Educational Psychology Review, Vol. 10, No.

1, 1998

The Distribution of Distributed Cognition:

Multiple Interpretations and Uses
Joyce L. Moore1,2 and Thomas R. Rocklin1

We argue that greater precision is needed in the use of the term distributed
cognition if this perspective is to significantly impact educational research. We
describe a continuum of interpretations of distributed cognition ranging from
a conceptualization of cognition as an individual phenomenon that is
influenced by factors external to the individual, to a conceptualization of
cognition as a social phenomenon that cannot be reduced to individual
psychological constructs. We discuss the issues raised by the papers in the
current volume, and locate the perspectives taken in these papers along the
distributed cognition continuum. The relationship between distributed cognition
and situated cognition is then examined, as these terms are often used
interchangeably. Finally, we discuss key issues for further research in distributed
cognition.
KEY WORDS: distributed cognition; situated cognition; context; commentary; educational
psychology.

INTRODUCTION

Perspectives on learning termed distributed cognition have become in-

creasingly popular in educational research. The term distributed cognition,
however, is currently used with a multitude of interpretations. The rapid
and widespread adoption of this term is accompanied by the risk that the
term comes to be used so broadly and imprecisely that it will mean very
little at all. "Distributed cognition" may slide from being a useful term in
our professional lexicon to being used ever more loosely until its original

1Divisionof Psychological and Quantitative Foundations, University of Iowa, Iowa City.

2Correspondence should be directed to Joyce Moore, 361 Lindquist Center, College of Edu-
cation, The University of Iowa, Iowa City, Iowa 52242.

97
1040-726X/98/0300-0097$15.00/0 © 1998 Plenum Publishing Corporation
98 Moore and Rocklin

meaning is lost (Nash, 1993). This is not a unique phenomenon in the edu-
cational literature. Terms such as "schema," "mental model," "situated cog-
nition" (Lave, 1991), and "structure" (Phillips, 1993) have suffered similar
fates. These terms are used by educational researchers to refer to diverse
constructs and processes, weakening their original power to describe im-
portant phenomena. In order to be informative, a term such as distributed
cognition must distinguish research in some way from research done under
other rubrics. The danger that follows from vague application of a term is
that the term will be used to sanction a piece of work, but have no impact
on either the implementation or interpretation of the work.
We believe that the emerging perspective of distributed cognition has
the potential to significantly influence educational research. Our goal in
this paper is to describe the landscape of distributed cognition, with the
hope that this will help avert a slide into "fuzzy generality of reference"
(Nash, 1993, p. 58). We summarize the issues addressed by the papers in
the current volume, describe several interpretations of distributed cogni-
tion, and then locate these papers along a continuum of distributed cog-
nition perspectives. We then briefly consider the relationship between
distributed cognition and situated cognition, as these terms are sometimes
used interchangeably. Finally, we present the key issues we see for further
research in distributed cognition.

ISSUES ADDRESSED IN CURRENT VOLUME

The papers in this volume present an array of methodologies, frame-

works, and issues relevant to the exploration of distributed cognition. King
(1998) has developed a model for peer tutoring that emphasizes the im-
portance of understanding the process of learning, as well as the actual
learning of particular subject matter. In the ASK to THINK—TEL WHY
model, same-age same-ability partners are trained to ask thought-provoking
questions and to generate elaborated explanations. According to King, cog-
nitive and metacognitive processes are distributed between the two part-
ners, as well as distributed between the individuals and the instructional
materials that are being used during tutoring.
King places her work within a Vygotskian framework. Internalization of
cognitive skill is an important mechanism in learning. In this sociocultural
approach, thinking is viewed as distributed between the one who is guiding
participation in an activity, and another who is appropriating knowledge and
skills. King describes what she labels "transactive" cognitive partnerships, the
characteristics of which include "mutual appropriation" and interdependence
of interaction. Rather than a more expert individual guiding learning and a
Interpretations of Distributed Cognition 99

