
In M. Jeannerod (Ed.), Attention and performance in determining the nature of the speed-accuracy
XIII (pp. 763-795). Hillsdale, NJ: Erlbaum. trade-off in aimed-hand movements. Journal of
Young, D.E., & Schmidt, RA. (1991). Motor programs as Experimental Psychology: Human Perception and
units of movement control. In N.I. Badler, B.A. Barsky, Performance, 14, 221-230.
& D. Zeltzer (Eds.), Making them move: Mechanics, Zelaznik, H.N., Schmidt, R.A., & Gielen, C.C.A.M.
control, and animation of articulated figures (pp. 129- (1986). Kinematic properties of rapid aimed hand
155). San Mateo, CA: Morgan Kaufmann. movements. Journal of Motor Behavior, 18, 353-372.
564 References

Zelaznik, H.N., Shapiro, D.C., & McColsky, D. (1981). Whetter, E. (2005). Timing variability in circle drawing
Effects of a secondary task on the accuracy of single and tapping: Probing the relationship between event
aiming movements. Journal of Experimental Psycho- and emergent timing. Journal of Motor Behavior, 37,
logy: Human Perception and Performance, 7, 1007- 395-403.
1018.
Zelaznik, H.N., & Spring, J. (1976). Feedback in response
Zelaznik, H.N., Shapiro, D.C., & Newell, K.M. (1978). On recognition and production. Journal of Motor
the structure of motor recognition memory. Journal Behavior, 8, 309-312.
of Motor Behavior, 10, 313-323.
Zelaznik, H.N., Spencer, R.M., & Doffin, J.G. (2000). Zernicke, R.F., & Roberts, E.M. (1978). Lower extremity
Temporal precision in tapping and circle drawing forces and torques during systematic variation of
movements at preferred rates is not correlated: non-weight bearing motion. Medicine and Science in
Further evidence against timing as a general-purpose Sports, 10, 21-26.
ability. Journal of Motor Behavior, 32, 193-199. Zernicke, R.F., & Schneider, K. (1993). Biomechanics
Zelaznik, H.N., Spencer, R.M.C., Ivry, R.B., Baria, A., and developmental neuromotor control. Child
Bloom, M., Dolansky, L., Justice, S., Patterson, K., & Development, 64, 982-1004.
A Battig, William F., 372 Bryan, William Lowe, 6, 430
Abbs, J.H., 169, 170 Baumeister, R.F., 126 Buekers, M.J., 265, 408
Abernethy, B., 148, 321 Bayley, N., 7 Bunz, H., 284
Abolafia, J.M., 73 Baylis, G.C., 120 Burgess-Limerick, R., 148
Abughaduma, R, 383 Beatty, J., 98 Burwitz, L., 356
Ackerman, P.L., 434, 436 Beek, P.J., 288, 436 Butler, M.S., 408
Adams, Jack A., 5, 11, 12, 14, 15, 92, Beevor, C.E., 7 Button, C., 374
160,182,315,356,358,363,383, Beilock, S.L., 124, 125, 127 Byrne, J., 467
426,430,431,438,440,462,472, Bekkering, H., 73
475, 477, 479, 483 Belen'kii, V.J., 172 c
Al-Ameer, H., 376 Benedetti, C., 424 Caird, J.K., 358
Albaret, J.-M., 374 Bennett, S.J., 357 Carello, C., 288
Alimi, A.M., 228, 229 Berkhout, J., 402 Carlsen, A.N., 192
Allard, F., 145, 146 Berkinblit, M.B., 187, 206, 256 Carlton, L.G., 145, 146, 233, 235, 236,
Allport, D.A., 102 Bernstein, Nikolai A., 7, 8, 39, 125, 244,401,418
Almeida, Q.J., 285 190,206,209,452 Carlton, M.J., 236, 402
Ammons, Carol, 10 Berthoz, A., 152 Carnahan, H., 379, 383, 408, 417
Ammons, Robert B., 10, 364, 467, 479 Biguer, B., 265 Carr, T.H., 127
Anderson, J.R., 430 Bilodeau, E.A., 10, 11, 362, 398, 410 Carron, A.V., 365
Anderson, M.C., 125 Bilodeau, l.M., 10, 11, 362, 398, 410, Carter, M.C., 45
Andrews, D.H., 486 412,413,414 Carter, S, 155
Angel, R.W., 173, 191 Bingham, G.P., 287 Cason, H., 406
Ansfield, M.E., 127 Bizzi, E., 180, 198, 252, 254, 257 Catalano, J.F., 369
Antoniou, A., 402 Bjork, R.A., 379, 381, 419 Cattell, James, 6, 125
Antonis, B., 102 Black, C., 370 Caulton, D., 218
Arbib, M.A., 269 Blankenship, W.C., 481 Chabris, C.F., 105
Archer, E.J., 462, 482 Blix,M., 7 Chaffin, D.B., 227
Armstrong, T.R., 213, 214, 386, 420 Bobrow, D.G., 102 Chamberlain, E., 6
Arnold, P., 488 Book, W.F., 6 Chapman, S., 149
Arnold, R.K., 401 Booth, B., 370 Cherry, E.C., 104
Asatryan, D.G., 250 Bootsma, R.J., 147, 148 Chiviacowsky, S., 418, 419
Aschersleben, G., 238 Boucher, L., 69 Christina, R.W., 366
Ascoli, K.M., 475 Bourne, L.E., 362 Chua, R., 192, 285
Ashford, D., 357 Bowen, K.F., 145 Cirstea, M.C., 403
Atkinson, R.C., 92 Boyce, B.A., 352 Clayton, T.M.H., 147
Brakefield, T., 481 Cohen, A., 218
B Bransford, J.D., 390 Cohen, Y., 114
Bachman, J.C., 35, 304, 348 Brebner, J., 72 Condon, S.M., 117
Bachrach, A.J., 127 Brenner, E., 140, 271 Coreas, D.M., 305
Backstein, D., 383 Brenner, J., 405 Cossu, G., 191
Baddeley, A.D., 364, 419 Bridgeman, B., 138 Crago, P.E., 167
Bahrick, H.P., 333, 342, 448 Briggs, G.E., 333, 383, 448 Craik, K.J.W., 10
Balasubramaniam, R., 238 Broadbent, D.E., 100 Cratty, B.J., 14
Baldissera, F., 289 Brogden, W.J., 383 Creamer, L.R., 356
Barrett, N, 146, 147 Brooks, V.B., 16, 196 Creem-Regehr, S.H., 281
Bartlett, C.J., 310 Brown, l.D., 37, 122 Cronbach, L.J., 298
Bartlett, F.C., 193, 207, 441 Brown, J., 91, 93, 472 Crossman, E.R.F.W., 258, 348
Basmajian, J.V., 404 Brown, S., 349 Cunningham, S., 379

