Psychology’s Role in Mathematics and Science

Education
Nora S. Newcombe Temple University
Nalini Ambady Tufts University
Jacquelynne Eccles University of Michigan
Louis Gomez University of Pittsburgh
David Klahr Carnegie Mellon University
Marcia Linn University of California, Berkeley
Kevin Miller University of Michigan
Kelly Mix Michigan State University

Improving mathematics and science education in the trends) was crystallized in a National Academy of Sciences
United States has been a matter of national concern for report titled Rising Above the Gathering Storm: Energizing
over half a century. Psychology has a vital role to play in and Employing America for a Brighter Future (National
this enterprise. In this article, the authors review the kinds Academy of Sciences, National Academy of Engineering,
of contributions that psychology can make in four areas: and Institute of Medicine, 2005). In addition, there have
(a) early understanding of mathematics, (b) understanding been an impressive number of high-level reports on various
of science, (c) social and motivational aspects of involve- aspects of this problem (summarized in Table 1).
ment in mathematics and science, and (d) assessment of In this article, we show how advances in psycholog-
learning in mathematics and science. They also examine ical research in fields such as cognitive psychology, devel-
challenges to psychology’s playing a central and construc- opmental psychology, cognitive science, and the emerging
tive role and make recommendations for overcoming those learning sciences or “science of learning” (White, Frish-
challenges. koff, & Bullock, 2007) offer new opportunities to address
Keywords: psychology, mathematics, science, education, these continuing questions. We present some examples of
assessment, motivation findings from recent research in four areas: early mathe-
matics understanding; science understanding; social and

E ver since the Sputnik flight over half a century ago, motivational aspects of involvement in science, technol-
there has been continuing concern about improving ogy, engineering and mathematics (STEM); and assess-
math and science education in the United States. ment of STEM learning. Findings such as the ones we
Will America have enough skilled and creative scientists discuss give grounds for optimism about psychology’s
and engineers to keep it competitive in an increasingly potential role in addressing the “gathering storm.” How-
technology-based world? Can it educate a general popula- ever, there are also challenges to using these insights. In
tion that can comprehend and contribute to the crucial each section, we make recommendations regarding how to
social, political, and economic issues raised by those sci-
entific and technological advances? There has been some
Nora S. Newcombe, Department of Psychology, Temple University;
progress (see http://www.nas.edu/sputnik/bybee4.htm for Nalini Ambady, Department of Psychology, Tufts University; Jacque-
reflections on what was learned in the first 40 years after lynne Eccles and Kevin Miller, Department of Psychology, University of
Sputnik), but nevertheless advances have been slower than Michigan; Louis Gomez, School of Education, University of Pittsburgh;
many would like. International comparisons suggest that David Klahr, Department of Psychology, Carnegie Mellon University;
the United States still lags behind many other countries in Marcia Linn, Graduate School of Education, University of California,
Berkeley; and Kelly Mix, Department of Counseling, Educational Psy-
math and science, consistently scoring around the middle chology, and Special Education, Michigan State University.
of the pack. For example, the Trends in International Math- This article is based on the work of a joint American Psychological
ematics and Science Study (TIMMS) found little measur- Association (APA)/Society for Research in Child Development (SRCD)
able change in the performance of American fourth and Presidential Task Force on Mathematics and Science Education convened
by APA President Sharon Stephens Brehm and SRCD President Aletha
eighth graders between 1995 and 2003 (see http://nces.ed Huston during 2007 and chaired by Nora S. Newcombe.
.gov/timss/). Recent data from assessments conducted by We thank Rena Subotnik, Ashley Edmiston, and Maya Bassford of
the Organization for Economic Cooperation and Develop- APA’s Education Directorate for their support and advice and John Hagen
ment’s Program for International Student Assessment in- for attending the initial meeting.
Correspondence concerning this article should be addressed to Nora
dicate no change in this situation between 2003 and 2006 S. Newcombe, Department of Psychology, Temple University, 1701
(see http://www.pisa.oecd.org/dataoecd/16/28/39722597. North 13th Street, Room 318, Philadelphia, PA 19122-6085. E-mail:
pdf). National concern about these trends (or the lack of newcombe@temple.edu

538 September 2009 ● American Psychologist

© 2009 American Psychological Association 0003-066X/09/$12.00
Vol. 64, No. 6, 538 –550 DOI: 10.1037/a0014813
maximize psychology’s contribution to educational large and diverse populations (e.g., low socioeconomic
change. We call for collaborative research that brings to- status (SES) as well as middle class) (National Mathemat-
gether investigators with expertise in psychology, STEM ics Advisory Panel, 2008).
education, STEM disciplines, and related fields to work A second example comes from research on fraction
actively and cooperatively to improve STEM education. concepts. Children typically have great difficulty learning
We encourage research that fits in Pasteur’s Quadrant fractions in school—starting in the early elementary grades
(Stokes, 1997), that is, basic research that is also use- and persisting into high school (Behr, Lesh, Post, & Silver,
inspired. This kind of research has the potential both to 1983; National Mathematics Advisory Panel, 2008). How-
enhance our understanding of basic psychological pro- ever, there is reason to think that this difficulty does not
cesses and to substantially improve STEM education, but stem from failure to comprehend part–whole relations be-
the connections between psychological data and educa- cause children seem to understand these relations in infor-
tional practice have often been difficult to forge. This is mal tasks (e.g., Frydman & Bryant, 1988; Goswami, 1989;
unfortunate, not only because psychological research has Paik & Mix, 2008). To illustrate, first- and second grade
the potential to enrich and ground educational practice but children readily matched fractional amounts of food (e.g.,
also because educational practice has the potential to enrich 1/4 of a watermelon ⫽ 1/4 of a banana), even though they
and ground psychological research. failed similar verbal tasks (Paik & Mix, 2008). Mack
(1993, 2000, 2001) described several case studies in which
Early Understanding of Mathematics she posed fraction problems to elementary school children.
When children arrive at kindergarten, they bring along Though the children initially struggled with these prob-
a range of concepts based on informal experiences. These lems, Mack showed that when they were reminded of
experiences accrue throughout childhood and likely pro- related informal experiences, such as dividing up a pizza,
vide the grounding for the acquisition of increasingly ab- they easily determined the solutions. In fact, very few such
stract symbols (Lakoff & Nunez, 2000). Psychological reminders were needed to achieve dramatic results.
research has amassed much information about this informal This finding points the way toward a powerful teach-
knowledge as well as about the mechanisms that likely ing tool for fractions, but it does not go far enough. For
move children from one level of understanding to another teachers to implement this idea on a larger scale, they
(National Mathematics Advisory Panel, 2008). In this sec- would need to know whether remindings work in larger
tion, we look at three examples of the fruits of psycholog- groups of children and with children from a range of
ical research and consider their relation to educational socioeconomic backgrounds (National Mathematics Advi-
practice. sory Panel, 2008). Perhaps individual children need to be
First, consider what we know about preschoolers’ reminded of the specific experiences that matter to them.
understanding of quantitative transformations (e.g., addi- Alternatively, there may be a way to sequence fraction
tion and subtraction). This knowledge has been tested in instruction for large groups that incorporates informal ex-
very young children using ingenious, nonverbal tasks. In periences more generally yet still engages most students.
one study, Huttenlocher, Jordan, and Levine (1994) had The answers to such questions are likely rooted in basic
children first watch as objects were successively hidden or research on analogical transfer and remindings (e.g., Gent-
removed from a hidden set and then indicate the resulting ner & Toupin, 1986; Gick & Holyoak, 1983), but this may
number by constructing an equivalent set. Other research- not be obvious to educators and to those who investigate
ers have presented similar problems and then measured educational practice. Psychologists have to play a role in
whether children reach into the hiding place to retrieve all bringing this information to the educational sphere and
of the items (Feigenson & Carey, 2003; Van de Walle, confirming that what they have observed in small-scale
Carey, & Prevor, 2000). Children can complete such tasks laboratory research translates into the classroom.
early in development—performing above chance for very A third example of research in early mathematical
small set sizes in late toddlerhood. development pertains to learning to coordinate various
One general implication of these findings is that teach- dimensions of quantity and understand their interactions.
ers should expect children to have this informal knowledge This question was a central aspect of Piaget’s research on
when they enter school. However, further research is conservation. For example, he found that when young
needed to make more specific connections to educational children were asked to judge the equivalence of two con-
practice. For example, does nonverbal calculation ability tainers of liquid, they tended to overemphasize the height
predict later mathematical achievement? Perhaps children of the liquid and ignore its width. He argued that this
who enter school without this ability are slower to learn mistake reflected an inability to maintain and coordinate
symbolic calculation procedures. If so, are there instruc- attention toward two dimensions simultaneously—a logical
tional interventions preschool teachers could use to close capacity deemed necessary for higher level thought. In-
the achievement gap early on? Without targeted research, it deed, the inability to coordinate multiple dimensions of
is not clear what these interventions might be. Furthermore, quantity may underlie a number of stumbling blocks in
in the cases where promising interventions for early child- later mathematical reasoning. For instance, children as old
hood mathematics have emerged from small-scale psycho- as 7 years have trouble understanding that when the same
logical experiments, it is unclear whether they merit na- amount of food is divided among more recipients, the sizes
tional implementation because they have not been tested in of the portions decrease (Correa, Nunes, & Bryant, 1998;