less expert individual appropriating knowledge and skills, as might be ex-

pected from a Vygotskian perspective, equal ability peers engage in mutual
appropriation in which both partners learn from each other.
King has used the constructs of distributed cognition effectively to pro-
vide an explanation for the efficacy she has observed for her ASK to
THINK—TEL WHY model. However, one problematic aspect of her
analysis is her characterization of transactive cognitive partnerships as
unique to interactions involving individuals who are roughly comparable
on a skill of interest. The notion of mutual appropriation is usually re-
garded to be an aspect of all interaction, not just interactions of individuals
who are similar on some dimension. Within the Vygotskian framework King
is drawing upon, there is always reciprocity in interaction and activity. Ap-
propriation is construed as multidirectional (Moschkovich, 1996; Newman,
Griffin, and Cole, 1989; Brown et al., 1993), and should not be viewed as
limited to the process by which a novice learns from an expert, but as an
aspect of all interaction.
Derry, DuRussel, and O'Donnell (1998) are working toward the de-
velopment of a model of thinking in interdisciplinary teams in order to
better understand the dynamics of group collaboration. They address the
issue of how group cognition drives the construction of intellectual products
within a group, whose goal is the design of products that would better en-
able team work. The particular groups that they are working with are in-
terdisciplinary teams that meet to further the agenda of the National
Institute for Science Education (NISE).
As a method, they use conversation analysis to examine group meet-
ings, and present an analysis of one meeting out of a series. In their frame-
work, they treat groups as cognitive entities that construct knowledge, and
they examine the processes by which conversation is translated into a group
product. They found that the degree to which a particular idea is distributed
(they consider an idea to be distributed across two individuals if it is men-
tioned by both of them) is not predictive of inclusion in a group product,
although they do not currently have an explanation for this result. Within
this framework, group cognition is analyzed using the same constructs as
individual cognition. Group interaction is accounted for through the cog-
nition of its individuals, but, unfortunately, there is no analytic unit pro-
vided for describing a uniquely interactive process.
Hewitt and Scardamalia (1998) consider the implications of the per-
spective of distributed cognition for classroom practices. They suggest that
classroom practices should be redesigned in order to create "knowledge
building communities"—groups that are dedicated to advancing the knowl-
edge of the collective. They offer CSILE (Computer Supported Intentional
Learning Environment) as a way to facilitate the distribution of knowledge
100 Moore and Rocklin

that will create knowledge building communities. CSILE is a student-gen-

erated networked computer database in which students enter "notes" on
particular topics. Notes are available to all students, and students are en-
couraged to build, comment, revise, and question each other's notes.
Through the technology of CSILE, a classroom environment is created that
allows intensive peer collaboration.
Hewitt and Scardamalia outline several design strategies for classroom
practices that support the distribution of knowledge. Methodologically,
Hewitt and Scardamalia suggest that psychologists study the person plus
their environment as an "indivisible unit of study." This is an alternative
position to that taken in the King and Deny et al. studies in which indi-
vidual and group processes were considered as analytically separable. Al-
though intriguing, the research implications of Hewitt and Scardamalia's
suggestion for studying a unit larger than an individual are unclear. They
delineate the design principles of a particular kind of an environment, but
do not discuss the research implications of those environments.
Lebeau (1998) characterizes the use of distributed cognitive tools, such
as medical histories and physical examinations, that are used in medical diag-
nosis, with the goal of addressing educational questions about learning to use
these tools. He uses the perspective of distributed cognition to reframe the
current portrayal of medical expertise by examining the implications of viewing
medical histories and physical examinations as distributed cognitive tools.
Lebeau constructs a framework by combining notions from sociocultural
psychology and ecological psychology. He uses the notion of appropriation
to characterize how medical students acquire the skills of medical diagnosis,
and the notion of affordances to discuss the support that exists for the activity
of medical diagnosis. Affordances are the support for particular activities cre-
ated by relevant properties of the world, which are specified relative to the
abilities of a particular organism (see Gibson, 1979; Greeno, 1994; Reed,
1988). Lebeau is concerned about the distribution of cognition between peo-
ple and physical artifacts, as well as the distribution of cognition between
people, but is primarily focused on the cultural aspects embedded in the prac-
tices in which medical students learn to engage. Despite his use of the term
affordances, Lebeau provides more an analysis of the practices of diagnosis,
than an analysis of the support (physical and social) for those activities.

DIFFERING INTERPRETATIONS OF DISTRIBUTED

COGNITION

Although the papers in this issue are quite diverse, the authors assume
that the more cognition is distributed the better a system (i.e., a pair of
Interpretations of Distributed Cognition 101

students, a classroom, a committee of academics, a physician, patient, and

other medical personnel) will function. The perspective of distributed cog-
nition is based on the assumption that characterizing cognition as distrib-
uted provides a more accurate conception of cognition than perspectives
that characterize cognition as an in-the-head phenomenon. However, it
does not necessarily follow from this assumption that the more cognition
is distributed the better a cognitive system will function. Rather than fo-
cusing on the degree of distribution, it seems more fruitful to model the
distributed functioning of cognitive systems and to analyze the properties
of the system that facilitate or hinder adaptive functioning. For example,
Lebeau (1998) analyzed the distributed functioning of a physician, patient,
other medical personnel, and medical artifacts, in the diagnosis of a medical
problem. This sort of analysis seems likely to lead to a greater under-
standing of the process of medical diagnosis, and to possible improvements
in the training of medical personnel. However, it is not necessarily the case
that medical diagnosis would be improved by distributing cognition as much
as possible throughout the system.
In addition, the papers in this issue focus on the design and use of
artifacts in the distribution of cognition. Artifacts include designed objects
such as tools, symbolic representations such as graphs, people in social re-
lations, and features of the natural environment (Pea, 1993). King discusses
the use of instruction sheets during peer tutoring, Derry et al. analyze re-
cord keeping within a team, Lebeau considers medical charts and other
medical artifacts, and Hewitt and Scardamalia address CSILE notes. Al-
though an analysis of the cognitive support provided by artifacts is theo-
retically productive, the study of distributed cognition must also include an
analysis of the processes of cognition and how interaction with artifacts
influences or constitutes these processes.
King, Derry et al., and Lebeau are concerned with the development
of individual competency. They are interested in the relationship between
group activity, the use of artifacts, and how these influence individual cog-
nition. Ultimately, they are interested in the "cognitive residue" (Perkins,
1993) that group activity provides an individual. In contrast, Hewitt and
Scardamalia consider the functioning of the classroom as a whole, and do
not analyze the functioning of particular students within a class.
Lebeau and King differ on their stance toward the relationship be-
tween appropriating skills and content knowledge. Lebeau states that the
medical interview is not a skill that can be learned independently from
medical knowledge of illnesses. In contrast, King teaches learning strategies
that are meant to be independent of any particular subject matter. An im-
plication of Lebeau's position is that teaching content-specific interactional
strategies would be more effective than attempting to teach content-free
102 Moore and Rocklin