II! 565
566 Author Index

D Flatt, A.E., 179 Haken, H., 284

Daffertshofer, A., 238 Flegal, K.E., 125 Halbert, J.A., 235, 236
Dail, T.K., 366 Fleishman, Edwin A., 9, 304, 305, Hall, K.G., 375
Dakin, C.J., 192 306, 307, 308, 309, 310, 311, 315, Hammerton, M., 159
Davis, R., 109 329,383,432,433,436,467,471 Hancock, P.A., 316
Debicki, D., 454 Forssberg, H., 185 Harrison, J.S., 193, 195
Dees, V., 421 Foulke, J.A., 227 Hart, T.J., 454
Defares, P., 130 Fowler, B., 272 Harter, Noble, 6, 430
Del Rey, P., 375 Fozard, J.L., 316 Hasan, Z., 167
Denier van der Gon, J.J., 40, 188, Francis, K.L., 316 Hasbroucq, T., 72
217, 257 Franks, I.M., 77, 192 Hawkins, B., 77, 145, 257
Deutsch, D., 100 Franks, J.J., 390 Hayes, A., 357
Deutsch, J.A., 100 Franz, E.A., 292, 293 Hayes, S.J., 452
Dewhurst, D.J., 161, 164, 167, 168, 202 Fritsch, G., 7 Head, H., 441
Diedrich, F.J., 281 Fuchs, A., 449 Heathcote, A., 349
Diener, H.C., 241 Fukson, 0.I., 187 Hebert, E.P., 376
Dienes, Z., 149 Fukuda, T., 169, 452 Hellebrandt, F.A., 452
Dijkstra, S., 11, 14, 92, 472, 475, 477 Fullerton, G.S., 6 Helsen, W., 263
Dodson, J.D., 128 Hempel, W.E., 432
G Hendrickson, G., 355
Donchin, E., 383
Danders, F.C., 58 Gable, C.D., 415 Henneman, E., 156
Donkelaar, P. van, 77 Galanter, E., 14, 75 Henry, Franklin M., 10, 12, 75, 162,
Donovan, J.J., 362 Gallistel, C.R., 16 163, 165, 193, 195, 196,202,304
Dornier, L.A., 416 Gandevia, S.C., 151 Herman, L.M., 475
Drazin, D.H., 81 Gardner, M., 120 Herrick, C.J., 7
Dubrowski, A., 383 Gelfan, S., 155 Heuer, H., 215, 220, 276, 451
Duck, T., 272 Genovese, E.D., 362, 366 Hick, W.E., 64
Dunbar, K., 104 Gentile, A.M., 395, 400, 430 Higgins, J.R., 173
Gentner, D.R., 220 Higgins, T., 452
E Geuze, R.H., 40, 188, 257 Hill, L.B., 6
Easterbrook, J.A., 129 Ghodsian, D., 215, 276, 451 Hird, J.S., 359, 360
Easton, T.A., 169 Gibson,James,63,78, 140, 142, 152 Hitzig, E., 7
Ebbinghaus, H.D., 341, 464 Gill, E.B., 146 Hobson, J.A., 481
Egstrom, G.H., 129 Girardi, G., 73 Hodges, N.J., 374, 452
Elliott, C.L., 319 Glencross, D.J., 146, 147 Hogan, J.C., 422
Elliott, D., 145, 146, 319, 472 Glover, S., 140 Holding, D.H., 386, 483
Ells, J.G., 476 Goodeve, P.J., 258 Hollands, J.G., 160
Elwell, J.L., 426 Goodman, D., 79, 272 Hollerbach, J.M., 217, 256
English, H.B., 405 Goodwin, G.M., 155 Hollingworth, H.L., 6
Ericsson, K.A., 351 Gopher, D., 102 Holst, E. von, 7, 8, 171
Erwin, C.I., 77 Gordon, G.B., 83, 84 Holt, K.G., 179
Espenschade, A., 7 Gordon, J., 378 Horan, J.J., 359
Evarts, E.V., 16, 167, 172 Gottsdanker, R., 111, 236 Hore, J., 454
Goulart, F., 191 Horn, R.R., 452
F Gracco, V.L., 169 Horsely, V., 7
Fairweather, M., 374 Granit, R., 16 Hoshizaki, L.E.F., 235
Farrell, J.E., 24 Gray, S., 454 Houk, J.C., 16, 156, 166, 167, 253
Farrell, M.J., 488 Greene, P.H., 206, 452 Houtz, S.J., 452
Feldman, A.G., 187, 206, 250, 252, 254 Greenwald, A.G., 111 Howell, M.L., 405
Feltz, D.L., 360 Gribble, P.L., 357 Hoyt, D.F., 280
Ferrier, D., 7 Grillner, S., 16, 184, 185, 205 Hufford, L.E., 383
Fischman, M.G., 408, 415 Grindley, G.C., 421, 426 Hughes, H.C., 69
Fisher, B.E., 378 Grossman, P., 130 Hull, C.L., 9, 14
Fisk, A.D., 107 Guadagnoli, M.A., 416, 447 Hulstijn, W., 405
Fitch, H.L., 206 Gurfinkel, V.S., 172 Hyman, R., 64, 65, 68
Fitts, Paul M., 10, 11, 72, 224, 228, Guthrie, E.R., 3
229,231,298,333,348,430,448,
449 H Immink, M., 370
Flach, J.M., 160 Hahn, R., 79 Irion, A.L., 5, 362, 478
 Author Index 567

Ishigami, Y., 124 Kveraga, K., 69 Mane, A.M., 383

Ivry, R.B., 241, 305 Kyllo, L.B., 352 Marchal Crespo, L., 387
Iwase, H., 228 Marsden, C.D., 165
L Marteniuk, R.G., 273, 422, 476
J Laabs, G.J., 476 Martens, R., 356
Jackson, G.M., 272 Laasko, M., 113 Martin, H.A., 468
Jackson, S.R., 272 Laban, R., 37 Masters, R.S.W., 354
Jaeger, L, 453 LaBerge, D., 83 Mathieson, B., 272
Jagacinski, R.J., 160 Lai, Q., 370 Mattar, A.A.G., 357
James, William, 93, 97, 98, 178, 196, Landers, D.M., 352, 359, 360 Matthews, P.B.C., 155
328 Landin, D., 376 McAllister, D.E., 483
Janelle, C.M., 129, 418 Lang, P.J., 405 McCabe, G., 292
Janis, I., 130 Langer, E.J., 117 McCabe, G.P., 233
Jansen, C., 271 Langfeld, H.S., 277 McCloskey, D.I., 155
Jeannerod, M., 265, 269 Langolf, G.D., 227 McColsky, D., 233
Jeeves, M.A., 78, 79 Lashley, Karl S., 179, 180, 193, 196, McCracken, H.D., 367, 369
Jeka, J.J., 290 198,209 McCullagh, P., 358, 424
Jennings, P., 218 Laszlo, J.I., 180, 181 McDonald, P.V., 117
Johansson, R.S., 168, 203 Latham, G.P., 352 McEvoy, S.P., 123
Jones, M.B., 433 Lavery, J.J., 413, 414, 415 McFarquhar, R., 77
Jones, S., 218 Lavie, N., 120 McGeoch, J.A., 362
Judd, C.H., 6, 355 Leadbetter, A., 120 McGown, C., 257
Lee, David, 140, 141, 147, 151, 152, McGraw, M.B., 7
K 153,370,371 McGuigan, F.J., 421
Kahneman, D., 98, 102, 113, 120, 129 Lee, T.D., 125, 285, 357, 362, 366, McLeod, P., 120, 149
Kalogeras, J., 90 374,378,384,403,408,417,424, McMahon., 114
Karlin, L., 111 447,456 McNevin, N., 354
Kavinsky, R.C., 272 Lee, W., 172 Mechsner, F., 289
Keele, Steve W., 11, 13, 101, 119, 144, Leidl, D., 383 Meijer, O.G., 5
145, 196, 218, 221, 241, 258, 263, Leonard, J.A., 69, 78, 79 Melton, Arthur W., 9
305,476 Lersten, KC., 384, 469 Merkel, J., 64
Kelso, B.A., 227 Leuba, J.H., 6 Merton, P.A., 12, 13, 165, 209
Kelso, J.A.S., 79, 179, 180, 272, 283, Levin, M.F., 403 Meulenbroek, R.G.J., 271
284,286,288,290,456,476 Lewbel, A., 436 Mewhort, D.J.K., 349
Kernodle, M.W., 401, 418 Lewis, D., 483 Meyer, D.E., 250, 259, 260, 358
Kerr, B., 117 Liao, C.M., 354 Meyers, J.L., 468
Kerr, R., 370 Lieberman, D.A., 422 Michon, J.A., 236
Kestenbaum, R., 111 Lin, C.-H., 378 Milgram, P., 146
Ketelaars, M., 77 Lindahl, Lawrence, 39, 402 Miller, G.A., 14, 75
Khan, M.A., 146 Lindeburg, F.A., 481 Milone, F., 475
Kim, Y.H., 263 Lindemann, 0., 73 Minas, S.C., 360
Kirch, M., 138 Liu, J., 422 Mishkin, M., 137
Kisselburgh, L., 145 Locke, E.A., 352, 383 Mohler, B.J., 281
Klapp, S.T., 77, 79 Logan, G.D., 194, 195, 196, 431 Mol, C.R., 40, 188, 257
Klein, R.M., 115, 124 Longman, D.J.A., 364, 419 Mone, S., 233
Kleiner, B.M., 369 Lordahl, D.S., 482 Monsell, S., 77
Knapp, B., 14 Lorge, I., 413, 419 Moray, N., 97
Knoll, R.L., 77 Lortie, C., 120 Morgan, R.L., 371, 372, 373, 374,
Kohl, R.M., 367, 368 Lotter, W.S., 304 376,377,378
Konczak, J., 453 Lough, S., 147 Morris, C.D., 390
Konick, A.F., 11, 14, 475 Morton, H.B., 165
Koning, P., 321 M Mosher, M., 272
Konzem, P.B., 385 MacKay, D.M., 450 Mowbray, C.H., 69
Krampe, R.Th., 351 MacKenzie, C.L., 228 Mowrer, O.H., 80, 81
Kristofferson, A.B., 220, 239 MacLeod, C.M., 104 Mulder, T., 405
Kritikos, A., 272 MacNeilage, P.F., 206 Muller, S., 322
Kunde, W., 276 MacRae, P.S., 316 Munoz, E., 191
Kurtz, T.S., 384 Madalena, J., 472 Murata, A., 228
Kvalseth, T.O., 227 Magill, R.A., 191, 378, 408, 424 Muybridge, Eadweard, 38, 43
568 Author Index