September 2009 ● American Psychologist 539

Table 1
Reports Addressing the Need for Improved Math and Science Education in the United States
Report Description

Overarching reports:
How People Learn: Brain, Mind, Experience Focuses on current research on learning necessary for deep understanding,
and School (National Research Council, effective teaching, and supportive environments. Six key topics regarding
2000) understanding and five regarding teaching and supportive environments
are discussed.
Rising Above the Gathering Storm: Makes recommendations on how America can maintain its position in the
Energizing and Employing America for a fields of science and technology. It states that revitalizing mathematics
Brighter Future (National Academy of and science education from kindergarten through 12th grade is essential,
Sciences, National Academy of along with finding and keeping scientists and engineers from both the
Engineering, & Institute of Medicine, United States and abroad.
2005)
From Neurons to Neighborhoods: The Closely examines variables that affect very young children during
Science of Early Childhood Development development.
(National Research Council & Institute of
Medicine, 2000)
Eager To Learn: Educating Our Preschoolers Illustrates the way in which young children are presently being educated,
(National Research Council, 2001b) using examples from the field of cognitive science as a framework. The
book offers conclusions and recommendations for early childhood
education.
Engaging Schools: Fostering High School Reviews current research on what shapes adolescents’ school engagement
Students’ Motivation to Learn (National and motivation to learn, including new findings on students’ sense of
Research Council & Institute of Medicine, belonging, and looks at ways these can be used to reform urban high
2004) schools. This book looks at various approaches to reform through
different methods of instruction and assessment, adjustments in school
size, vocational teaching, and other key areas.
Mathematics education:
Adding It Up: Helping Children Learn Describes in detail variables associated with learning mathematics.
Mathematics (National Research Council,
2001a)
Review and Appraisal of the Federal “This report presents the results of an analysis of the federal government’s
Investment in STEM Education Research investment in learning and education research within the domains of
(National Science and Technology science, technology, engineering and mathematics (STEM)” (p. 2). The
Council, 2006) National Science and Technology Council’s Education and Workforce
Development Subcommittee created a STEM task force whose goals were
to review the current federal investment in research in learning and
education for Grades K–20, as well as to provide recommendations for
strengthening the federal education research portfolio to improve “STEM
learning and educational practices in the long run” (p. 2).
Foundations for Success: The Final Report of The National Mathematics Advisory Panel was created by President Bush to
the National Mathematics Advisory Panel advise the President and the Secretary of Education on the best use of
(National Mathematics Advisory Panel, scientifically based research on the teaching and learning of
2008) mathematics, with a focus on algebra. Five task groups and three
subcommittees were created; their evidence guidelines and initial
findings are outlined in this report.
Science education:
Learning to Think Spatially: GIS as a Support Spatial thinking is a cognitive skill that can be used in everyday life, the
System in the K–12 Curriculum (National workplace, and science to structure problems, find answers, and express
Research Council, 2006b) solutions using the properties of space. It can be learned and taught
formally to students with appropriately designed tools, technologies, and
curricula. This report explains the nature and functions of spatial thinking
and shows how spatial thinking can be supported across the K–12
curriculum through the development of appropriate support systems.

540 September 2009 ● American Psychologist

Table 1 (continued)
Report Description

Taking Science to School: Learning and What is science for a child? How do children learn about science and how
Teaching Science in Grades K–8 (National to do science? Drawing on a vast array of work from neuroscience to
Research Council, 2007) classroom observation, Taking Science to School provides a
comprehensive picture of what is known about teaching and learning
science from kindergarten through eighth grade. By looking at a broad
range of questions, this book provides a basic foundation for guiding
science teaching and supporting students in their learning.
Ready, Set, Science! Putting Research to This volume is designed as a practitioner-oriented accompaniment to
Work in K–8 Science Classrooms Taking Science to School: Learning and Teaching Science in Grades K–
(Michaels, Shouse, & Schweingruber, 8. It summarizes the findings in Taking Science to School and then goes
2007) on to present detailed case studies of real classroom experiences that
illustrate the complexities that science teachers grapple with every day:
selecting and designing rigorous and engaging instructional tasks,
managing classrooms, orchestrating productive discussions with culturally
and linguistically diverse groups of students, and helping students make
their thinking visible using a variety of representational tools. The aim is
to make the implications of research clear, accessible, and stimulating for
a broad range of science educators.
America’s Lab Report: Investigations in High Laboratory experiences as a part of most U.S. high school science curricula
School Science (National Research have been taken for granted for decades, but they have rarely been
Council, 2006a) carefully examined. What do they contribute to science learning? What
can they contribute to science learning? What is the current status of labs
in our nation’s high schools as a context for learning science?
Assessment:
Knowing What Students Know: The Science At a time when traditional testing is subject to increasing criticism, research
and Design of Educational Assessment suggests that new, exciting approaches to assessment may be on the
(National Research Council, 2001c) horizon. Advances in the sciences of how people learn and how to
measure such learning offer the hope of developing new kinds of
assessments—assessments that help students succeed in school by making
as clear as possible the nature of their accomplishments and the progress
of their learning. Knowing What Students Know essentially explains how
expanding knowledge in the scientific fields of human learning and
educational measurement can form the foundations of an improved
approach to assessment. These advances suggest ways that the targets of
assessment—what students know and how well they know it—as well as
the methods used to make inferences about student learning can be
made more valid and instructionally useful. Principles for designing and
using these new kinds of assessments are presented, and examples are
used to illustrate the principles. Implications for policy, practice, and
research are also explored.
High Stakes: Testing for Tracking, Promotion, Everyone is in favor of “high education standards” and “fair testing” of
and Graduation (National Research student achievement, but there is little agreement as to what these terms
Council, 1999) actually mean. High Stakes looks at how testing affects critical decisions
for American students. As more and more tests are introduced into the
country’s schools, it becomes increasingly important to know how those
tests are used—and misused—in assessing children’s performance and
achievements.

Miller, 1984; Sophian, Garyantes, & Chang, 1997). This Errors and misconceptions persist well into adolescence
inverse relationship underlies a whole class of mathematics (Behr et al., 1983).
concepts that are based on part–whole relations, including Psychological research on number and amount rela-
fractions, division, and proportions. Each of these concepts tions has the potential to explain why this is so. However,
is notoriously difficult for children to learn in school. this connection may not be immediately apparent because