learning strategies. For example, learning what counts as an acceptable

mathematical explanation should be more effective than learning that in
general it is important to justify one's interpretations (cf. Cobb and Yackel,
1996).
These papers illustrate the range of interpretations of distributed cog-
nition. One perspective on the continuum of interpretations of distributed
cognition is an elaboration of information-processing psychology, which fits
nicely with a standard cognitive science approach (Nardi, 1996). Cognition
is conceptualized as an individual phenomenon, but the importance of fac-
tors external to the individual is recognized. Knowledge is something that
individuals can possess, but sometimes that knowledge is distributed around
a variety of artifacts and other individuals. Group processes are seen as a
way to facilitate individual learning. Thus, individual and distributed cog-
nitions are viewed as distinct phenomena that exist in an interdependent
dynamic interaction (Salomon, 1993a). We might call this the "individual-
plus" view of distributed cognition (using terms from Perkins, 1993; Lave,
1991). This perspective issues no challenge to current notions of individual
cognition, but advocates the consideration of additional influences on in-
dividual cognition. Group processes are often described by the same con-
structs and processes as individual cognitive processes (e.g., Hutchins,
1995a,b; Salomon, 1993b). One research question that follows from this
perspective is how working with others or various artifacts, such as com-
puters, influence individual cognition.
The Derry et al. paper provides a clear example of the "individual-
plus" approach. They support the idea that information-processing con-
structs can be applied to the workings of groups, and that the psychology
of individuals and the psychology of distributed cognitions should be ac-
commodated within the same theoretical framework (explicitly drawing on
the work of Salomon, 1993b). We will have more to say later about the
challenges of accommodating the psychology of distributed cognition in the
same framework as individual cognition. Derry et al. use an information-
processing model to provide insight into the difficulties in the processing
of information by groups. This analysis differentiates the influence of dis-
tributed and individual cognition on the products of a group.
King's work on peer tutoring also falls into the individual-plus camp.
The study is couched in the sociocultural terms of internalization and ap-
propriation, but King's concern is primarily about the effects of interacting
with others and artifacts on individual cognition and metacognition. Ap-
propriation and internalization are seen as the mechanisms by which indi-
viduals become more expert (i.e., how they construct the skills and
knowledge that others have). Again, individual cognition is the primary con-
cern, with interactions with the world—including other individuals—seen
Interpretations of Distributed Cognition 103

as an explanatory mechanism for individual cognition. Internalization is the

mechanism by which an individual learns from the social world (i.e., how
distributed cognition becomes individual). Thus, within the individual-plus
approach, the notion of internalization provides an explanation for how
the social world and the individual come to share a good deal in common
(Lave, 1991).
The individual-plus perspective captures much of the work being done
in distributed cognition. There is, however, an alternative position that
moves further from the traditional notions of information-processing psy-
chology. This interpretation of distributed cognition is that "learning, think-
ing, and knowing are relations among people in activity in, with, and arising
from the social and culturally structured world" (Lave and Wenger, 1991,
p. 51). In this view, cognition is neither fully internalized as knowledge
structures, nor fully externalized as activity (Lave and Wenger, 1991). This
position rejects both the information-processing paradigm, as well as avoids
a relapse into behaviorism (however, for a different interpretation of this
issue see Nardi, 1996). Cognition cannot be explained as either a primarily
in-the-head phenomenon, nor as activity alone. Cognition is viewed as a
social phenomenon, referred to as the "social-only" perspective of distrib-
uted cognition. Social processes are considered as cognition (L. B. Resnick,
1991), which cannot be reduced to individual psychological constructs.
Theorists who adhere to this perspective of distributed cognition argue that
a view of cognition is incomplete if it does not consider social and cultural
aspects. The notion of cognition is fundamentally distorted when it is stud-
ied as a purely psychological phenomenon (Lave, 1988). Research questions
that follow from this perspective include the question of how social norms
are dialectically related to group activity.
The Hewitt and Scardamalia paper on the design principles of CSILE
captures some aspects of this second perspective. The authors emphasize
the work of the community over the work of an individual. One goal of
the project is to create an environment in which students in a CSILE class-
room view their own work in the context of group work. Hewitt and Scar-
damalia suggest the intriguing possibility that a CSILE classroom is a
self-improving system. If a classroom is either a self-organizing system (e.g.,
M. Resnick, 1991) or a dynamic system (e.g., Thelen and Smith, 1994),
then there are characteristics of the system that cannot be found by con-
sidering characteristics of individuals alone. This implies a need to abandon
top-down approaches to the study of group behavior (i.e., stop looking for
who's running the system) and consider new constructs to explain the be-
havior of the classroom as a whole. Although intriguing, the research im-
plications of considering the classroom as a self-improving system are not
clear from the paper.
104 Moore and Rocklin