N Prablanc, C., 265 Seeger, C.M., 72

Nacson, J., 479 Pribram, K., 14, 75 Seibel, R., 69
Nagelkerke, P., 77 Proske, U., 151 Seidler, R.D., 341
Namikas, G., 482 Provins, K.A., 180 Serrien, D.J., 290, 403
Nashner, L.M., 152, 165 Ptito, A., 403 Seymour, W.D., 382, 384, 385
Navon, D., 102 Shaffer, L.H., 102, 215
Q Shannon, C.E., 10
Neal, R.J., 321
Neisser, U., 14 Quesada, D.C., 80, 329 Shapiro, D.C., 45, 191, 214, 218, 233,
Nelson, T.O., 341 Quilter, R.E., 316 370,376,420,443,485,486
Neumann, 0., 102, 107, 114, 122, 468 Quinn, J.T., 77, 219, 257, 405 Shea, C.H., 215, 218, 367, 368, 370,
Newell, A., 348 374,376,377,378,389,404,415
R Shea, J.B., 371, 372, 373, 422
Newell, K.M., 235, 236, 244, 401,
402,405,452 Rack, P.M.H., 250 Sherrington, C.S., 7, 179
Newsom-Davis, J., 165 Radosevich, D.J., 362 Sherwood, D.E., 219, 243, 244, 246,
Nichols, T.R., 166, 253 Ragan, J., 128 273,407,408
Nicholson, D.E., 420 Raibert, M.H., 209 Shiffrin, R.M., 92, 105, 106
Nissen, M.J., 83 Reddish, P.E., 147 Shik, M.L., 183
Noble, C.E. (1968), 11, 315 Reeve, T.G., 408 Shirley, M.M., 7
Noble, M.E., 356, 448 Reinkensmeyer, D.J., 387 Sidaway, B., 408
Norman, D.A., 97, 102, 127 Rejall, A.E., 6 Simmons, D.C.V., 123
Norman, J., 137 Reynolds, B., 363 Simon, D.A., 379, 381, 419
Reynolds, P., 102 Simons, D.J., 105
0 Reynolds, S.L., 316 Singer, R.N., 129, 315
Ogden, W.C., 83 Rhoades, M.V., 69 Skoglund, S., 13, 156
Ollis, S., 374 Rich, S., 329, 432, 433 Slater-Hammel, A.T., 194
Orlovskii, G.N., 183 Robinson, C.H., 272 Small, A.M., 69
Osgood, C.E., 483 Rogers, D.E., 75, 196 Smeets, J.B.J., 140, 271
Romaiguere, P., 72 Smith, J., 180
p Rosenbaum, D.A., 74, 77, 79, 271, Smith, J.L., 16, 188
Pal'tsev, Y.I., 172 473,474 Smith, W.M., 145
Park, J.-H., 218 Rosenbloom, P.S., 348 Smith-Munyon, V.L., 408
Parker, J.F., 304, 467, 471 Rossignol, S., 185 Snoddy, G.S., 348, 350, 430
Partridge, M.J., 452 Roth, K., 215, 220 Snyder, C.R., 102
Pashler, H., 116 Rothstein, A.L., 81, 401 Solley, W.H., 353
Passmore, S.R., 125 Rothwell, J.C., 181, 191 Southard, D., 452
Patashnik, 0., 77 Russell, D.G., 438, 440 Southard, D.L., 272
Patrick, J., 475 Ryan, E.D., 468 Sparrow, W.A., 405
Pepper, R.L., 475 Sperling, A., 138
Perkins-Ceccato, N., 125 s Sperling, G., 90
Peters, M., 263 Salmoni, A.W., 81, 397, 419, 421 Spirduso, W.W., 316
Peterson, J.R., 11, 228, 231 Salthouse, T.A., 318 Stanton, N.A., 83
Peterson, L.R., 91, 93, 472 Savelsbergh, G.J.P., 148 Starkes, J.L., 263
Peterson, M.J., 91, 93, 472 Schlosberg, H., 61 Stein, R.B., 186
Pew, R.W., 11, 14, 214, 216, 313, 355, Schmidt, R.A., 77, 80, 83, 84, 191, Stelmach, G.E., 111, 367, 369, 476
447 196,208,210,215,236,237,243, Stephenson, J.M., 146
Phillips, J.R., 402 244,246,255,257,266,276,288, Stephenson, R.W., 311
Pick, H.L, 281 313,329,344,370,381,403,415, Sternberg, Saul, 58, 77
Pigott, R.E., 376 416,420,437,438,440,441,443, Stevens, L.T., 6, 239
Pilloff, D., 127 451,454,475,479,485 Stevenson, M.R., 123
Plamondon, R., 228, 229 Schmidt, R.C., 287 Stickgold, R., 481
Pokorny, R.A., 305 Schneider, D.M., 266 Stuart, D.G., 16, 151, 156
Polanyi, M., 355 Schneider, G.E., 148 Suddon, F.H., 415
Polit, A., 198, 252, 254, 257 Schneider, K., 454 Summers, J.J., 5, 214, 263, 291
Pollock, B.J., 357 Schneider, W., 105, 106, 107, 431 Sweezy, R.W., 486
Posner, M.I., 11, 14, 83, 84, 102, 114, Schroeder, W.H., 355 Swinnen, S.P., 275, 276, 283, 285,
119, 144, 145, 475 Schulman, H.G., 111 290,403,420,422,424,453,471
Possamai:, C., 72 Schumsky, D.A., 398
Poulter, D.R., 354 Seal, J., 72 T
Poulton, E.C., 78, 159 Sears, T.A., 165 Tandy, R.D., 416
 Author Index 569

Taub, E., 180 Vercruyssen, M., 316 Williams, H.G., 305

Taylor, C.R., 280 Vereijken, B., 452 Wilson, D.M., 180
Terzuolo, C.A., 215 Vickers, K.J., 322 Wimmers, R.H., 288
Tesch-Romer, C., 351 Viviani, P., 215 Wing, A.M., 220, 239, 240, 244, 271
Thaman, C., 233 von Holst, E., 7, 8, 171 Winstein, C.J., 378
Thelen, Esther, 317 Wishart, L.R., 318, 379
Thomas, J.R., 359 w Woodward, M., 123
Thompson, W.B., 281 Wadman, W.J., 40, 188, 191, 192, 221, Woodworth, R.S., 6, 61, 141, 143,
Thon, B.,374 257 144, 145,224,269,278,483
Thorndike, E.L., 6, 413, 419, 483 Walker, N., 481 Woollacott, M.H., 165, 305
Thuring, J.P., 217 Wallace, S.A., 476 Worringham, C.J., 6
Tickner, A.H., 123 Walter, C.B., 191, 244, 275, 403, 453 Wright, C.E., 77
Tipper, S.P., 120, 121 Walters, C.E., 452 Wright, D.L., 370, 378, 389, 408, 415
Todd, J.A., 263 Wang, Y.T., 415 Wrisberg, C.A., 422
Toole, T., 376 Warren, W.H., 152, 281 Wu,A.D.,378
Trachtman, D., 380 Waters, L.K., 383 Wulf, G., 215, 354, 381, 404, 418, 419
Trevarthen, C.B., 147
Watts, S., 454
Trowbridge, M.H., 406 y
Weaver, W., 10
Trumbo, D., 356
Weber, E., 7 Yanowitz, B.A., 422
Tsutsui, S., 374
Weeks, D.J., 319 Yao, W., 415
Turvey, M.T., 17, 206, 254, 288, 452
Wegner, D.M., 127 Yardley, L., 120
u Weigelt, M., 276, 354 Yerkes, R.M., 128
Ulrich, L., 356 Weinberg, R.S., 128 Young, D.E., 191, 403, 451, 485
Ulrich, R., 241 Welch, J.C., 117 Young, D.S., 147, 152
Ungerleider, L.G., 137 Welford, A.T., 100, 110, 111, 115, 228 Young, M.S., 83
Upton, G., 422 Welsh, T.N., 285
Weltman, G., 129 z
v Westbury, D.R., 250 Zaal, F.T.J.M., 287
Valls-Sole, J., 191, 192, 221 Westling, G., 168, 203 Zanone, P.G., 456
Van Cott, H.P., 383 Wetzel, M.C., 16 Zavala, A., 383
vander Velden, H., 453 White, J.L., 437, 440 Zehr, E.P., 186
van Donkelaar, P., 77 Whiting, H.T.A., 16, 146, 148, 452 Zelaznik, H.N., 77, 79, 145, 233, 292
van Emmerik, R.E.A., 452 Wickens, C.D., 159, 355 Zernicke, R.F., 454
van Wieringen, P.C.W., 147, 288 Wiener, N., 10 Zuberec, S., 146
Vaughan, J., 271 Williams, A.M., 129, 452 Zuckerman, J., 356
Note: Page numbers followed by an italicized for t refer to the figure or table on that page, respectively.