September 2009 ● American Psychologist 541

the tasks used in these experiments are about sharing, scale problems and need large-scale data. Educators would
rather than the kinds of part–whole problems typically need to know, for example, what the best “down and dirty”
posed in school. Research that could directly link the two way is to measure math readiness in massive groups of
topics would help bridge this gap. For example, it would be preschoolers. If I test 1,000 preschoolers, how many are
desirable to know whether training on number and amount likely to lack the math readiness skills identified in psy-
relations in sharing situations would lead to improvement chological research? What happens to these children in the
on fraction comparison tasks, like those used in schools. long term— do these deficits predict achievement in later
Each of our three examples offers both promise and grades? Psychologists could assist in this effort by either
frustration. There are data with clear implications for math- replicating their findings in larger populations or partnering
ematics education, and yet further research is needed to with educational researchers who routinely do so. In a few
answer questions vital to bridging the gap between theo- areas, we are beginning to see this idea come to fruition. For
retical understanding of basic cognitive processes and what example, the Cognitive Tutors and other technology-linked
classroom practices are best. What leads to these gaps and instructional systems used by researchers at the Pittsburgh
how can we bridge them? One obstacle is that although, as Science of Learning Center have now logged over 30,000
we noted earlier, any attempt to improve math and science hours of student instruction time and over 6 million student–
education lies squarely in Pasteur’s Quadrant (Stokes, tutor interactions, producing data available for extremely thor-
1997), psychologists and educators tend to come at this ough analysis of fine-grained learning paths.
“engineering design” question from different perspectives The encouraging fact is that when researchers have
and tend to parse the learning landscape in different ways. managed to bridge the gap between psychology and math-
Some psychologists tend to focus on general processes and ematics education, their work has yielded some striking
conceptual primitives related to learning and motivation insights. For example, by looking at the contexts in which
that can be incorporated into a number of different struc- children acquire mathematics, recent work has uncovered
tures. For example, they may study how children learn the origins of individual differences in early mathematical
division, but the choice to study this problem is somewhat competence. Klibanoff, Levine, Huttenlocher, Vasilyeva,
arbitrary: Division problems may be more a vehicle for and Hedges (2006) looked at the kinds of mathematical
studying part–whole reasoning than an end in themselves. input children were exposed to in different preschools,
In contrast, many educators tend to parse the landscape in finding huge differences by SES in the nature and amount
terms of school subjects and instructional topics. In the of mathematical talk that preschool teachers produced.
educational literature, research on science concepts might These differences in experience seem a plausible source for
not be linked to research on math concepts, whereas in the the large SES-related differences in mathematics achieve-
psychology literature, these topics are seen as sharing com- ment when children enter school, differences that are the
mon learning mechanisms, including analogical reasoning, best predictor of later school mathematics achievement
the influence of language on cognition, symbol grounding, (Duncan et al., 2007). Yet these SES-related differences are
and so forth. Moreover, with respect to broad approaches to far from immutable. Ramani and Siegler (2008) found that
different types of instructional procedures, these different per- substantial differences between children from low- and
spectives give rise to long-standing and highly contentious middle-income families on number line estimation tasks
debates, such as the recent exchanges about the theoretical and were eliminated after children were engaged for only one
empirical basis of constructivist approaches to science and hour in playing simple board games that required counting.
math instruction (Hmelo-Silver, Duncan, & Chinn, 2007; This study demonstrates how research bridging education
Kirschner, Sweller, & Clark, 2006; D. Kuhn, 2007; Schmidt, and psychology not only reveals new insights about basic
Loyens, van Gog, & Paas, 2007; Sweller, Kirschner, & Clark, cognition but can also point the way toward an evidence-
2007). In addition, psychology may become preoccupied with based intervention.
questions that seem not to have direct links to education at all In summary, psychologists have a great deal to con-
(e.g., the innateness of concepts). tribute to the improvement of mathematics learning. Al-
Psychologists can take several steps to bridge this kind though this may require them to adapt some of their com-
of gap. For example, we could write and publish review mon research practices, doing so will pay off in terms of
papers that are targeted at the education audience, in which both practical utility and a deeper scientific understanding
we spell out how seemingly disparate educational problems of the nature of mathematical development.
are united at a process level. Another step is for psychol-
ogists to familiarize themselves with the educational liter- Understanding Science
ature on problems that relate to the processes they study. A
third step is to take their own research programs closer to Science education faces a somewhat different set of chal-
addressing problems in school learning by demonstrating lenges than mathematics education does. For one thing,
that specific process deficits can account for failure to learn there is much less agreement about what constitutes “core
certain material or that addressing these processing deficits knowledge” in science than in mathematics, and there is,
will lead to improved school learning in a variety of topics. accordingly, less consensus about optimal sequencing of
In addition, we note that psychologists tend to prefer different scientific domains. For another, science educators
small-scale studies that isolate and closely examine various aim to convey not only the content of science but also the
cognitive processes, whereas educators deal with large- processes whereby scientific knowledge is acquired, re-

542 September 2009 ● American Psychologist

fined, revised, extended, and disseminated, including Anderson, 1999), and understanding the logic of experi-
modes of argumentation and the social and professional mental design (Chen & Klahr, 1999; Tschirgi, 1980). These
context of the scientific enterprise. Although these issues heuristics and skills are considered important targets for
are clearly important for those choosing to become math- research and for education because they are assumed to
ematicians, in science the issues are important even for reflect domain generality and transferability (D. Kuhn et
those who seek only to understand the field and to make al., 1995; Ruffman, Perner, Olson, & Doherty, 1993). Psy-
informed decisions about scientific issues. For example, in chological research has also identified important parallels
the normal mathematics curriculum, one is quite unlikely to between historical and philosophical aspects of science
hear questions of the type “Where did the method of long (T. S. Kuhn, 1962) and the process of cognitive develop-
division come from?” or “How sure are we of the method of ment (Carey, 1985; Koslowski, 1996). It has identified the
integration by parts in calculus?” whereas such questions ways in which domain-specific knowledge evolves: some-
about the source and certainty of many scientific hypotheses times via the gradual elaboration of existing theories through
are considered to be part of a real appreciation of science. the accretion of new facts and knowledge and at other times
The recent National Academy of Sciences volume by the replacement of one theoretical framework by another.
Taking Science to School (National Research Council, Research in this tradition (Carey, 1988, 1991; Chi, 1992; Linn
2007) identifies several aspects of science that are essential & Hsi, 2000) places less emphasis on the mastery of domain-
to convey to students (see also Lehrer & Schauble, 2006, general logic, heuristics, or strategies and more on processes
for an extensive treatment of the different forms of scien- of conceptual or theory change.
tific knowledge that children must acquire). Educators want Last, science involves culturally established norms of
students to gain three things: knowledge about the natural argumentation, disagreement, presentation standards, and
world, understanding of the process by which that knowl- shared goals of truth seeking. These may be quite distinct
edge is generated, and an appreciation of the social and from everyday nonscientific discourse. The participatory
participatory nature of science. nature of science learning highlights that science is a cul-
Beginning with knowledge about the natural world, ture made up several subcultures that differ in large and
we note that there is a vast, almost overwhelming, amount small ways (Latour, 1999; Longino, 2002). Psychological
of such knowledge, ranging from the periodic table, to plate research based on this perspective emphasizes the fact that
tectonics, to Newton’s laws, and so on. Indeed, some of the individual scientists or groups of scientists are part of a
most difficult decisions in the creation of any science wider social environment, inside and outside science, with
curriculum concern what to include, what to leave out, and which they are in constant communication and that has
what to teach first. Cogent arguments can be made for just strongly shaped their knowledge, skills, resources, motives,
about any sequence of domain coverage, and the inevitable and attitudes.
result of the deliberations of those setting state science One of the challenges of science instruction is that
standards is an accumulation of topics based as much on rather than entering the science class as “empty vessels”
advocacy as on sound psychological and pedagogical prin- into which knowledge can be “poured,” novice science
ciples (Gross et al., 2005). The end result, as many have learners bring to the science classroom an accumulated
lamented, is a science curriculum that is a mile wide and a configuration of preconceptions about the natural world.
mile deep (Li & Klahr, 2006). One possible contribution Some of these are partially correct and easy to modify
that psychology can make to this difficult problem is to (Vosniadou & Brewer, 1994), whereas others are funda-
characterize the different domains in terms of their cogni- mentally at variance with the concepts to be acquired and
tive and motivational demands and to use our knowledge of require radical reconceptualization (Chi, 1992, 2005). In
developmental psychology to guide curriculum selection the former case, instruction based on analogical processes
and sequencing. may be very effective (Clement, 1993), but in the latter
In addition to what has been discovered, students need case, much research remains to be done to determine how
to know about how those discoveries were made. Psycho- best to overcome such misconceptions.
logical research has tended to focus on basic, domain- Another factor that makes it hard to teach science is
general cognitive processes including formal logic, heuris- that the language of science is confusingly similar to ev-
tics, and problem-solving strategies that span a wide range eryday language but may mean something quite distinct.
of scientific methods. The pioneering work in this area The literature suggests that adolescent science learners
includes that of Inhelder and Piaget (1958) on formal import everyday understanding of words into the science
operations, studies by Bruner, Goodnow, and Austin classroom in a way that can lead them down the garden
(1956) on concept development, and investigations by Wa- path or confuse them in subtle ways that teachers fail to
son (1960, 1968) on hypothesis-testing strategies. More recognize. Psychologists who study words and language
recent studies have focused on problem-solving strategies might be able to shed light on the conditions that cause
for coordinating theory and evidence (Klahr, 2000; D. everyday prior knowledge to be a source of confusion. In
Kuhn, Garcia-Mila, Zohar, & Andersen, 1995; D. Kuhn & addition to these conceptual and linguistic challenges, sci-
Park, 2005), mastering counterfactual reasoning (Leslie, ence instruction must deal with a puzzling paradox in the
1987), distinguishing patterns of evidence that do and do nature of children’s acquisition of causal reasoning pro-
not support a definitive conclusion (Amsel & Brock, 1996; cesses. On the one hand, there is evidence for surprisingly
Beck & Robinson, 2001; Fay & Klahr, 1996; Vellom & sophisticated causal reasoning in very young children