The Lebeau paper on physician-patient interactions captures aspects

of both interpretations of distributed cognition. He views the medical in-
terview as a "socially shared tool" to facilitate individual cognition. Lebeau
claims that individuals appropriate the tools of distributed activity, and he
is interested in how individuals change as a result of this appropriation.
Lebeau uses the term appropriation to mean adopting the practices of a
community of practitioners. As with King's concept of internalization, the
concept of appropriation explains how distributed cognition might become
individual cognition. Lebeau believes various people contribute to joint
problem solving, but he focuses on the improvement of individuals within
a system, rather than the functioning or improvement of the system as a
whole.
As mentioned above, Deny et al. differentiate the influence of distrib-
uted and individual cognition on the products of a group. Note that this
distinction between individual and distributed cognition would not make
sense within the social-only perspective. Because the social-only perspective
views cognition as a social phenomenon, a distinction between individual
and group cognition is meaningless.
Salomon (1993b) opposes the social-only perspective, arguing that psy-
chologists have gone too far in dismissing the individual from analyses. One
cannot construct an adequate theory of distributed cognition without an
explicit role for individual cognition. A major issue dividing the individual-
plus and the social-only perspectives is whether all cognition is viewed as
distributed, or whether only some cognition is viewed as distributed. In the
former case, the assumption is that cognition is distributed by definition,
and the research issues involve how cognition is distributed in various situ-
ations. In the latter case, cognition is viewed as distributed in some situ-
ations and individual in others. When cognition is viewed as individual, it
is characterized in terms of information-processing constructs. We continue
the discussion of the role of information-processing constructs in distributed
cognition in a later section.

THE RELATIONSHIP BETWEEN DISTRIBUTED

COGNITION AND SITUATED COGNITION

Complicating the portrayal of the uses of the notion of distributed cog-

nition is the fact that the term distributed cognition is sometimes not distin-
guished from the use of the term situated cognition. For example, in the
current issue, Hewitt and Scardamalia do not discriminate between the use
of "distributed" and "situated." A position labeled as "situative/pragmatist-
sociohistorical" in Greeno, Collins, and Resnick (1996) views knowledge as
Interpretations of Distributed Cognition 105

"distributed among people and their environment, including the objects, ar-
tifacts, tools, books, and the communities of which they are a part" (p. 17).
Although labeled as "situative," this position shares much with the interpre-
tations of distributed cognition. There are strong similarities between the
various interpretations of distributed and situated cognition, critiques of the
positions (e.g., Vera and Simon, 1993; Anderson, Reder, and Simon, 1996),
and applications of the positions to the development of learning environ-
ments and empirical work (e.g., Salomon, Perkins, and Globerson, 1991;
Young, 1993; Choi and Hannafin, 1995). Lave (1991) has differentiated three
genres of situated approaches that help illustrate the similarities of these ap-
proaches to the interpretations of distributed cognition.
In an approach Lave labeled "cognition plus," research on individual
cognition remains unchallenged, but social factors are now considered as
one of many influences on individual cognition. The view of situated cog-
nition that Lave has described as cognition plus and the individual-plus
view of distributed cognition are similar both in name and concept. In ad-
dition, they are both similar to Perkins' (1993) notion of "person-plus" to
describe a person and the surround as the proper unit of analysis. These
positions do not challenge the tenets of current information-processing psy-
chology but encourage the exploration of additional factors that might in-
fluence individual cognition.
In a second approach that Lave labeled "interpretative," situatedness
is found in language and/or social interaction. There is no world inde-
pendent of individuals' construction of it, and meanings are located in ne-
gotiated social interaction. Here the term "situated" is not physical but
social. The position that Lave has labeled interpretive does not seem to
have a direct counterpart in the distributed cognition literature, and it
seems most closely aligned with a radical constructivist position (e.g., von
Glasersfeld, 1984, 1990).
Lave labeled the third approach "situated social practice," which is
similar to the social-only position. In this approach, learning is a relation
among people "engaged in activity in, with, and arising from the socially and
culturally structured world" (Lave, 1991, p. 67, emphasis in original). Cog-
nition is situated in the historical development of ongoing activity. The situ-
ated cognition genre that Lave has labeled situated social practice and the
distributed cognition genre that Salomon (1993a,b) has described as cog-
nition as an irreducibly social phenomena are also similar. It is significant
that Lave is often cited as a source for both these positions (e.g., in Salo-
mon, 1993a,b for distributed cognition; in Greeno, Collins, and Resnick,
1996 for situated cognition).
Despite the similarities between distributed and situated cognition out-
lined above, and the fact that many researchers do not distinguish between
106 Moore and Rocklin