A movement phases in 141 directed 99

Abilities stroboscopic conditions 146 and distracted driving 99, 122-
collection of 433 visual illusions in 139f 124
correlations among 304 Air force testing 9 dual-task situations 99, 117-120
defined 301-302 Alpha-gamma coactivation 155 as effort 98
in everyday activities 310 Alzheimer's disease 320, 470 filter theories 100-102
generalmotor 302-303 Amnesia 470 fixed capacity theory 100
identified by factor analysis 307- Anticipation 78 flexible allocation of capacity 102
308 with advance information 79 inattention blindness and 104-
as limiting factors 302 costs-benefit analysis 83-85 105
prediction and 303-304 effector 78, 234, 249 introspection of 97, 98
versus skill 302 false 84 limited capacity of 97, 98-99
specificity of 304-305 foreperiods and 80-82 multiple resource theories 102
transfer among 304 implications 82-83 physiological measures of 98
varieties of 302 perceptual 78 pupil diameter and 98
Absolute constant error ( ICE I) receptor 78, 234 selective 97, 99
computation 30-31 spatial (event) 78-80 by stage of information
interpretation 31 temporal 80-82 processing lOlf
Absolute deviation 30 types of 78 Stroop effect 103-104
Absolute error (AE) Anticipatory actions task demands 99
computation 30 controlling whole-body actions temporal coordination and 286-
interpretation 30 151-153 287
versus total variability 30 interceptive control 148-151 theories of 100-103
Absolute retention 463 vision and 148-153 types 97-98
Acceleration 41 Anxiety 127, 128. See also arousal visual 123, 488
Accuracy measures 26, 31-33. See Aptitude. See abilities Attentional focus 124
also speed-accuracy trade-off Arousal choking under pressure 126-127
Acquiring Ball Skill (Whiting) 16 attention and 98 effects in golfers 126f
Acquisition of Skill (Bilodeau) 10 cue-utilization hypothesis 129 internal versus external 124-126
Action-centered interference 120- hypervigilance 129-130 ironic effects and 127
122 inverted U principle 128 practice and 354
Action slips 98 perceptual narrowing 129 Audition 153
Adjacent trial effect 434 and performance 128f, 129f Augmented feedback 394-395. See
Afferent neurons 154, 155 practice specificity and 389 also knowledge of performance
Age 315 Attention 97. See also arousal; (KP); knowledge of results
motor development and 316 attentional focus; interference (KR)
motor performance in older action-selection view of 102-103 associational functions 426
adults 316-318 anxiety and 127-128 classifications 393
practice variability and 370 automatic processing and 105- dimensions of 394t
reaction time effects 318f 108 informational functions 425
Aiming errors 223 change blindness and 104-105 versus inherent feedback 393-394
Aiming tasks "cocktail party problem" 104 learning and 398-400
errors with eyes open versus competition for 98-99, 103 and learning degradation 426-
closed 144f consciousness and 98 427
lights on versus off 144f controlled processing and 105- learning versus performance
manual aiming 143 108 effects 400

Ill 571
572 Subject Index

Augmented feedback (continued) Central intermittency 10 elaborative processing hypothesis

motivational function 425-426 Central nervous system 157 of 377-378
as performance variable 400 Central pattern generators 182-185 generalizability of 374-377
research applications 398 in humans 188 hypotheses about 377-381
research paradigms 395-398 maintaining original pattern 185 learner-contingent scheduling
Autogenetic reflexes 164 Cerebellar damage 7 381-382
Automaticity 12, 69, 107, 120 Change blindness 104-105 learner response to 372
and learning 339-340 Choice reaction time 65f, 66f, 144, nature of material practiced 372
stage oflearning 431 318f order of practice items 372
Automatic movements 12, 69, 107, Chunking 91 performance effects versus
120,339-340,449-450 Closed-loop systems 438. See also learning 381
Automatic processing 98 feedback; feedforward; reflexes practical implications 381
effector level 135 randomized blocks of trials 376-
B error 135 377
Bachman ladder climb task 35, 304, error-correction 438-440 reconstruction hypothesis of 378-
348,455,468 evidence against Adams' theory 380
Backward chaining 383 440 schema theory and 445-446
Balance executive level 135 serial practice 376
attention demands 117 expanded model for movement subject influences 375
control systems 151 control l59f task complexity 374-375
as general ability 304 feedback 135, 438-440 Continuation tasks 237-239
on stabilometer 32, 354 models 136f, l59f, 17lf Continuous tasks 22, 117
vestibular system and 153 movement duration and 158-161 versus discrete tasks in skill
vision and 151-152 origination 15 retention 469-471
Ballistic movements. See rapid reference mechanisms 135 part versus whole practice of 383-
movements reference of correctness 160-161 384
Basal ganglia 319 sensory information sources 136 retention of skill 467-468
Bernstein Perspective 452-455 Closed skills 24 Controlled processing 98
Bias 27 "Cocktail party problem" 104 Coordination 263-264. See also
Bilateral-transfer paradigm 218 Cognitive games 321 temporal coordination
Bimanual coordination. See Cognitive psychology 58 absolute 293
coordination Cognitive Psychology (Neisser) 14 anti-phase pattern 43
Biofeedback 404-405 Compensatory tracking 22 bimanual 271-277
Biomechanics field 4 Complexity continuous 277
Blocked practice psychological refractory period costs and benefits of bimanual
effects on performance and and 111 279
retention 382f reaction time and 77 directional preference 289
practice variability and 370-371 stimulus-response compatibility discrete 264
versus random 371-374 and 74, 196 displacement plots 42f
Bottom-up processing 99 Concept-formation tasks 421 end-state 276-277
Brain activity Consciousness eye-head-hand 264-265
and attention 98 attention and 98 gait transitions 278-282
measures of 47-48 reflexive corrections and 162-164 head to arm in golf 274
during tickling 175 Consolidation 480-481 in-phase 43
Brain damage 7, 138 Constant error (CE) interlimb 293
Brain imaging techniques 47-48 application 28-29 kinematics 41-43
Brain structures 7 computation 27 maintenance tendency 278
interpretation 27 preparatory postural reactions
c units of 27 172-173
Callosotomies, individuals with Constant practice 368, 369, 370, 445 relative 293
319 Constant proportion test 220 relative phase 43
Canadian Society for Psychomotor Constraint-induced therapy 319, 350 social 293-294
Learning and Sport Psychology Construct validity 25 spatial 273, 275-276, 292-293
10 Context-conditioned variability 207 stability 43
Capability. See abilities Context specificity 389 units of action 265-268
Capacity interference 99 Contextual interference. See also upper and lower limb 289-290
Ceiling effects 332-333 practice Coordination kinematics 41-43
Cell phones. See distracted driving blocked versus random practice Coordinative structures 206
Central control mechanisms 182 371-374 Corollary discharge 171, 361
 Subject Index ll'l 573