September 2009 ● American Psychologist 543

(Gopnik & Schulz, 2007), but on the other hand, there is science and math educators that are strongly supported by
evidence that even adults have great difficulty isolating our research.
causal factors in simple everyday contexts and that absent
explicit instruction they have a difficult time designing
Social and Motivational Issues
empirical investigations that can minimize causal ambigu- Over the last 30 years, many psychologists have studied the
ity. motivational influences on learning and engagement in
One fundamental goal of science education, as stated math and science. Much of this work began in the 1970s
in Taking Science to School (National Research Council, with a concern regarding the underparticipation of females
2007), is that children should be able to “know, use, and in STEM courses in high school and college and the un-
interpret scientific explanations of the natural world” (p. 2). derrepresentation of females in STEM careers. Although
In addition, science education must clarify the distinction there has been a major increase in the participation of
between what science is and is not. More specifically, it women in various STEM college majors and professions,
needs to convey what kinds of questions can be answered females and minorities are still underrepresented in some
by scientific inquiry and what kinds cannot. For example, disciplines including mathematics, physics, chemistry,
in the common “day at the pond” experience used in computer science, and engineering, especially at the higher
elementary science education, the following types of ques- levels such as full professorships (Hyde & Linn, 2006;
tions are grist for the scientist’s mill: “What makes the National Science Foundation, 2008). In addition, overall,
pond look green?” “Why is one kind of vegetation on this American students are underrepresented in STEM college
side, and another on the other side?” In contrast, it is majors compared with the proportions of students in those
important to delimit the scope of science from questions of majors in other countries (National Science Foundation,
meaning or social convention such as “Why are we here?” 2008). One way to understand this situation is in terms of
or “Why do we drive on the right?” social and motivational issues.
So, in summary, we know a lot about early scientific Eccles, Barber, and Josefowicz (1999) proposed a
thinking. What don’t we know? An important issue is that comprehensive theoretical model of the motivational and
our generalizations are based on specific topics, specific social influences on students’ engagement and performance
types and ages of students, and particular instructional in mathematics. Drawing on the classic expectancy value
strategies. However, to date, psychological research has theory of motivation, they linked achievement-related
explored only a small part of this overall space. We do not choices (such as the decision to take advanced math and
have an overarching and interconnected theory to address science courses in high school, or the decision to seek
all of these phenomena in a systematic fashion or one that training to enter a STEM field, or the willingness to engage
can suggest practical instructional strategies (pedagogy, fully in learning STEM material in primary and secondary
materials, texts, sequencing, etc.) for a broad range of school) to two sets of beliefs: the individual’s expectations
topics in the sciences (chemistry, physics, biology, etc.) for success (“Can I do the task?”) and the importance or
and the wide range of ages over which science is taught in value the individual attaches to the various options per-
school. ceived by the individual as available (“Do I want to do the
Our suggestions for future research in this area are task?”). These beliefs are related to cultural norms, expe-
consistent with the broad set of recommendations made in riences, aptitudes, and those personal beliefs and attitudes
the recently published Institute of Education Sciences prac- that are commonly assumed to be associated with achieve-
tice guide Organizing Instruction and Study to Improve ment-related activities (see Eccles, Wigfield, & Schiefele,
Student Learning (Pashler et al., 2007, p. 2): 1998), including causal attributions, beliefs regarding the
nature of intelligences, the input of parents, peers, and
1. Space learning over time. teachers, culturally based beliefs about both the nature of
2. Interleave worked example solutions with problem- achievement domains and the “appropriateness” of partic-
solving exercises. ipation, self-perceptions and self-concepts, perceptions of
3. Combine graphics with verbal descriptions. the task itself, and the processes and consequences associ-
4. Connect and integrate abstract and concrete repre- ated with identity formation.
sentations of concepts. Beginning with the question “Can I do the task?”
5. Use quizzing to promote learning. extensive work based on self-efficacy theory has docu-
6. Help students allocate study time efficiently. mented the critical importance of confidence in one’s abil-
7. Ask deep explanatory questions. ity to master the material being taught in STEM classes for
persistence and performance, particularly when one is
These recommendations are based on several decades faced with difficult material. Using both survey and inter-
of work in the learning sciences as applied to math and vention methods, many psychologists have demonstrated
science education. Perhaps the most notable feature of this the power of increased confidence in one’s math ability for
practice guide is that it indicates that the strength of the sustaining engagement and performance in STEM learning
evidence supporting most of the recommendations is “low” activities (Bandura, 1997; Pajares, 1996; Wigfield, Eccles,
to “moderate” in most cases, and “strong” in only a few Schiefele, Roeser, & Davis-Kean, 2006). Much of the
(Pashler et al., 2007, p. 2). Clearly, much work remains to recent work has focused on the psychological and social
be done before psychologists can make suggestions to forces underlying individual differences in the answer. One

544 September 2009 ● American Psychologist

major thrust in this area is grounded in Dweck’s work on cause teachers’ confidence in their students’ ability to mas-
theories of intelligence along with the recent work in ter the material is essential for supporting the students’
achievement goal theory. Dweck (1986, 2002) identified confidence in their own ability. Unfortunately, negative
two different theories of intelligence that are likely to affect racial, ethnic, gender, and social class stereotypes can lead
students’ continuing engagement in STEM through their teachers and school districts to communicate low expecta-
impact on individuals’ confidence in their ability to master tions for the academic achievements of some groups of
difficult STEM material. She proposed that entity theorists, students. Such processes are quite prevalent in the STEM
who believe that intelligence is fixed, are likely to lose fields.
confidence in their ability to master math and science when Full engagement in learning STEM also requires a
faced with difficult problems and preliminary failure expe- desire or willingness to do the task (Wigfield et al. 2006).
riences. In contrast, incremental theorists, who believe that Research has helped us understand the psychological and
intelligence is malleable, are more likely to attribute learn- social factors influencing the answer to the question “Do I
ing difficulties to lack of experience and to continue to have want to do this task?” as well as provided methods to
confidence in their ability to master the material despite increase the odds that the answer will be yes. Two of the
initial failures. most widely researched psychological influences relate to
Advocates of achievement goal theory link individu- intrinsic motivation and both personal and social identities.
als’ theories of intelligence to more general motivational On intrinsic motivation and interest, beginning with
orientations (e.g., Anderman, Maehr, & Midgley, 1999; the early work of Lepper and his colleagues and continuing
Midgley, 2002). These scholars argue that entity theorists with the work of Deci and Ryan and their colleagues,
tend to focus on performance goals (doing better than psychologists have shown the potential benefits of intrinsic
others and getting high grades), leading them to focus on motivation (i.e., doing something purely for the sake of
relative ability, to avoid challenging situations unless they personal enjoyment; Deci & Ryan, 2002) and the potential
are certain of success, and to perform less well when pitfalls of extrinsic motivation (i.e., doing something for an
challenged. In contrast, because incremental theorists focus external reward; Deci, Koestner, & Ryan, 1999; Lepper &
more on learning/mastery and improvement and believe Greene, 1978). Similarly, interest theorists have shown that
that failure is due to lack of knowledge rather than lack of students process interesting STEM materials much more
“ability,” they do well when challenged and persist despite deeply and fully than noninteresting STEM materials (Hidi,
initial difficulties. This line of reasoning has led to a large 1990). Intrinsic motivation and interest can be cultivated by
body of research studies focused on the differences in using different teaching strategies. Cordova and Lepper
mastery/learning goals versus ability/performance goals. (1996) suggested two ways in which this might occur:
By and large, this work has shown that approaching STEM through the use of contextualization and of personalization
learning situations with learning/mastery goals leads to and choice. Teachers often seek to present information in
improved performance and greater persistence. For exam- an abstract form, deliberately decontextualizing it from the
ple, Grant and Dweck (2003) followed college students in students’ everyday experience. This is thought to give the
an introductory chemistry course and found that those who student the ability to generalize the abstract knowledge in
had learning goals processed the course material more different domains. This manner of presentation, however,
deeply and were more likely to integrate the course mate- often has the unfortunate consequence of undermining chil-
rial across units. These students also earned higher grades dren’s interest in the subject matter because they do not see
than their peers who did not endorse learning goals, even any practical utility in it or how it could be applied to their
after Grant and Dweck controlled for SAT scores. Whether everyday curiosities and interests. By harnessing children’s
these effects are mediated by students’ confidence in their intrinsic curiosities about their world and applying abstract
ability to master the material needs to be assessed. concepts, such as math, to experiences they come across
Educational interventions designed to increase the every day (e.g., at home), teachers can likely increase
prevalence of learning/mastery goals do lead to increases in children’s intrinsic motivation (Bruner, 1966). Second, by
engagement with, and mastery of, STEM material. For using characters, themes, objects, and so forth that are
example, Farrell and Dweck (1985) found that eighth-grade already of high interest to children, educators can person-
children instructed to have learning goals in a week-long alize difficult and abstract concepts (e.g., teaching fractions
unit in their science class attained higher scores on a by dividing portions of a pizza), which makes them more
transfer-of-skills test, worked harder on the test, and were accessible and ultimately more memorable (Hidi, 1990;
more likely to try to apply what they learned to solve novel Linn, Bell, & Davis, 2004; Linn & Hsi, 2000).
problems compared with children instructed to have per- Other groups of developmental and social psycholo-
formance goals. Similarly, interventions designed to gists have stressed the role of personal and social identities
change children’s theories from entity to incremental pro- for students’ involvement in STEM. Eccles and her col-
duced performance and motivational gains (Blackwell, leagues focused on what they call attainment value, which
Trzesniewski, & Dweck, 2007). Finally, the National Re- they define in terms of the fit between activities such as
search Council and Institute of Medicine (2004) report on STEM courses and STEM professions and the individual’s
engaging schools stressed the importance of teachers and own needs, personal interests, and personal values. Longi-
school personnel having high expectations for all students’ tudinal studies have shown that gender differences in stu-
ability to master the material being taught precisely be- dents’ decisions to enroll in advanced mathematics are