these terms in practice, Pea (1993) reserves the term "situated" to mean
tied to the physical world. Pea's interpretation of the situated approach
emphasizes that a reasoner can exploit features of the physical world, avoid-
ing the need to perform mental symbol manipulations unless necessary
(e.g., Scribner, 1988). He also criticizes the work on situated cognition for
failing to acknowledge the fundamental role of artifact design in the activity
of the reasoners, which he sees as an important contribution of distributed
cognition. However, it should be noted that the proponents of situated cog-
nition explicitly include the social, as well as the physical world, as impor-
tant aspects of their position (e.g., Lave and Wenger, 1991; Suchman, 1987;
Greeno, Smith, and Moore, 1993; Greeno et al., 1996).
In our view, much of the work labeled as situated or distributed cogni-
tion falls under the cognition/individual-plus rubric, and seems equally well
described by either nomenclature. Thus, at this point, it is not clear that the
distinction is useful. A more useful distinction might be how the individual
is conceived, as in the distinction between individual-plus and social-only. We
believe that this distinction will continue to be pivotal for work in distributed
cognition; research questions, methodologies, and empirical findings will be
organized around this theme, implicitly or explicitly. In the next section, we
discuss several issues that we believe are important, but understudied, with
respect to distributed cognition. These issues may be viewed differently from
the individual-plus or social-only perspectives but need to be addressed re-
gardless of perspective. One important issue is the development or adoption
of research methodologies to address new concerns raised by distributed cog-
nition. A central question is whether methodologists should focus on indi-
viduals interacting with other people and artifacts, the functioning of a
cognitive system, or some combination of the two. Another key issue contin-
ues to be the role of information-processing constructs in the development
of a theory of cognition as distributed. Will information-processing constructs
continue to have a role in the study of distributed cognition, or will they be
replaced by other theoretical constructs? Finally, those who study distributed
cognition have highlighted the functioning of individuals in groups, raising
questions about the role of individual differences in group activity, as well as
the political dimensions of individual interaction.

KEY ISSUES FOR RESEARCH

What impact will adopting a distributed cognition perspective have on

research methodology? An important issue is how current paradigms will
need to be changed to adequately address the questions raised by adopting
a distributed cognition perspective. Both interpretations of distributed cog-
Interpretations of Distributed Cognition 107

nition raise the possibility that psychologists may need to study issues that
have traditionally been the venue of other social sciences.
According to Salomon (1993b), the individual-plus approach implies
that studies should neither be restricted to individuals, nor controlled, ar-
tificial settings. These properties are illustrated by each of the papers in
this issue. Peer tutoring in a classroom setting, group meetings, patients'
histories collected by medical personnel, and a classroom database are
naturalistic settings that involve the interaction of multiple individuals.
Salomon (1993b) also distinguishes the division of cognitive labor from
shared cognitive labor. The division of cognitive labor involves an individual
shifting cognitive responsibility onto a tool(s) or a person(s). When con-
sidering the division of labor, the focus is on the individual. In contrast,
shared cognitive labor occurs when individuals jointly engage in some ac-
tivity. This is collaborative activity, where each action has the potential to
change the joint activity. When considering shared cognitive labor, the fo-
cus is on the cognition of multiple individuals. King's study (1998) of peer
tutoring illustrates this distinction. King characterizes peer tutoring as in-
volving a student off-loading cognitive responsibilities onto another student
as well as onto the tutoring instructions. In addition, she discusses the proc-
ess by which both students jointly engage in understanding written material.
Although the individual-plus approach may shift the research focus, it does
not seem to require a departure from current research techniques.
A tenet of the social-only interpretation of distributed cognition is that
the "mind" cannot be studied independently of the culturally organized set-
tings within which people function. However, it is not obvious what it means
operationally to study the person/environment as an "indivisible unit of
study" (Hewitt and Scardamalia, 1998). The individual-plus approach con-
ceptualizes cognition as divided among an individual, objects, and other peo-
ple; whereas the social-only approach conceptualizes cognition as "stretched
over" people and objects (Lave, 1991). Hewitt and Scardamalia are not pri-
marily concerned with the individual work of the children in CSILE class-
rooms, choosing instead to look at the behavior of the classroom as a whole.

APPLICABILITY OF INFORMATION-PROCESSING
CONSTRUCTS

Can the techniques and constructs that have developed in the study
of individual cognition be applied to the study of distributed cognition?
Clearly, there are varying positions on this issue within the distributed cog-
nition approaches. Salomon (1993b) and Derry et al. (1998) explicitly state
that they are attempting to apply the constructs of individual cognition to
108 Moore and Rocklin