Correction models Discrete tasks 21, 117 constant error 27-28

Crossman-Goodeve model 258- versus continuous in skill relationships among scores 31
259 retention 469-471 total variability 29-30
optimized submovement 259 and part versus whole practice variable error 27-28
Corrections 257-258 384-385 Errors
consciousness of 162-164 practice distribution and 365-367 in aiming 223
feedback-based 233 retention of skill 468-469 in decision making 87-88
in feedforward systems 173-174 Dissociations 137, 470 with false anticipation 84
identification of 260 Distracted driving 122-123 objective 437
latencies of 161, 162/ cell phone use 123 pedal misapplication example 203
measurement of 260 handheld versus hands-free program execution 202-204
position and velocity traces of 173/ phones 124 program selection 201-202
Correct rejection. See signal- sources of distraction 123-124 in signal detection 87-88, 89/
detection theory Distributed practice. See practice subjective 437
Correlation. See also factor-analytic distribution in timing 234
studies; scattergrams Donders' subtractive method 58, 59 triggered reactions in 204-205
adjacent-trial effects 434 Dorsal rhizotomy 179 Exercise and Sport Sciences Reviews
among skills 304 Double-stimulation paradigm 108 14
intertrial analyses 433-434 Down syndrome 319 Experimental methods 298,
multiple 314 300. See also motor learning
and regression 50 E experiments
remoteness effects 433-434 Ebbinghaus-Titchener illusion 138, Expertise 299
superdiagonal form 434 139/ Exteroceptors 136
Correlation coefficient 53-54 Effector anticipation 78, 249 Eye movements
Cost-benefit analysis Efference copy 171, 172 in eye-head-hand coordination
of anticipation 83 Effort 265
of bimanual coordination 279 learning and 340 measures of 46-47
Criterion tasks measures of 340 saccadic 170-172
in practice versus variable- mental 98 tracking set-up 48/
practice 367, 368/ Elaborative processing hypothesis
simulation and 486 377-378 F
transfer of learning and 481 Electroencephalography (EEG) 47 Factor-analytic studies
Crossed-extensor reflex 185 Electromyography (EMG) 37, 45-46 criticisms of approach 310
Crossman-Goodeve model 258-259 agonist and antagonist activity factor analysis 305-306
Cues 189/ factor loadings 306
discrimination by relevancy 129 triple-burst pattern 189 factor matrix 306, 307/
separation and recall 475-476 Empirical equations 49 motor abilities identified by 307-
Cue-utilization hypothesis 129 determining constants 49-50 308
Cuff technique 181 graphical method 49/ for prediction of job performance
Cutaneous receptors 156-157, 168 interpreting 50 309
Cybernetics (Wiener) 10 for linear relationships 49 role of factors in skills 308, 310
uses of 50 False starts 82
D Engineering psychology 10 Fatigue 9, 362
Deafferentiation 179 Equilibrium-point theories 199 Feedback. See also augmented
animal studies 179-180 evidence for 254-255 feedback; knowledge of results
dorsal rhizotomy 179 extensions of 256-257 (KR)
in humans 180-182 length-tension diagram 250 accumulated 394
sensory neuropathy 181-182 limitations of 257 in closed-loop systems 135
Degrees of freedom a and 'Y models 252-254 concurrent versus guidance 386
coordination in golf putting 274 mass-spring mechanisms 250-252 definitions 393
exploiting 454-455 Ergonomics 7, 10 inherent 393-394
freezing 452 Error-correction. See corrections kinematic 402-404
problem 8, 189-190 Error detection kinetic 405
releasing 452-454 capabilities 374, 436-438, 445 for object manipulation 168-169
reorganizing 452-453 capability development 354 response-produced 178
Deliberate practice 351 in feedforward systems 173-174 and speed-accuracy trade-off 233
Dichotic-listening paradigm 104 Error measures tactile 168-169
Differential approach. See absolute constant error 30-31 video 401-402
individual differences absolute error 30 virtual 282, 488
574 Subject Index

Feedforward 170 triggers for transitions 280, 281 Imagery 359,361,479

efference copy in limb control vision and 281 Implicit speech 361
172 visual sensation and speed 281, Impulse-timing view 210
error detection/ correction 173 282f impulses 210-211
for memory retrieval 75 Gamma efferent motor neurons 154 order of events 211
models of 174 Gamma loop 164-167 phonograph analogy 213
preparatory postural reactions Gearshift analogy 450-451 relative force 213
172-173 Gender 313, 315, 318f relative timing 211-213
saccadic eye movements 170-172 Generalized motor programs Impulse-variability theory 242
self-tickling example 174, 175 (GMPs) and ballistic-timing tasks 244,
Final common path 7 changing parameters of 219-220 245-246
Fitts' law computer model 208 batting and 248-249
applications 228-229 force-duration interaction 218 effect of movement amplitude
bimanual coordination and 271- impulse-timing view 210-213 244-245
273 invariant features 208-210 effect of movement time 245, 248f
defined 224-226 muscle selection parameter 218- force variability 242-244
discrete Fitts' task 228 219 initial impulse modeling 244
exception to 230-231 overall duration parameter 213- limitations of 247, 250
generality of 228-229 216 spatial errors 246-247
importance of 229 overall force parameter 216-218 Inattention blindness 104-105
index of difficulty 226 phonograph analogy 213 Index of difficulty. See Fitts' law
interpretation of 226-227 retention of 471 Individual differences 6, 297. See
modifications 228 General motor ability 12, 302-303 also abilities
origination 11 Gerontology 316 age effects 315-318
relation to speed-accuracy trade- Goal setting 352-353 differential approach 297-298,
offs 227 Golgi tendon organs 7, 156 300-301
Fitts' paradigm 224, 225f Goniometers 43 experimental approach 297, 298,
Fitts' task 224, 228-229 Graphics tablets 43 300
Flight simulators 436, 486f Grasping movements 268-271 expertise 299
Floor effects 332-333 Guidance 386-388 expert versus novice 321-324
FMRI. See functional magnetic learning stages and 434-436
resonance imaging (fMRI) H modular view 305
Force variability 242-244 Habit 328, 430 neurological impairment 318-320
Foreperiods Haken-Kelso-Bunz Model 284-285, reliability and 301
aging 81-82 455-456 studies using 432-433
constant-duration 80 Handwriting 209-210 in timing ability 305
and false starts 82 Haptic feedback 488 Induced-motion paradigm 138
variable-duration 80-81 Heart rate 98 Information feedback. See feedback
Forgetting 379, 462. See also Henry-Rogers experiment 75-77 Information processing. See also
retention loss Henry's specificity hypothesis 304- attention; memory; signal-
Forward models 174 305 detection theory
Functional magnetic resonance Hick's law automatic 105-108
imaging (fMRI) 47-48, 98 exceptions to 69 black box model 57f, 58
Functional stretch reflex 167 intercept 68 chronometric approach of study 58
Functional tuning 199 interpreting 66, 68 computer metaphor 57
slope 68 with consistent-mapping
G Hierarchical control model 205-206, conditions 106-107
Gait 447-452 controlled 105-108
in animals 278-280 Hierarchy of habits 430 expanded model 6lf
energy costs 280 HKB model. See Haken-Kelso-Bunz limited capacity for 37
generator models 184 Model three stages of 58, 60
half-center model 182 Home advantage 389 of unexpected information 103
human research 280-282 Human factors 7, 10 with varied-mapping conditions
oxygen consumption 280f Human Movement Science 14 106-107
patterns 182, 279 Huntington's disease 319 Inherent feedback 393-394
role of vision 281 Hypervigilance 129-130 Inhibiting actions 193
selection 280 anecdotal evidence for 193
stability and 281 anticipatory actions 194-196
transitions 278-282 Iconic memory 471-473 rapid discrete actions 193-194
 Subject Index 575