September 2009 ● American Psychologist 545

mediated primarily by gender differences in the value that sociocultural stereotypes can have significant effects on
the students attached to mathematics. Furthermore, and academic performance. These effects are found in children
more important, young women think that advanced math as young as 5 to 7 years of age (Ambady, Bernieri, &
and physics are less important and enjoyable than the many Richeson, 2000).
other advanced high school courses they could be taking Fortunately, there are ways to protect against the
instead, particularly courses linked to the biological and negative effects of stereotypes. For example, Ambady,
social sciences. Not surprisingly, females are now as likely, Paik, Steele, Owen-Smith, and Mitchell (2004) have found
if not more likely, than their male counterparts to major in that stereotype threat can be attenuated through individua-
the biological and social sciences and to go into advanced tion. Women in a gender-primed condition who focused on
training and careers in the medical, biological, and social other aspects of their individual identity performed signif-
sciences. In their longitudinal follow-ups of adolescents as icantly better than women who were in a gender-primed but
they make the transition to adulthood, Eccles and her nonindividuated condition and as well as women in a
colleagues have found that females’ STEM career deci- non-gender-primed condition. Thus, individuation might
sions were closely linked to their desire to have a career be one potential mechanism that negatively stereotyped
that allowed them to directly help other people: Women groups can use to buffer stereotype threat. Since stereotype
with STEM interests and a desire to help other people threat may provide clues as to why fewer minorities and
avoided engineering in favor of the biological and medical women excel at STEM disciplines and careers, additional
sciences (Eccles et al., 1999). Interestingly, interventions research should focus on other fruitful means to assuage its
based on making physics more interesting to females by negative consequences.
using more human biological examples of physical princi- In summary, psychological evidence supports the con-
ples have been quite successful at increasing females’ clusion that motivational beliefs and the social processes
engagement in physics classes. linked to the development of these beliefs explain a sub-
Recently, social psychologists have also examined the stantial amount of individual and group differences in both
role of social identities and social stereotypes as important learning and engagement in STEM. Furthermore, we know
sociocultural factors influencing the underrepresentation of that student confidence and interest in STEM subject areas
women and minorities in STEM-related areas. For instance, decline on average across the K–12 school years. Finally,
Steele and Aronson’s (1995) work on “stereotype threat” we know that interventions based on the motivational prin-
has shown that when put in an academic performance ciples outlined in this section can be effective in both
situation that is believed to be diagnostic of their intellec- decreasing group differences in, and increasing average
tual ability, and when their race is made salient, African levels of, students’ engagement and performance in STEM
American students feel an extra burden to perform subject areas. However, many specifics remain to be
well—so as to not confirm, to themselves or others, the worked out, and it is telling that a recent Institute of
negative stereotype. This extra burden of anxiety interferes Education Sciences Practice Guide on encouraging female
with their performance and paradoxically impairs it. Much participation in the STEM disciplines could endorse many
of this work has focused specifically on mathematics. suggested practices only at low or moderate levels of
Gender stereotypes about math and science also run confidence in the evidence (Halpern et al., 2007). More
deep, and stereotype-threat effects have been found for collaborative research is needed between psychologists and
women in domains associated with negative gender stereo- educational researchers to design programs that can be
types, such as mathematics, when their gender identity is implemented in large-scale reform initiatives and to iden-
made salient. For example, in one study some women were tify which programs work best for which subpopulations.
led to believe that their performance on a particular math
task was gender relevant, and others were led to believe
Assessment
that the same math task was not gender relevant. Those We focus on assessment in the context of the No Child Left
participants who believed their performance was relevant Behind Act of 2001—a policy initiative that has mandated
to gender performed worse in comparison to men; and the annual assessment of mathematics and, more recently, in-
participants who believed their performance was not re- creased testing in science with the goal of improving learn-
lated to gender performed as well as men (Spencer, Steele, ing. Research in the learning sciences draws attention to the
& Quinn, 1999). potential consequences of this policy (National Research
It is interesting that positive stereotypes also affect Council, 1999). However, assessments by themselves do
performance in mathematics. Shih, Pittinsky, and Ambady not improve performance. To make increased assessment
(1999) subtly activated different aspects of the identity of meaningful and effective, the assessments should mandate
Asian American women and examined their mathematics the sorts of activities that characterize a science-literate
performance. In the domain of mathematics in this culture, individual. When they are effective, tests aligned with an
Asians are positively stereotyped, whereas women are neg- effective curriculum can increase the impact of the curric-
atively stereotyped. In accordance with prevailing socio- ulum. Ideally, tests themselves will serve as learning
cultural stereotypes, participants whose gender identity was events, contributing to the curriculum and at the same time
activated performed worse than control participants, but measuring student progress and informing teachers of in-
those whose Asian identity was activated performed sig- structional needs. In the worst scenario, tests have the
nificantly better. Thus social identities and their associated unintended consequence of motivating unproductive cur-