the study of groups. Intriguing questions are raised about whether the proc-
esses of cognition that are distributed are the same as the processes of
individual cognition. If one imposes the psychology of individuals on
groups, then groups must have memory and memory processes. Where do
short-term memory and long-term memory get stored in distributed sys-
tems? Do groups have metacognition? Even from a stance in which indi-
vidual cognition plays an important role, the processes and structures of
information-processing psychology may not map onto group cognition.
Nickerson (1993) suggests that individual cognition and distributed cogni-
tion are so different that different terms should be used.
Hutchins (1995a) takes the more intermediate stance that the same infor-
mation-processing constructs applicable to individuals are applicable to groups,
buy they are manifested differently at the individual and group level. He be-
lieves that systems larger than individuals have cognitive properties in their own
right that are not reducible to the cognitive properties of individuals. Hutchins
applies the classical cognitive science approach to a unit of analysis that is larger
than a person, and has developed a model to demonstrate that a cognitive prop-
erty such as confirmation bias or memory retrieval efficiency is manifested dif-
ferently at the individual and group level (Hutchins, 1991, 1995b).
Finally, the proponents of the social-only approach to distributed cog-
nition claim that information-processing constructs have fundamentally dis-
torted the study of cognition. There is no room for information-processing
constructs in this perspective unless one can characterize the social inter-
action that produces shared understanding.

GROUP ACTIVITY

One clear implication of the distributed cognition approach, regardless

of perspective, is that psychologists find some way to study the functioning
of cognitive systems. These systems may involve groups of interacting in-
dividuals (e.g., Deny et al., 1998), or groups of individuals operating in a
system in which cultural artifacts play an important role (e.g., Lebeau,
1998), or individuals interacting with artifacts. There are two issues that
we propose are understudied with respect to group functioning - the role
of individual differences and the role of political and power differentials.

DIFFERENCES AMONG INDIVIDUALS

One of the more serious challenges is the incorporation of differences

among the people that comprise any particular unit of study. At a funda-
Interpretations of Distributed Cognition 109

mental level, the study of cognition is about the ways in which information
is transformed by various units or processes. In traditional information-
processing models these units might be sensory memory, working memory,
and long term memory. In a model of distributed cognition, the units must
include, at some level, the people (as well as the inanimate tools) involved
in the interaction. The differences among these people have important im-
plications for how the information flows and is transformed in the system.
Consider, for example, one way in which people involved in a distrib-
uted cognition process might differ. Some of the people might be women,
and others men. As Tannen (1994) has pointed out, men and women in
many situations—including work—communicate in qualitatively and quan-
titatively different ways. Men are more likely to see asking questions as
putting the questioner in a weak position. They are therefore more reluc-
tant to ask questions and are more careful about when and where they ask
questions. Another example involves what Tannen calls "ritual opposition."
Men more frequently engage in heated arguments and more vigorously at-
tack each other and each other's ideas. Frequently, women respond to these
attacks without accepting their ritual nature. They may give up on a idea
that is challenged, assuming that the vigor of the challenge is an indication
of the low quality of their idea.
Both of these examples suggest that a model of distributed cognition
that does not consider the gender of the group members is seriously in-
complete. If men and women differ in the situations and ways in which
they ask questions, is it possible to understand the processes involved in
the ASK to THINK–TEL WHY interactions (King, 1998) without includ-
ing gender in the analysis? Similarly, if one wants to understand—as do
Derry et al. (1998)—why certain ideas are included in a "to-do" list and
others are not, it might be relevant to examine the gender of the originator
of each of the ideas.
Gender is not the only characteristic that affects the ways in which
people communicate. Another example involves the role of status and
power in the use of indirectness. Messages can vary from quite direct
("Please type, proof, and print this by 3:00.") to quite indirect ("Do you
think you have time to type this this afternoon?"). The use of indirectness
is quite complex, but there are clearly some situations in which low power
and status are associated with indirectness. Imagine, for example, the dis-
tributed cognition of a faculty surgeon and a surgical resident (i.e., a sur-
geon in training). As they work to mobilize the patient's colon, the resident
might remark, as though asking for information, "Is that linear structure
you're considering cutting the ureter?" rather than saying "Stop, you're
about to cut the ureter!" This example speaks directly to the question of
which properties of cognition are distributed and how those properties are
110 Moore and Rocklin

distributed. In this case, the skills, knowledge, and other attributes (perhaps
including simply location at the operating table) needed to recognize an
anatomical structure are at least slightly unevenly distributed. Whether the
problem (i.e., cutting the right structures and leaving other structures in-
tact) is solved effectively depends on how knowledge, skills, and informa-
tion are distributed with respect to status and power. More specifically, the
problem is more likely to be solved effectively if the knowledge, skills, and
information are associated with high status and power.
To close, we emphasize two points. First, the examples we cite involve
mean differences between groups of people (e.g., between men and
women). Although it is of interest to know that particular ways of commu-
nicating are associated with particular characteristics of the communicators,
it is likely more important to study the variety of ways in which people
communicate (e.g., use of questions) and to include those in models of
distributed cognition than it is to study the group differences. We expect
that although such subject variables as gender might provide an appropriate
starting point, ultimately effective models of distributed cognition will in-
clude styles of communication, for example, rather than these subject vari-
ables. In other words, ritual opposition is more common among men than
women, but it is the ritual opposition, whether it comes from a man or a
woman, that is of interest. Secondly, we emphasize that the examples we
have chosen are intended only to be illustrative. We make no claims that
these are the most important issues in the domain of communication styles
for distributed cognition researchers to consider.