Inhibition of return (IOR) 114-115 Kinematics 39 362-365

Inner ear 153 acceleration 41 effort and 340
Intelligence 303, 434 coordination 41-43 generalizability as criterion 341-
Interception location analysis 39-41 342
control 148-151 velocity 41 goal-setting effects on 352-353
moving targets 146-147 Kinesthetic sensitivity test 432 habit and 328, 430
spatial information 149-151 Kinetic feedback 405 individual differences and 344,
temporal information 148-149 Knowledge of performance (KP) 431-436
and time to contact 140 395, 400-405 memory and 89-90, 341
Interference 100. See also contextual Knowledge of results (KR) 395. See motor stage 431
interference also KR-scheduling perceptual 355-356
action-centered 120-122 bandwidth 407-408 versus performance effects 351,
capacity 99 erroneous 408-409 381,400
contextual 445-446 in observational learning 358 to performance paradox 377
effect of length of SOA 109, llOf optimal bandwidth 427 "rate" of 345
effects of practice 111 origins of paradigm 396 speed of decision and 340
grouping effect 110 precision 405-406 stages of 430-431
inhibition of return 114-115 qualitative vs. quantitative 406- variables 338, 345
during KR-delay intervals 421- 407 without awareness 355
422, 424 KR-scheduling Learning curves 331
as measure of attention 99 absolute-frequency 409-412 Limb movement accuracy 11
proactive 474, 475 blocked versus random 417 Linear relationships 49
psychological refractory period interference during KR-delay Locomotion. See also gait
108-110 421-422 reflex involvement 185-187
refractoriness for practical interference during post-KR- reflex reversal 170
situations 112-113 delay 424 vision and 152-153
retention loss and 474-475 interpolated activities 421-424 Long-loop reflexes 166-167, l68t
retroactive 474, 475 intertrial interval 421 Long-term memory (LTM) 90, 93
revised single-channel view 115- KR-delay interval 419-421 Looming 63, 141
116 post-KR-delay interval 421
separation between responses relative-frequency 409-412 M
113-114 self-regulated 417-419 Magnet effect 278
single channel hypothesis 100- statistical summaries 416-417 Magnetoencephalography (MEG)
102, 110 subjective estimation during KR- 47
stimulus complexity 111 delay 422-424 Maintenance tendency 278
stimulus-response compatibility summary length 415-416 Man-machine interactions 10
111 temporal locus of 419-421 Mashburn task 32, 483
structural 99 trials delay method 412-415 Massed practice 361
Interference criterion 99 Mass-spring mechanisms 250-252
Internet 18 L The Mathematical Theory of
Interoceptors 136 Ladder climb task 35, 468 Communication (Shannon and
Interresponse interval (IRI) 113 Law of Effect 6, 425 Weaver) 10
Intersensory facilitation 61 Law of practice 347-348 Measurement devices 43-45
Inverted U principle 128 generality of 348-350 Measurement systems
Ironic effects 127 learning progress and 350 objectivity 24
time scales 348 reliability 25
J Lead-up activities 385-386 sensitivity 24
Joint receptors 156 Learning 327-329. See also motor validity 25-26
Journal of Experimental Psychology 10 learning; observational Measures. See also errors; kinematic
Journal of Human Movement Studies 14 learning; schema theory; measures
Journal of Motor Behavior 14 transfer of accuracy 26, 31-33
Journals 14, 17, 18 "amount" of 342 eye movements 46-47
analogies, use of 354 of movement magnitude 35-36
K analogous states to forgetting movement outcome 26
Kinematic feedback 402-404 462t physiological 37
Kinematic measures and automaticity 339-340, 431, of retention 463-465
acceleration 41 449-450 secondary tasks 36-37
location 39-41 cognitive task elements 360 of speed 33-35
velocity 41 distribution-of-practice effects on of time 33-35
576 Subject Index

Mechanical principles, knowledge permanence of 329 force production 242-244

of 355 task difficulty and 447 inhibition 156
Memory 88. See also retention; Motor learning experiments length-tension curve 250, 251.f
short-term memory (STM) between-subject variance 331-332 253f
box theories of 90 ceiling and floor effects 332-333 stiffness 166, 250, 252
decay with time 90 double designs 337 Muscle spindles 154-156
direct versus indirect influences group results 337-338 alpha-gamma coactivation 155
on 89-90 learning variables 338 control of muscle stiffness 166
dissociations 470 performance curves 329-331 equatorial region 154
iconic 471-473 performance measures 331 fibers 154
learning and 89-90, 341 performance variables 338 and gamma loops 164-166
literal storage of information 91 retention designs 336-337 innervation 154
long-term 93 scoring sensitivity 333-334 long-loop reflexes 166-167
motor program retrieval 75 transfer designs 336-337 neuron activation 154
postmovement 473 within-subject variance 332
as process versus result of process Motor memory 461-462 N
467 Motor neurons 154, 155 Negative asynchrony 238
recall experiments 90, 91 Motor programs. See also Negative transfer
and rehearsal 91, 93 Generalized motor programs conditions for 483
search and retrieval 89 (GMPs) experiments 465
span of 91 acquisition of sequencing 450- retroactive design 483-484, 485f
theoretical frameworks 88-89 451 of timing 485-486
types 90 central versus peripheral Neural imaging 48
"Memory Drum" theory 12, 75 invariance and 220-221 Neural networks 182, 183
Memory trace 440 as combinations of reflexes 451- Neurological disorders 320-321
Mental practice 359-361 452 Neurophysiology field 4
Mesencephalic preparation 183-184 creation of 450-452 North American Society for the
Military research 9 decentralized theory 221 Psychology of Sport and
Mirror tracing task 349.f 361, 470 degrees of freedom and 189-190 Physical Activity (NASPSPA)
Modeling. See observational doubts about 221 10
learning errors types 201-205 Novel movements 369, 370, 443,
Monosynaptic stretch reflex 163 hierarchical levels of control 205- 445
Motivation 206 Novelty problem 207-208
augmented feedback and 425-426 movement pattern consistency
goal setting and 352-353 450 0
for learning 351-353 need for revision 208 Objectivity 24
as state of activation 128 novelty problem 207-208 Object manipulation
task importance 352 preparation 75 feedback 168-169
Motor abilities. See abilities rapid movement advance goal postures and 271
Motor commands 437 programming 196, 198 intrinsic and extrinsic object
Motor Control 14 retrieval 75 properties 270
Motor development 7, 315, 316 sensory information and 198-201 reaching-grasping 268-271
Motor equivalence 205 statistical invariance and 220 Observational learning 356
Motor learning. See also learning; storage problem 206-207 of dynamic skills 357
schema theory Motor skills 21. See also abilities knowledge of results and 358-359
Bernstein perspective 452-455 Motor stage 431 model skill level and 357-359
cognitive perspectives on 446- Movement characterization 37-39 scheduling observations 359
447 Movement classification 21 of spatial information 357
cognitive stage 430-431 Movement end point 250 of strategies 356-357
definitions 327, 328 Movement planning 140, 264, 269, of temporal information 357
fixation stage 431 270,271,361 using videotapes 356
Haken-Kelso-Bunz Model 284- Movement time (MT) 34-35 Occlusion methods 322
285, 455-456 Muller-Lyer illusion 139 Open-loop systems 177-178
hierarchical control perspective Multiple correlation 314 Open skills 23
447-452 Multiple regression 314 Optical array 142
individual differences and 431- Muscle fibers 154 Optical flow 63, 78, 137, 142
436 Muscle receptors 154-156 Optic ataxia 138
motor stage 431 Muscles Optimized-submovement models
observability 328-329 elasticity 250 259
 Subject Index 577