546 September 2009 ● American Psychologist

ricular changes such as increased test practice or elimina- decision makers need confidence that the assessments offer
tion of curricular activities that are not directly measured items that have characteristics aligned with our best under-
by the test. standing of student cognition and learning in mathematics
Analysis of state mathematics and science tests, for and science.
example, shows that they rarely measure important abilities Although the various stakeholders in the field of as-
such as using evidence to form arguments, interpreting sessment do not agree about the construct that underlies
contemporary dilemmas, or comprehending the nature of science and mathematics learning, investigators have in-
science. As a result, tests deter teachers from teaching the creased our understanding about how students learn and
skills that are valuable for science-literate individuals. how assessments tap the learning that has occurred (Na-
Some teachers infer that practice on test items would be the tional Research Council, 1999, 2000, 2006a). Measuring
best way to improve performance, and textbooks regularly literacy has been an important focus of research in science
include standardized items as part of class tests. When they and mathematics education. There is evidence that valid
are evaluated on standardized test performance, many math and reliable measures of the kinds of complex reasoning
and science teachers abandon inquiry goals and teaching that students need in order to operate in today’s world can
for understanding and substitute memorization and drill on be used in research and in large-scale assessments. Re-
multiple-choice questions requiring the recall of facts (Au, search on assessment offers promising ways to measure
2007). Although research shows that improving perfor- argumentation (Osborne, Erduran, & Simon, 2004), inte-
mance on multiple-choice, fact-oriented tests results from grated understanding (Clark & Linn, 2003; Linn & Hsi,
instruction that stresses coherent understanding (Cobb, 2000), coherent understanding (DeBoer, 2005), and under-
Wood, Yackel, & Perlwitz, 1992), schools and teachers are standing of the nature of scientific advance (diSessa, 2000).
often afraid to depart from practices that align with the We strongly encourage research programs to develop
perceived demands of the assessment. and validate these kinds of tests in current educational
To ensure that increased assessment impacts learning, settings. To illustrate, the Technology Enhanced Learning
we need to understand what should be measured and how and Science (TELS) center has conducted a large-scale
the results can lead to improved teaching and learning. As cohort comparison study to assess the impact of inquiry
articulated in the National Academy of Sciences publica- modules delivered using a technology-enhanced learning
tion titled Knowing What Students Know (National Re- environment and powerful visualizations of scientific phe-
search Council, 2001c), we need improved design, use, and nomena on student learning (Linn, Lee, Tinker, Husic, &
interpretation of assessments to ensure that they contribute Chiu, 2006). In this study, the TELS center participants
to learning. The report underscores the value of collabora- contrasted current high-stakes items typically administered
tive research, calling for “increased and sustained multi- in a multiple-choice format with explanation items from
disciplinary collaboration around theoretical and practical some high-stakes tests, as well as from research. The in-
matters of assessment” (p. 11). Knowing What Students vestigators found that the items that required students to
Know specifically articulates what the authors of the report develop an argument and explain their reasoning were far
call “the assessment triangle” (p. 296). The triangle links a more sensitive to inquiry instruction than were the items
model of student cognition and learning in the domain, that asked students to select an answer in a multiple-choice
beliefs about the kinds of observations that will provide format.
evidence of students’ competencies, and an interpretation For the explanation items, the TELS research group
process for making sense of the evidence. The authors of found that it is important not only to have items that require
the report call for using the assessment triangle to itera- reasoning but also to have scoring rubrics that capture the
tively refine these three elements of the assessment in order accomplishments. By scoring high-stakes test items using a
to align the goals of instruction with the outcome measures. rubric that rewarded making connections and building ar-
The report is replete with examples of inappropriate test guments, TELS increased the sensitivity of these items.
use resulting from limitations in implementation of the With the rescoring, these items were effective at distin-
assessment triangle. Lack of alignment can lead to serious guishing between inquiry instruction and typical instruc-
errors in both designing curriculum materials (when the tion. Aligning the goals of instruction with the assessment
goals of the curriculum materials do not reflect the goals of and the scoring rubric is crucial to gaining the benefit
the assessment) and in assessing students (when students promised by the No Child Left Behind Act of 2001.
do not have the opportunity to learn the material tested). Technology-enhanced instruction can provide a far
In addition, to implement effective solutions, decision more detailed set of information for teachers and students
makers need more informative models of student cognition than is possible with typical curriculum materials. Using
and learning, curriculum designers need to design materials materials developed and delivered with a learning environ-
based on research findings and to conduct research on how ment allows gathering of information about student perfor-
instruction contributes to long-term understanding, and mance in the moment. Computer-delivered instruction can
schools need to identify combinations of instruction and include embedded assessments that ask students to reflect
assessment that jointly lead to improved science under- on their understanding or to judge their own learning, and
standing. Ideally, individuals making decisions about in- the computer can construct quite sophisticated models not
struction will have evidence about how specific materials only of students’ subject matter knowledge but also of their
provide opportunities for students to learn. In addition, strategies for using, and sometimes “gaming,” such sys-

September 2009 ● American Psychologist 547

tems (Aleven, Stahl, Schworm, Fischer, & Wallace, 2003; the assessments used in mathematics and science
Koedinger & Aleven, 2007). They can also log the kinds of and provide more immediate feedback to teachers
investigations that learners conduct. This information can ● research to investigate the relationship between
guide learners, inform the design of instructional materials, knowledge gained in classrooms, in laboratory ex-
and help teachers use their time effectively. The informa- periences, and in field settings, in order to better
tion that these environments can gather is relatively easily understand the relationships between what has been
communicated to teachers. Teachers can use this informa- called formal and informal learning, as well as to
tion to modify their instruction and to improve the activities assess factors such as anxiety and stereotype threat.
that their students encounter.
In summary, assessment is currently at a rather primi- Conclusion
tive level. We have the opportunity, using technology-en- Psychology is a broad discipline that encompasses many
hanced materials and current research in cognition, to gain a areas of research important to successful education in
far more precise and detailed understanding of the trajectories mathematics and science. In addition, it is a key discipline
that students follow, the kinds of difficulties that they face, along with cognitive science, neuroscience, computer sci-
and the sorts of learning experiences that advance their un- ence, and other fields in the establishment of a new science
derstanding. Similarly, we have an opportunity to gather very of learning that has exciting potential to provide deep
powerful and useful information from a broad range of indi- insights into the nature of human learning and how best to
cators about teacher learning, teacher performance, and enhance it at all ages and in a variety of disciplines. In
teacher response to evidence of student progress. Further- order to make these goals realizable, however, psychology
more, teachers can use this broad array of information to will need to learn to do research in multidisciplinary con-
determine what works for the students they teach. texts, in specific public policy environments, and with an
We offer a number of recommendations. First, as each eye to useful application as well as pure knowledge. Such
of the issues discussed in this report suggests, interdisci- alliances are in fact emerging, and can be facilitated by
plinary research has the potential to improve STEM learn- professional organizations including the American Psycho-
ing. As a starting point, we encourage the convening of logical Association and the Society for Research in Child
summits of all stakeholders to discuss contemporary topics Development.
and come up with a roadmap for research efforts in this
field. Multidisciplinary collaborations need to involve psy- REFERENCES
chometricians, learning scientists, STEM educators, disci-
pline specialists, technologists, and classroom teachers. Aleven, V., Stahl, E., Schworm, S., Fischer, F., & Wallace, R. M. (2003).
Partners will need to jointly negotiate understanding of the Help seeking and help design in interactive learning environments.
Review of Educational Research, 73, 277–320.
cognition of student learning, the design of observations
Ambady, N., Bernieri, F. J., & Richeson, J. A. (2000). Toward a histology
that will provide evidence of student competencies, and the of social behavior: Judgmental accuracy from thin slices of the behav-
interpretation of the results for classroom use, student use, ioral stream. In M. P. Zanna (Ed.), Advances in experimental social
policymaking, and long-range planning. To build consen- psychology (Vol. 32, pp. 201–271). San Diego, CA: Academic Press.
sus and support an emergent research program, these sum- Ambady, N., Paik, S., Steele, J., Owen-Smith, A., & Mitchell, J. (2004).
Deflecting negative self-relevant stereotype activation: The effects of
mits should include the following possible foci: individuation. Journal of Experimental Social Psychology, 40, 401–
408.
● How can we enable teachers to take advantage of Amsel, E., & Brock, S. (1996). The development of evidence evaluation
the diagnostic information available in classrooms skills. Cognitive Development, 11, 523–550.
and to identify additional sources of information? Anderman, E. M., Maehr, M. L., & Midgley, C. (1999). Declining
● How can we create assessments that measure the motivation after the transition to middle school: Schools can make a
goals of science articulated in recent reports such as difference. Journal of Research and Development in Education, 32,
131–147.
Taking Science to School (National Research Coun- Au, W. (2007). High-stakes testing and curricular control: A qualitative
cil, 2007) and Knowing What Students Know (Na- metasynthesis. Educational Researcher, 36, 258 –267.
tional Research Council, 2001c)? Bandura, A. (1997). Self-efficacy: The exercise of control. New York:
● How can we align assessment with instruction, con- W. H. Freeman.
sistent with the recommendations of Knowing What Beck, S. R., & Robinson, E. J. (2001). Children’s ability to make tentative
interpretations of ambiguous messages. Journal of Experimental Child
Students Know (National Research Council, Psychology, 79, 95–114.
2001c)? Behr, M. J., Lesh, R., Post, T. R., & Silver, E. A. (1983). Rational number
concepts. In R. Lesh & M. Landau (Eds.), Acquisition of mathematics
Second, we recommend that funds be allocated to concepts and processes (pp. 91–126). New York: Academic Press.
support important and promising questions that clearly Blackwell, L., Trzesniewski, K., & Dweck, C. S. (2007). Implicit theories
need to be researched. These include the following: of intelligence predict achievement across an adolescent transition: A
longitudinal study and an intervention. Child Development, 78, 246 –
263.
● research to understand how the current high-stakes
Bruner, J. S. (1966). Toward a theory of instruction. Cambridge, MA:
tests are influencing motivation, as well as under- Harvard University Press.
standing Bruner, J. S., Goodnow, J. J., & Austin, G. A. (1956). A study of thinking.
● research to understand how technology can improve New York: Wiley.