POLITICAL CONTEXT

To understand the ways in which cognitive systems work psychologists

will need to confront issues such as political context and power differentials
among individuals. To date, this has been outside the venue of research on
cognition. In order to cope with these issues, research techniques may need
to be expanded in order to adequately address these issues, perhaps pil-
fering from other social sciences.
For example, political issues play no role in the current papers, or in
many studies of distributed cognition. In addition, most would not question
the fact that power differentials influence interactions among individuals
(e.g., Linde, 1988); however, this issue is also not addressed in much of the
work on distributed cognition. Lebeau (1998) states that cognition is distrib-
uted between a physician and a patient, and Deny et al. (1998) are interested
in the interactions between team leaders and team members. In both of these
examples, cognitive activity is taking place in a context of a broader political
Interpretations of Distributed Cognition 111

arena, as well as an arena of personal status differentials. For example, in

the Deny et al. study, there is a need to satisfy a funding agency (i.e., NISE),
and a desire to increase the number of perspectives represented within the
team. Are these issues that should be included in an analysis of the distribu-
tion of knowledge and the cognitive functioning of this team? These factors
certainly have the potential to influence the cognitive processing of this
group, yet they are not factors that have traditionally been studied with re-
spect to individual cognition. As the unit of analysis increases, the unresolved
issue is how broad a stroke of context must be painted in order to describe
the behavior of a group or individual validly.

CONCLUSIONS

We have suggested two approaches to the study of distributed cogni-

tion, a viewpoint that is rapidly gaining in popularity in the educational
literature. We discussed several interpretations of situated cognition, an-
other perspective that has recently gained many advocates. We compared
the two perspectives and suggested similarities in how the two terms are
being used. We then considered the implications of the notion of distributed
cognition for educational research.
We press for greater clarity in the term distributed cognition. Although
consensus is not necessarily the most desirable outcome, a more detailed
and explicit definition is needed. The potential impact of the perspective
of distributed cognition on the field of educational psychology is significant,
and we would like to avoid that impact dissolving into a sea of "fuzzy gen-
erality of reference" (Nash, 1993, p. 58). A new perspective will affect the
discipline only if it changes the types of questions asked or the research
approaches used. To merely sanction existing research with a new term,
without affecting its implementation or interpretation, will not move the
field forward.

REFERENCES

Anderson, J. R., Reder, L. M., and Simon, H. A. (1996). Situated learning and education.
Educ. Res. 25(4): 5-11.
Brown, A. L., Ash, D., Rutherford, M., Nakagawa, K., Gordon, A., and Campione, J. (1993).
Distributed expertise in the classroom. In Salomon, G. (ed.), Distributed Cognitions: Psy-
chological and Educational Considerations, Cambridge University Press, New York, pp.
188-228.
Choi, J., and Hannafin, M. (1995). Situated cognition and learning environments: Roles, struc-
tures, and implications for design. Educ. Technol. Res. Devel. 43(2): 53-69.
Cobb, P., and Yackel, E. (1996). Constructivist, emergent, and sociocultural perspectives in
the context of developmental research. Educ. Psychol. 31(3/4): 175-190.
112 Moore and Rocklin