Otolith organs 153 group versus individual from testing 315

Output chunking 114 performance 349 Preprogrammed tasks 233
Overlearning 329 guidance and 386-388 Preselection effect 476-477
injury prevention and 387 Proactive interference 474-475
p
instructional set for 353-354 Proactive transfer 466
Panic (hypervigilance) 129-130 knowledge of mechanical Probe technique 37, 117, 118-120
Parallel processing 58, 60, 100 principles 355 Procedural skills 486, 487
Parkinson's disease 319 law of 347-350 Process-dissociation 98
Pattern recognition 61 mental versus physical 359-360 Process orientation 14
Perceptual and Motor Skills Research on- and off-task conditions 351/ Professional journals 14
Exchange 10 part versus whole 382-386 Progression-regression hypothesis
Perceptual anticipation 78 power function 348, 349 448-449
Perceptual learning 355-356, 385 predictability of skilled Proprioception 7. See also closed-
Perceptual narrowing 129 performance and 433 loop systems
Perceptual trace 438 retention and 470-471 Proprioceptors 136, 153
Perceptual training 356, 385 safety and 365, 387 cutaneous 156-157, 168
Performance. See also practice as simplification process 435 ensemble characteristics 157
curves 329-331 temporary versus permanent input to central nervous system
errors 98 improvements 351 157
improvement versus learning therapy as 319, 350 jointreceptors 156
351,381,400 verbal instruction 353-355 muscle receptors 154-156
length of rest period and 362 Practice distribution 361 vestibular system 153
retention testing and 462, 463.f defining terms 361 Psychological journals 14
464, 468.f 469/ discrete versus continuous tasks Psychological refractory period
variables 338, 345 365-367 108-110. See also interference
Performance-learning paradox 377 effects on learning 362-365 Pupil dilation 37, 98, 340
Peripheral vision 129 effects on performance 362 Pursuit rotor task 31, 359, 360
Physical therapy 319, 350 fatigue and 362, 363 Pursuit tracking 22
Plans and the Structure of Behavior length of retention interval 363
(Miller, Galanter, and Pribram) massed 361 R
14, 75 permanence of learning 363-364 Random practice 447
Point of no return 194, 195, 196 safety issues 365 versus blocked 371-374
Point-to-point computation 160 time scale 364 effects on performance and
Polyrhythms 263, 291-292, 384 total practice time 364-365 retention 382/
Ponzo illusion 139 Practice specificity 388 knowledge of results and 417-419
Positive transfer 465 context 389 Rapid movements 161. See also
Positron emission tomography sensory and motor 388-389 speed-accuracy trade-off
(PET) 47 transfer-appropriate processing agonist-antagonist patterning
Posture 389-390 190-191
control of 151-152 Practice variability 367 central control of 188-189
coordination and 271, 276, 294 age of learner and 370 programming in advance 196,
discrimination 308 experimental design for 368t 198
muscle stiffness and 166 generalizability and 369, 370, 371 spatial errors 246
preparatory reactions 172-173 interpreting effects 371 startled actions 191-192
reflexes involved 186 nature of task 370 types of 224
secondary task actions 117, 119, 120 novel tasks and 369 Reaching-grasping 268-271
Potentiometers 43 retention effects 367 Reaction time (RT) 33-35. See also
Power law 228 scheduling 370-371 Hick's law
Practice. See also contextual schema theory and 367, 371, 443 age effects on 318/
interference transfer effects 367-369 choice RT 34, 64, 65.f 66.f 144,
amount 347, 470-471 Precuing technique 79, 83 318/
amount of 347 Prediction 312-313. See also test comparison to other response
blocked versus random 371-374 development types 168t
deliberate 351 from general motor ability notion critical events in 34/
effects on RT delay in double- 303-304 effects of stimulus properties 61
stimulation 111 of individual differences 434 gender effects on 318/
exponential function 349, 350 practice and 433 and information processing 58
fear-reduction 385, 387 from a relationship between movement accuracy effects 77
focus of attention 354-355 variables 52-53 and movement complexity 76, 77
578 Subject Index

Reaction time (RT) 33-35. See also bypassing 79 direction of relationship 51-52
Hick's law (continued) familiarity effects 69 prediction 52
movement duration effects 77 interference 74 strength of relationship 52
and number of movement parts 77 learned associations 69, 74 Schema 14
and S-R alternatives 64 measuring amount of learning 441-443
for varied- and consistent- information 66 types 442-443
mapping 107/ number of alternatives and 64, Schema theory 438
Recall 68t on contextual interference 445-
experiments 90, 91 uncertainty and 66, 67, 68 446
memory 441 Response time 35 error detection 445
schema 442 Retention knowledge-of-results frequency
Receptor anticipation 78 absolute 463 445
Reciprocal inhibition 165, 185 amnesia and 470 limitations of 445-446
Reciprocal innervation 7 amount of practice and 470 on novel movements 443, 445
Recognition memory 441 characteristics of memory 467 practice variability and 367, 371,
Recognition schema 442-443 for continuous tasks 467-468 443
Reconstruction hypothesis 378-380 discrete tasks 468-469 Secondary tasks. See also attention
Reference tests 311 effect of temporary factors on continuous 117-118
Reflexes 462,464 discrete 117
autogenetic 164 of generalized motor programs measures of 36-37
combinations of 451-452 versus parameters 471 probe technique 117, 118-120
cutaneous 186 learning within trials 471 problems with 120
gamma loop 164-167 measuring 463-465 proficiency trade-offs with
latencies in corrections 161, 162/ memory-set effect 473-474 simultaneous tasks 102
in locomotion 185-187 relation to transfer 462 Selective attention 97, 99
long-loop 166-167, l68t relative 463-465 Self-organization theory 284-285
monosynaptic stretch 164-166 variability-of-practice effects 367 Semicircular canals 153
movement control and 169 of verbal-cognitive components Sensory consequences 442, 443
myotatic 168t 469-470 Sensory neurons 154
posture and 186 Retention interval 462 Sensory neuropathy 179, 181-182
prior instructions and 167 Retention loss Sensory receptors 7
role in skilled actions 170 brief postmovement memory Serial processing 58, 60, 106
as a stereotyped response 186 473-477 Serial tasks 23
triggered 167-168 consolidation and 480-481 and part versus whole practice
Reflex-reversal phenomenon 170, cue separation 475-476 382-383
185-186 errors by retention-interval and response-chaining and 178
Refractory period. See interference reinforcements 473/ Set hypothesis 477, 478-480
Regression 50 iconic memory 471-473 Short-term memory (STM) 91-93
Rehearsal 91, 93, 359. See also interference and 474-475 Short-term motor memory (STMM)
practice preselection effect 476-477 14, 92-93, 472
Reinvestment theory 126 repetition spacing 477 Short-term sensory store (STSS)
Relative retention 463-464 warm-up decrement 477-480 90-91
Reliability 25, 301 Retention test 462 versus short-term memory 91-92
Remoteness effect 433-434 Retroactive interference 474-475 trace-decay theory and 472
Repetitive movement-timing tasks Retroactive transfer 466 Signal-detection theory 85
237-242 Rhythm 183,263,290,291,384.See criterion setting 86-87
Research Quarterly 10 also gait decision-making errors 87-88
Response-chaining hypothesis 178- Root mean square error (RMSE) 29 detection sensitivity 88
179, 178.f 182 RT. See reaction time (RT) effects of shifting criterion 89/
Response initiation 60 error reduction 87
Response programming 74-75. See s misses and false alarms 87
also reaction time (RT) Saccades 69,72, 171 strength of sensory information
feedforward 75 Saccule 153 85-86
Henry-Rogers experiment 75-77 Savings score 341 Simon effects 73-74
and movement complexity 76, 77 Scattergrams 50 Simulation
tuning 75 between age and running speed financial savings 467
Response selection 63-64. See also 52/ in medical field 487, 488
Hick's law; stimulus-response between age and running time physical 486-488
compatibility 51/ pilot training 436, 486-487
 Subject Index 579