548 September 2009 ● American Psychologist

Carey, S. (1985). Conceptual change in childhood. Cambridge, MA: MIT Gentner, D., & Toupin, C. (1986). Systematicity and surface similarity in
Press. the development of analogy. Cognitive Science: A Multidisciplinary
Carey, S. (1988). Conceptual differences between children and adults. Journal, 10, 277–300.
Mind and Language, 3, 167–181. Gick, M. L., & Holyoak, K. J. (1983). Schema induction and analogical
Carey, S. (1991). Knowledge acquisition: Enrichment or conceptual transfer. Cognitive Psychology, 15, 1–38.
change? In S. Carey & R. Gelman (Eds.), The epigenesis of mind: Gopnik, A., & Schulz, L. (Eds.). (2007). Causal learning: Psychology,
Essays on biology and cognition (pp. 257–291). Hillsdale, NJ: Erlbaum. philosophy, and computation. New York: Oxford University Press.
Chen, Z., & Klahr, D. (1999). All other things being equal: Acquisition Goswami, U. (1989). Relational complexity and the development of
and transfer of the control of variables strategy. Child Development, 70, analogical reasoning. Cognitive Development, 4, 251–268.
1098 –1120. Grant, H., & Dweck, C. S. (2003). Clarifying achievement goals and their
Chi, M. T. H. (1992). Conceptual change within and across ontological impact. Journal of Personality and Social Psychology, 85, 541–553.
categories: Examples from learning and discovery in science. In R. Gross, P., Goodenough, U., Lerner, L. S., Haack, S., Schwartz, M.,
Giere (Ed.), Cognitive models of science: Minnesota Studies in the Schwartz, R., & Finn, C. E., Jr. (2005). The state of state science
Philosophy of Science (pp. 129 –186). Minneapolis: University of Min- standards 2005. Retrieved March 16, 2009, from Thomas B. Fordham
nesota Press. Institute website: http://www.edexcellence.net/detail/news.cfm?news_id⫽
Chi, M. T. H. (2005). Commonsense conceptions of emergent processes: 352.
Why some misconceptions are robust. Journal of the Learning Sci- Halpern, D., Aronson, J., Reimer, N., Simpkins, S., Star, J., & Wentzel, K.
ences, 14(2), 161–199. (2007). Encouraging girls in math and science (NCER 2007–2003).
Clark, D., & Linn, M. C. (2003). Designing for knowledge integration: Washington, DC: National Center for Education Research, Institute of
The impact of instructional time. Journal of the Learning Sciences, 12, Education Sciences, U.S. Department of Education. Retrieved March 16,
451– 494. 2009, from http://ies.ed.gov/ncer/pubs/practiceguides/20072003.asp.
Clement, J. (1993). Using bridging analogies and anchoring intuitions to Hidi, S. (1990). Interest and its contribution as a mental resource for
deal with students’ preconceptions. Journal of Research in Science learning. Review of Educational Research, 60, 549 –571.
Teaching. 30, 1241–1257. Hmelo-Silver, C. E., Duncan, R. G., & Chinn, C. A. (2007). Scaffolding
Cobb, P., Wood, T., Yackel, E., & Perlwitz, M. (1992). A longitudinal, and achievement in problem-based and inquiry learning: A response to
follow-up assessment of a second-grade problem-centered mathematics Kirschner, Sweller, and Clark (2006). Educational Psychologist, 42,
project. Educational Studies in Mathematics, 23, 483–504. 99 –107.
Cordova, D. I., & Lepper, M. R. (1996). Intrinsic motivation and the Huttenlocher, J., Jordan, N. C., & Levine, S. C. (1994). A mental model
process of learning: Beneficial effects of contextualization, personal- for early arithmetic. Journal of Experimental Psychology: General,
ization, and choice. Journal of Educational Psychology, 88, 715–730. 123, 284 –296.
Correa, J., Nunes, T., & Bryant, P. (1998). Young children’s understand- Hyde, J. S., & Linn, M. C. (2006). Gender similarities in mathematics and
ing of division terms in a noncomputational task. Journal of Educa- science. Science, 27, 599 – 600.
tional Psychology, 90, 321–329. Inhelder, B., & Piaget, J. (1958). The growth of logical thinking from
DeBoer, G. E. (2005, January). Standardizing test items. Science Scope, childhood to adolescence. London: Routledge & Kegan Paul.
28(4), 10 –11. Kirschner, P. A., Sweller, J., & Clark, R. (2006). Why minimal guidance
Deci, E. L., Koestner, R., & Ryan, R. M. (1999). A meta-analytic review during instruction does not work: An analysis of the failure of con-
of experiments examining the effects of extrinsic rewards on intrinsic structivist, discovery, problem-based, experiential and inquiry-based
motivation. Psychological Bulletin, 125, 627– 668. teaching. Educational Psychologist, 41, 75– 86.
Deci, E. L., & Ryan, R. M. (2002). The paradox of achievement: The Klahr, D. (2000). Exploring science: The cognition and development of
harder you push, the worse it gets. In J. Aronson (Ed.), Improving discovery processes. Cambridge, MA: MIT Press.
academic achievement: Impact of psychological factors on education Klibanoff, R., Levine, S. C., Huttenlocher, J., Vasilyeva, M., & Hedges,
(pp. 61– 87). San Diego, CA: Academic Press. L. (2006). Preschool children’s mathematical knowledge: The effect of
diSessa, A. A. (2000). Changing minds: Computers, learning and literacy. teacher “math talk.” Developmental Psychology, 42, 59 – 69.
Cambridge, MA: MIT Press. Koedinger, K., & Aleven, V. (2007). Exploring the assistance dilemma in
Duncan, G. J., Claessens, A., Huston, A. C., Pagani, L. S., Engel, M., experiments with Cognitive Tutors. Educational Psychology Review,
Sexton, H., et al. (2007). School readiness and later achievement. 19, 239 –264.
Developmental Psychology, 43, 1428 –1446. Koslowski, B. (1996). Theory and evidence: The development of scientific
Dweck, C. S. (1986). Motivational processes affecting learning. American reasoning. Cambridge, MA: MIT Press.
Psychologist, 41, 1040 –1048. Kuhn, D. (2007). Is direct instruction the answer to the right question?
Dweck, C. S. (2002). The development of ability conceptions. In A. Educational Psychologist, 42, 109 –113.
Wigfield & J. S. Eccles (Eds.), Development of achievement motivation Kuhn, D., Garcia-Mila, M., Zohar, A., & Andersen, C. (1995). Strategies
(pp. 57– 88). San Diego, CA: Academic Press. of knowledge acquisition. Monographs of the Society for Research in
Eccles, J. S., Barber, B., & Josefowicz, D. (1999). Linking gender to Child Development, 60(4, Serial No. 245).
educational, occupational, and recreational choices: Applying the Kuhn, D., & Park, S. (2005). Epistemological understanding and intellec-
Eccles et al. model of achievement-related choices. In W. B. Swann, tual values. International Journal of Educational Research, 43, 111–
J. H. Langlois, & L. A. Gilbert (Eds.), Sexism and stereotypes in 124.
modern society: The gender science of Janet Spence (pp. 153–192). Kuhn, T. S. (1962). The structure of scientific revolutions. Chicago:
Washington, DC: American Psychological Association. University of Chicago Press.
Eccles, J. S., Wigfield, A., & Schiefele, U. (1998). Motivation to succeed. Lakoff, G., & Nunez, R. E. (2000). Where mathematics comes from: How
In W. Damon (Series Ed.) & N. Eisenberg (Vol. Ed.), Handbook of the embodied mind brings mathematics into being. New York: Basic
child psychology (5th ed., Vol. 3, pp. 1017–1095). New York: Wiley. Books.
Farrell, E., & Dweck, C. S. (1985). The role of motivational processes in Latour, B. (1999). Pandora’s hope: Essays on the reality of science
transfer of learning. Unpublished manuscript, Harvard University. studies. London: Cambridge University Press.
Fay, A., & Klahr, D. (1996). Knowing about guessing and guessing about Lehrer, R., & Schauble, L. (2006). Scientific thinking and science literacy.
knowing: Preschoolers’ understanding of indeterminacy. Child Devel- In W. Damon, R. Lerner, K. A. Renninger, & I. E. Sigel (Eds.),
opment, 67, 689 –716. Handbook of child psychology: Vol. 4. Child psychology in practice
Feigenson, L., & Carey, S. (2003). Tracking individuals via object-files: (6th ed., pp. 153–196). Hoboken, NJ: Wiley.
Evidence from infants’ manual search. Developmental Science, 6, 568 – Lepper, M. R., & Greene, D. (1978). The hidden costs of reward. Hills-
584. dale, NJ: Erlbaum.
Frydman, O., & Bryant, P. (1988). Sharing and the understanding of Leslie, A. M. (1987). Pretense and representation: The origins of “theory
number equivalence by young children. Cognitive Development, 3, of mind.” Psychological Review, 94, 412– 426.
323–339. Li, J., & Klahr, D. (2006). The psychology of scientific thinking: Impli-