Deny, S. J., DuRussel, L. A., and O'Donnell, A. M. (1998). Individual and distributed cog-
nitions in interdisciplinary teamwork: A developing case study and emerging theory. Educ.
Psychol. Rev. 10: 25-56.
Gibson, J. J. (1979). The Ecological Approach to Visual Perception, Houghton Mifflin, Boston.
Greeno, J. G. (1994). Gibson's affordances. Psychol. Rev. 101(2): 336-342.
Greeno, J. G., Smith, D. R., and Moore, J. L. (1993). Transfer of situated learning. In Det-
terman, D. K., and Sternberg, R. J. (eds.), Transfer on Trial: Intelligence, Cognition, and
Instruction, Ablex, Norwood, NJ, pp. 99-167.
Greeno, J. G., Collins, A., and Resnick, L. (1996). Cognition and learning. In Berliner, D.,
and Calfee, R. (eds.), Handbook of Educational Psychology, Simon and Schuster Macmil-
lan, New York, pp. 15-46.
Hewitt, J., and Scardamalia, M. (1998). Design principles for distributed knowledge building
processes. Educ. Psychol. Rev. 10: 75-96.
Hutchins, E. (1991). The social organization of distributed cognition. In Resnick, L. B., Levine,
J. B., and Teasley, S. D. (eds.), Perspectives on Socially Shared Cognition, American Psy-
chological Association, Washington, DC, pp. 283-307.
Hutchins, E. (1995a). How a cockpit remembers its speeds. Cogn. Sci. 19: 265-288.
Hutchins, E. (1995b). Cognition in the Wild, MIT Press, Cambridge, MA,
King, A. (1998). Transactive peer tutoring: Distributing cognition and metacognition. Educ.
Psychol. Rev. 10: 57-74.
Lave, J. (1988). Cognition in Practice: Mind, Mathematics, and Culture in Everyday Life, Cam-
bridge University Press, Cambridge.
Lave, J. (1991). Situating learning in communities of practice. In Resnick, L. B., Levine, J.
M., and Teasley, S. D. (eds.), Perspectives on Socially Shared Cognition, American Psycho-
logical Association, Washington, DC, pp. 63-82.
Lave, J., and Wenger, E. (1991). Situated Learning: Legitimate Peripheral Participation, Cam-
bridge University Press, Cambridge.
Lebeau, R. (1998). Cognitive tools in a clinical encounter in medicine: Supporting empathy
and expertise in distributed systems. Educ. Psychol. Rev. 10: 3-24.
Linde, C. (1988). The quantitative study of communicative success: Politeness and accidents
in aviation discourse. Lang. Soc. 17: 375-399.
Moschkovich, J. N. (1996). Moving up and getting shorter: Negotiating shared descriptions
of linear graphs. J. Learn. Sci. 5: 241-279.
Nardi, B. (1996). Studying context: A comparison of activity theory, situated action models,
and distributed cognition. In Nardi, B. (ed.), Context and Consciousness: Activity Theory
and Human-Computer Interaction, MIT Press, Cambridge, MA, pp. 69-102.
Nash, W. (1993). Jargon: Its Uses and Abuses, Blackwell, Cambridge, MA.
Newman, D., Griffin, P., and Cole, M. (1989). The Construction Zone: Working for Cognitive
Change in School, Cambridge University Press, Cambridge, UK.
Nickerson, R. S. (1993). On the distribution of cognition: Some reflections. In Salomon, G.
(ed.), Distributed Cognitions: Psychological and Educational Considerations, Cambridge
University Press, New York, pp. 229-261.
Pea, R. (1993). Practices of distributed intelligence and designs for education. In Salomon,
G. (ed.), Distributed Cognitions: Psychological and Educational Considerations, Cambridge
University Press, New York, pp. 47-87.
Perkins, D. N. (1993). Person-plus: A distributed view of thinking and learning. In Salomon,
G. (ed.), Distributed Cognitions: Psychological and Educational Considerations, Cambridge
University Press, New York, pp. 88-110.
Phillips, D. C. (1993). Paul Hirst's structure, or, the uses and abuses of an overworked concept.
In Barrow, R., and White, P. (eds.), Beyond Liberal Education: Essays in Honor of Paul
H. Hirst, Routledge, New York, pp. 79-93.
Reed, E. S. (1988). James Gibson's ecological approach to cognition. In Costall, A., and Still,
A. (eds.), Cognitive Psychology in Question, St. Martin's Press, New York, pp. 142-173.
Resnick, L. B. (1991). Shared cognition: Thinking as social practice. In Resnick, L B., Levine,
J. M., and Teasley, S. D. (eds.), Perspectives on Socially Shared Cognition, American Psy-
chological Association, Washington, DC, pp. 1-20.
Interpretations of Distributed Cognition 113

Resnick, M. (1991). Beyond the centralized mindset. In Birnbaum, L. (ed.), Proceedings of

the International Conference on the Learning Sciences, Association for the Advancement
of Computing in Education, Charlottesville, NC, pp. 389-396.
Salomon, G. (1993a). Editor's introduction. In Salomon, G. (ed.), Distributed Cognitions: Psy-
chological and Educational Considerations, Cambridge University Press, New York, pp.
xi-xxi.
Salomon, G. (1993b). No distribution without individuals' cognition: A dynamic interactional
view. In Salomon, G. (ed.), Distributed Cognitions: Psychological and Educational Consid-
erations, Cambridge University Press, New York City, pp. 111-138.
Salomon, G., Perkins, D. N., and Globerson, T. (1991). Partners in cognition: Extending human
intelligence with intelligent technologies. Educ. Res. 20(3): 2-9.
Scribner, S. (1988). Studying working intelligence. In Rogoff, B., and Lave, J. (eds.), Everyday
Cognition: Its Development in Social Context, Harvard University Press, Cambridge, MA,
pp. 9-40.
Suchman, L. A. (1987). Plans and Situated Action, Cambridge University Press, Cambridge.
Tannen, D. (1994). Talking from 9 to 5: How Women's and Men's Conversational Styles Affect
Who Gets Heard, Who Gets Credit, and What Gets Done at Work, William Morrow, New
York.
Thelen, E., and Smith, L. B. (1994). A Dynamic Systems Approach to the Development of Cog-
nition and Action, MIT Press, Cambridge, MA.
Vera, A. H., and Simon, H. A. (1993). Situated action: A symbolic interpretation. Cogn. Sci.
17(1): 7-48.
von Glasersfeld, E. (1984). An introduction to radical constructivism. In Watzlawick, P. (ed.),
The Invented Reality, Norton, New York, pp. 17-40.
von Glasersfeld, E. (1990). An exposition of constructivism: Why some like it radical. In Davis,
R. B., Maher, C. A., and Noddings, N. (eds.), Constructivist views of the teaching and
learning of mathematics. J. Res. Math. Educ.: Monogr. 4: 19-30.
Young, M. F. (1993). Instructional design for situated learning. Edu. Technol. Res. Devel. 41(1):
43-58.