transfer and 467, 486, 487 stimulus detection 60-61 directional preference 285, 289
in virtual environments 488 Stimulus onset asynchrony (SOA) egocentric preference 285
Single channel theories 10, 111 108 in-phase 282
bottleneck location 100-101, 101- effect of orienting cue and intention and 286-287
102 stimulus locations 114, l16f interlimb pairings 290, 29lf
evidence against 111 effect on refractoriness 109, llOf, perception and 287-289
filters 100-102 112 phase transitions 284
revised view 115-116 in inhibition of return paradigm polyrhythms and 291-292
of undifferentiated capacity 100 ll5f self-organization 284-285
Single-photon emission computed separation between responses spatial orientation 285-286
tomography (SPECT) 47 and 113f, 114 unintended phase transitions
Skin resistance 98 and S-R compatibility 112f 283-284
Ski simulator task 35, 36f, 452, 453f Stimulus-response compatibility upper- and lower-limb 289-290
Social coordination 293-294 69-71, 111 Temporal information
Spacing effect 379, 477 and complex actions 74 about time to contact 140, 148
Spatial anticipation 78-80 end-comfort effect 74 interceptive control and 148-149
Spatial coordination intensity and 72-73 learning through observation 357
biasing 292 population stereotypes and 71-72 Temporal occlusion 322
in bimanual coordination 273, response patterns 71 Test development
275-276 Simon effects 73 criterion variable 313-314
end-state 276-277 SNARC effect 73 posing the problem 313
force outputs 273 spatial and anatomical task analysis 313
novel combination development relationships 72 validation 314-315
292-293 stimulus pattern and 70-71 Tests. See also measurement systems
Spatial information Stop-signal paradigm 194 face valid 25, 313
interceptive control and 149-151 Stress predictor 313
observational learning of 357 regression of learning 449 of retention 462
Spatial occlusion 322 as state of activation 128 standardized 25
Spatial orientation test 432 throwing proficiency as function of transfer 462
Specificity hypothesis 304-305. See of 129f validity 25
also practice specificity Stretch reflex 155 Test scores
Speech 169-170, 187,206,361 Stroke patients 319, 350 absolute retention 463
Speed-accuracy trade-off 35, 223- Stroop effect 103 difference score 463
224. See also equilibrium-point Structural interference 99 percentage scores 464, 466-467
theories; Fitts' law; impulse- Subjective estimations 422-424 predictor 314
variability theory; timing Subtractive method 58, 59 savings score 464, 467
correction models of 257-260 Synchronization tasks 237-239 Therbligs 37, 38
distance programming 242 Time-and-motion studies 6, 38
feedback hypothesis 233 T Time on target (TOT) 32
linear 231-233 Tactile information Time-to-contact (T,) 140, 148, 153
location programming 242 for balance 151 Timing. See also temporal
logarithmic 224, 227 feedback 168-169 coordination
movement-time goal hypothesis receptors 156-157 clock and motor variance 240-241
233-234 Task analysis 311-312 discrete anticipation 234-235
Spinal cord 157, 187-188 Task difficulty 447 discrete movement 235-236
Spinal preparations 182 Task orientation 14 errors 234
Spinal tuning 172 Tau 141, 148,153 individual differences 305
Stabilometer 32 Tau coupling 141 learning through modeling 357
Standardized tests 25 Tau gap 152 negative transfer of 485-486
Startle reflex 191-192 Taxonomies 310 practice variability and 370
Stimulus factor-analytic 310-311 reaching and grasping 269-271
clarity 61 task analysis 311-312 repetitive movement 237-239
complexity and RT 111 Temporal anticipation 82-83 types of 241
effect on reaction time 61 Temporal coordination 282 velocity effect 236-237
intensity 61 anti-phase 282, 292 Wing-Kristofferson model 239-240
modality 61 attention and 286-287 TMS. See transcranial magnetic
Stimulus detection 60-61 between-person task 288f stimulation (TMS)
Stimulus identification 60 bimanual timing patterns 282-283 Tonic neck reflex 169, 185
pattern recognition 61, 63 complex timing patterns 290-292 Top-down processing 99
580 Subject Index

Total variability (E) Unconscious behavior 98 locomotion and 152-153

versus absolute error 30 Units of action 265-266 object identification versus
computation 29-30 identifying 266 interaction 137
interpretation 30 methods in identification 266-268 for perception 137
Trace-decay theory 472 U.S. Air Force Human Resources role in gait transitions 281
Tracking tasks Research Center 9 sensory neuropathy and 181f
compensatory 22 U.S. Army Air Force Psycho-Motor training myth 484
examples of 22-23 Testing Program 9 ventral system 137
measures of accuracy 31-32 Utricle 153 in whole-body actions 151-153
pursuit 22 Visual agnosia 138
Training. See learning; practice v Visual attention 123, 488
Transcranial magnetic stimulation Validity 25-26 Visual feedback. See also aiming
(TMS) 48, 98 Variable error (VE) tasks
Transfer application 28-29 iconic memory and 471-472
defined 465 computation 28 intercepting moving targets 146-
drill practice 481 interpretation 28 147
experiments 465 Variable practice. See practice moving to stationary targets 141,
intertask 372, 481-483 variability 143
measures of 466-467 Velocity 41 practice specificity and 388-389
negative 465, 483-486 Velocity effect 237 on time to contact 140-141
positive 465 Verbal-cognitive abilities 22, 431, time to process 141, 143-148
proactive 465-466 469-470 uncertainty 144-146
retention and 462 Verbal learning 14, 421, 478, 479, Visual illusions 138, l39f
retroactive 465-466 483 Visual proprioception 140, 142
similarity and 483 Verbal-motor stage 431 Visual-search tasks 106
simulation and 486-488 Vestibular system 153 Visual streams 137-140
of timing 485-486 Vestibulo-occular reflex 265
variability-of-practice effects 367- Video feedback 401-402 w
369 Video games 488 Warm-up decrement 456, 477
Transfer-appropriate processing Videotapes 43 as forgetting 478
389-390 Virtual reality 43, 488 as loss of set 478-480
Transfer surface 483 Vision. See also visual feedback Weber's law 236
Transfer test 462 for action 137 Whole-body actions 151-153. See
Triggered reactions 167-170, 204- anticipatory actions and 148-153 also locomotion
205. See also reflexes balance and 151-152 Wineglass effect 168
Tuning 75 detection of distance 142f Wing-Kristofferson model 239-241
Two-hand coordination task 32 dissociations 137-138 Working memory 92
dorsal system 137
u y
gait transitions and 281
Uncertainty 67 interceptive control 148-151 Yerkes-Dodson law 128
 UT IH.E AUTHORS

R ichard A. Schmidt, PhD, is profes-

sor emeritus in the department of
psychology at UCLA. He currently runs
his own consulting firm, Human Perfor-
mance Research, working in the area of
human factors and human performance.
Known as one of the leaders in research
on motor behavior, Dr. Schmidt has more
than 35 years' experience in this area and
has published widely.
The originator of schema theory, Dr.
Schmidt founded the Journal of Motor
Behavior in 1969 and was editor for 11 years.
He authored the first edition of Motor Con-
trol and Learning in 1982, followed up with
a second edition of the popular text in 1988, The authors and their wives: Dick Schmidt, Laurie Wishart,
and collaborated with Tim Lee for the third Gwen Gordon, and Tim Lee.
edition in 1999 and fourth edition in 2005.
Dr. Schmidt received an honorary doctorate from Catholic University of Leuven,
Belgium, in recognition of his work. Schmidt is a member of the North American Society
for the Psychology of Sport and Physical Activity (of which he was president in 1982),
the Human Factors and Ergonomics Society, and the Psychonomic Society. Dr. Schmidt
has received the C.H. McCloy Research Lectureship from the American Alliance for
Health, Physical Education, Recreation and Dance.
His leisure-time activities include sailboat racing, amateur Porsche racing, and skiing.

T imothy D. Lee, PhD, is a professor in the department of kinesiology at McMaster

University in Hamilton, Ontario, Canada. He has published extensively in motor
behavior and psychology journals since 1979. More recently, he has contributed as an
editor to Journal of Motor Behavior and Research Quarterly for Exercise and Sport and as an
editorial board member for Psychological Review. Since 1984 his research has been sup-
ported by grants from the Natural Sciences and Engineering Research Council of Canada.
Dr. Lee is a member and past president of the Canadian Society for Psychomotor
Learning and Sport Psychology (SCAPPS) and a member of the North American Society
for the Psychology of Sport and Physical Activity (NASPSPA), the Psychonomic Society,
and the Human Factors and Ergonomics Society. In 1980 Dr. Lee received the inaugural
Young Scientist Award from SCAPPS; in 1991-92 he received a Senior Research Fellow-
ship by the Dienst Onderzoekscoordinatie, Catholic University in Leuven, Belgium; and
in 2005 he presented a prestigious Senior Scientist Lecture at NASPSPA.
In his leisure time, Dr. Lee enjoys playing hockey and golf. He has maintained a
lifelong fascination with blues music and would one day love to put years of motor
learning study into practice by learning to play blues guitar.
581
 You'll find
other outstanding
otor behavior resources at

nKinetics.com

-
ustralia .................................................... 08 8372 0999
nada .................................................. 1-800-465-7301
urope .......................................... +44 (O) 113 255 5665
ew Zealand .............................................0800 222 062

~CJ HUMAN KINETICS

~~ The Information Leader in Physical Activity & Health
lb\) P.O. Box 5076 •Champaign, IL 61825-5076 USA