September 2009 ● American Psychologist 549

 cations for science teaching and learning. In J. Rhoton & P. Shane National Research Council & Institute of Medicine. (2004). Engaging
(Eds.), Teaching science in the 21st century (National Science Teachers schools: Fostering high school students’ motivation to learn. Washing-
Association and National Science Education Leadership Association). ton, DC: National Academies Press.
Arlington, VA: NSTA Press. National Science and Technology Council. (2006, October 6). Review and
Linn, M. C., Bell, P., & Davis, E. A. (2004). Specific design principles: appraisal of the federal investment in STEM education research. Retrieved
Elaborating the scaffolded knowledge integration framework. In M. C. March 16, 2009, from http://www.ostp.gov/galleries/NSTC%20Reports/
Linn, E. A. Davis, & P. Bell (Eds.), Internet environments for science STEMEducationResearchOctober06.pdf
education (pp. 315–340). Mahwah, NJ: Erlbaum. National Science Foundation. (2008). Women, minorities, and persons of
Linn, M. C., & Hsi, S. (2000). Computers, teachers, and peers: Science disability in science and engineering (NSF 07–315). Arlington, VA:
learning partners. Mahwah, NJ: Erlbaum. Author. (Available at http://www.nsf.gov/statistics/wmpd/employ.htm)
Linn, M. C., Lee, H.-S., Tinker, R., Husic, F., & Chiu, J. L. (2006, August No Child Left Behind Act of 2001, Pub. L. No. 107–110, 115 Stat. 1425
25). Teaching and assessing knowledge integration in science. Science, (2002).
313, 1049 –1050. Osborne, J., Erduran, S., & Simon, S. (2004). Enhancing the quality of
Longino, H. (2002). The fate of knowledge. Princeton, NJ: Princeton argumentation in school science. Journal of Research in Science Teach-
University Press. ing, 41, 994 –1020.
Mack, N. K. (1993). Learning rational numbers with understanding: The Paik, J. H., & Mix, K. S. (2008). Do Korean fraction names promote part–whole
case of informal knowledge. In T. P. Carpenter, E. Fennema, & T. A. reasoning? Journal of Cognition and Development, 9, 145–170.
Romberg (Eds.), Rational numbers: An integration of research. Studies Pajares, F. (1996). Self-efficacy beliefs in academic settings. Review of
in mathematical thinking and learning (pp. 85–105). Hillsdale, NJ: Educational Research, 66, 543–578.
Erlbaum. Pashler, H., Bain, P., Bottge, B., Graesser, A. Koedinger, K., McDaniel,
Mack, N. K. (2000). Long-term effects of building on informal knowledge M., & Metcalfe, J. (2007). Organizing instruction and study to improve
in a complex content domain: The case of multiplication of fractions. student learning: A practice guide (NCER 2007–2004). Washington,
Journal of Mathematical Behavior, 19, 307–332. DC: National Center for Education Research, Institute of Education
Mack, N. K. (2001). Building on informal knowledge through instruction Sciences, U.S. Department of Education.
in a complex content domain: Partitioning, units, and understanding Ramani, G. B., & Siegler, R. S. (2008). Promoting broad and stable
multiplication of fractions. Journal for Research in Mathematics Edu- improvements in low-income children’s numerical knowledge through
cation, 32, 267–295. playing number board games. Child Development, 79, 375–394.
Michaels, S., Shouse, A. W., & Schweingruber, H. A. (2007). Ready, set, Ruffman, T., Perner, J., Olson, D. R., & Doherty, M. (1993). Reflecting on
science! Putting research to work in K-8 science classrooms. Wash- scientific thinking: Children’s understanding of the hypothesis– evi-
ington, DC: National Academies Press. dence relation. Child Development, 64, 1617–1636.
Midgley, C. (2002). Goals, goal structures, and adaptive learning. Mah- Schmidt, H. G., Loyens, S. M. M., van Gog, T., & Paas, F. (2007).
wah, NJ: Erlbaum. Problem-based learning is compatible with human cognitive architec-
Miller, K. F. (1984). Child as the measurer of all things: Measurement ture: Commentary on Kirschner, Sweller, and Clark (2006). Educa-
procedures and the development of quantitative concepts. In C. Sophian tional Psychologist, 42, 91–97.
(Ed.), Origins of cognitive skills: The Eighteenth Annual Carnegie Shih, M., Pittinsky, T. L., & Ambady, N. (1999). Stereotype susceptibil-
Symposium on Cognition (pp. 193–228). Hillsdale, NJ: Erlbaum. ity: Identity salience and shifts in quantitative performance. Psycholog-
National Academy of Sciences, National Academy of Engineering, & ical Science, 10, 81– 84.
Institute of Medicine. (2005). Rising above the gathering storm: Ener- Sophian, C., Garyantes, D., & Chang, C. (1997). When three is less than
gizing and employing America for a brighter future. Washington, DC: two: Early developments in children’s understanding of fractional
National Academies Press. quantities. Developmental Psychology, 33, 731–744.
National Mathematics Advisory Panel. (2008). Foundations for success: Spencer, S., Steele, C. M., & Quinn, D. (1999). Stereotype threat and
The final report of the National Mathematics Advisory Panel. Wash- women’s math performance. Journal of Experimental Social Psychol-
ington, DC: U.S. Department of Education. ogy, 35, 4 –28.
National Research Council. (1999). High stakes: Testing for tracking, Steele, C. M., & Aronson, J. (1995). Stereotype threat and the intellectual
promotion, and graduation (J. P. Heubert & R. M. Hauser, Eds.). test performance of African Americans. Journal of Personality and
Washington, DC: National Academy Press. Social Psychology, 69, 797– 811.
National Research Council. (2000). How people learn: Brain, mind, Stokes, D. E. (1997). Pasteur’s quadrant: Basic science and technological
experience, and school (J. D. Bransford, A. L. Brown, & R. R. Cocking, innovation. Washington, DC: Brookings Institution Press.
Eds.). Washington, DC: National Academies Press. Sweller, J., Kirschner, P. A., & Clark, R. (2007). Why minimally guided
National Research Council. (2001a). Adding it up: Helping children learn teaching techniques do not work: A reply to commentaries. Educational
mathematics (J. Kilpatrick, J. Swafford, & B. Findell, Eds.). Washing- Psychologist, 42, 115–121.
ton, DC: National Academy Press. Tschirgi, J. E. (1980). Sensible reasoning: A hypothesis about hypotheses.
National Research Council. (2001b). Eager to learn: Educating our Child Development, 51, 1–10.
preschoolers (B. T. Bowman, M. S. Donovan, & M. S. Burns, Eds.). Van de Walle, G. A., Carey, S., & Prevor, M. (2000). Bases for object
Washington, DC: National Academy Press. individuation in infancy: Evidence from manual search. Journal of
National Research Council. (2001c). Knowing what students know: The Cognition and Development, 1, 249 –280.
science and design of educational assessment (J. W. Pellegrino, N. Vellom, R. P., & Anderson, C. W. (1999). Reasoning about data in middle
Chudowsky, & R. Glaser, Eds.). Washington, DC: National Academy school science. Journal of Research in Science Teaching, 36, 179 –199.
Press. Vosniadou, S., & Brewer, W. F. (1994). Mental models of the day/night
National Research Council. (2006a). America’s lab report: Investigations cycle. Cognitive Science, 18, 123–183.
in high school science (S. R. Singer, M. I. Hilton, & H. A. Schwein- Wason, P. C. (1960). On the failure to eliminate hypotheses in a concep-
gruber, Eds.). Washington, DC: National Academies Press. tual task. Quarterly Journal of Experimental Psychology, 12, 129 –140.
National Research Council. (2006b). Learning to think spatially: GIS as a Wason, P. C. (1968). Reasoning about a rule. Quarterly Journal of
support system in the K-12 curriculum. Washington, DC: National Experimental Psychology, 20, 273–281.
Academies Press. White, G., Frishkoff, G., & Bullock, M. (2008). Bridging the gap between
National Research Council. (2007). Taking science to school: Learning and psychological science and educational policy and practice. In S. K.
teaching science in grades K-8 (R. A. Duschl, H. A. Schweingruber, & A. Thurman & C. A. Fiorello (Eds.), Applied cognitive research in K–3
W. Shouse, Eds.). Washington, DC: National Academies Press. classrooms (pp. 227–264). New York: Routledge.
National Research Council & Institute of Medicine. (2000). From neurons Wigfield, A., Eccles, J. S., Schiefele, U., Roeser, R. W., & Davis-Kean, P.
to neighborhoods: The science of early childhood development (J. P. (2006). Development of achievement motivation. In W. Damon (Series
Shonkoff & D. A. Phillips, Eds.). Washington, DC: National Academy Ed.) & N. Eisenberg (Vol. Ed.), Handbook of child psychology (6th ed.,
Press. Vol. 3, pp. 933–1002). New York: Wiley.

550 September 2009 ● American Psychologist