I.  INTRODUCTION

This study has, as it's title implies, two primary objectives which are woven together throughout the presentation.  First, the study is about the lives of fishermen who live on an island off the west coast of Sweden.  The people of this community have, for hundreds of years, engaged in various forms of fishing as their primary means of existence.  Fishing has played a central role in the lives of these people—early on in subsistence terms, later on in economic terms, and always in terms defining who they are to themselves and others.  The study is also about reconceptualizing a theory of culture and cognition.  The primary aspect of this second objective entails formulating—through empirical means—principles about the ways behavior is (and becomes) organized, and about the consequences of this organization for social and cognitive life.  In the theory of distributed cognition employed herein, cognitive actions are mediated by internal structures which result from embodied experience with (and, simultaneously, result in) the structure and processes of social practices and cultural histories of which the actions are part.  The theory entails an information processing account of the material, social, and historical means for organizing behavior.

In order to accomplish these objectives, different units of analysis than those typical of anthropology and cognitive science are employed.  At the same time, the analysis also draws upon important elements of methodology and theory from both of these fields.  For instance, while most of cognitive science chooses to bound the system of analysis at the skin of the individual—effectively ignoring the role of extra-individual and historical phenomena in the explanation of cognitive events—the analysis of this study does not privilege the boundaries of individuals, per se.  While Chapter 2 does address the ways individuals organize experience (and thereby make meaningful certain forms of behavior for themselves as individuals), the mechanisms which enable this organization are seen to be constructed through participation in a material, social, and historical order.  Only through these mechanisms—as instantiated in this larger context—does the organization of experience come about, and only through an examination of the working of these mechanisms in context does the organization of experience become tangible to us as analysts.  Chapter 3 presents a descriptive account of the practice of pair-trawling for herring.  It demonstrates how historical processes entailing change and persistence in technology and social organization have shaped the nature of resources employed by participants in their activities as fishermen.  These resources are shown to be important mediators of cognitive performance but are rarely considered in traditional treatments by cognitive science.

On the other hand, anthropological theory of cognition, while sensitive to the historical properties of human living, has been unable to unify theories of culture and behavior.  Culture is generally taken to be the mental contents, understandings, or knowledge of (often idealized) members of a social group—leaving the processes by which organized behavior is accomplished either quite mysterious or tautologically grounded in cultural knowledge.  Again, a shift in the units of analysis proves helpful—this time from the static description of idealized conceptual structure to the dynamic description of embodied performance in real situations, shaped by material structures and socio-historical processes.

Chapter 4 provides an analysis of a real-time sequence of events which takes place during the activity of searching for where to set the trawl by a pair-trawling team of ships.  Cognitive events are here analyzed in terms of captains' time-course constructions and manipulations of resources for accomplishing the familiar, yet complexly dynamic, tasks involved.  These resources are seen to be both external and internal to individuals—in the social and material means of the practice and the abilities acquired through performance with these means—creating the functional systems by which complex problems and computations are made simpler and the practice reproduced.  The system of activity so constituted is shown to exhibit emergent properties which are not simply the sum of the abilities of the individuals involved.  Rather, the system exhibits properties which result from the interactions between participants and the tools at their disposal.

Some of the most key cognitive resources turn out to be those created in the communication which takes place between captains of the team working within a loosely cooperative fleet of such pair-trawling teams.  A large part of this process entails the establishment of intersubjectively shared accounts of fish sightings which are mediated by sonar representations of herring and the linguistic coding of those representations.  Chapter 5 presents an experiment conducted to investigate the nature of this intersubjectivity as a product of performing in different practices (employing different techniques) of fishing.  The experiment provides solid evidence for the proposition that the sharing of interpretive and communicative skills regarding sonar displays of fish is patterned upon the specific histories of experience fishermen have engaging each other as interlocutors and interpreters in these tasks.  The experiment also provides evidence suggesting that the linguistic representations employed to talk about sonar displays and fish are important cognitive resources employed in establishing the meanings of those sightings for practitioners.

In the remainder of this introductory chapter, the theoretical framework used to guide the data collection and analysis of the study is presented.  First, a brief account of the intellectual history of cognitive anthropology, including an assessment of the current state of theory in this and allied disciplines of cognition, is discussed.  This discussion leads to a statement of the failings of current theory to accommodate a number of important aspects of the relationship between culture and cognition.  Finally, a theory of "distributed cognition" is outlined as a set of conceptual tools for addressing these problems and unifying the chapters which follow.

The state and history of theory in Cognitive Anthropology

There is a history of at least 35 years in cognitive anthropology of trying to describe the "native's point of view" by making explicit local conceptual categories which underlie native interpretations of the world.  In early work, this entailed reconstructing indigenous semantic systems from the relationships between constituent terms of the respective languages (cf. Conklin, 1962; Frake, 1962; Goodenough, 1956; Lounsbury, 1956).  In its cross-cultural form, this research often sought to provide evidence for biological properties which might constrain knowledge structures and the cultural phenomena which entail this knowledge (cf. Berlin & Kay, 1969; Kay & McDaniel, 1978; Romney & D'Andrade, 1964; Wallace, 1961).  Later forms of research have been extended to address relationships between the organization of knowledge and: perception, inferences, emotions, conceptualization, language, motivation, and reasoning.  (See edited volumes by D'Andrade & Strauss, 1992; Dougherty, 1985; Holland & Quinn, 1987).  This effort has been multifaceted and productive but is not without shortcomings.  Before addressing these issues, it is worth briefly considering how this branch of anthropology differs from other kinds of anthropological investigation.  I begin with an empirical claim:

Cultural anthropology (of all stripes) entails describing interpretive processes: the ways individuals located in time and space (within a cultural history and social and physical space), evaluate their situations with some consequence (be it actions, emotions, thoughts, feelings, or what have you).

All modern ethnographic accounts include the specific and regular acts of interpretation which individuals perform, as part of the researcher's attempt to communicate a picture of the social life of some historically bounded group of people.  Whether one is describing an ethos, a world view, a ritual, a performance, or a social structure, examples of interpretive acts of individuals in the group are used to make the description convincing.  Take, for example, the nuances of a speech act—the range of contexts in which it is used, the variety of forms it takes, how it compares with other similar and complement acts, the reactions it engenders, the emotions it evokes, the expectations it produces—all are used as evidence in support of the claim that it is a speech act, and not an accidental puff of wind, that is at stake.

Exemplifying the problem of (and for) ethnography—in this case making sense of an act we might want to call a ‘wink'—Geertz (1973:7) makes a similar claim:  "[B]etween what Ryle calls ‘thin description' of what the rehearser (parodist, winker, twitcher ...)  is doing (‘rapidly contracting his right eyelids') and the ‘thick description' of what he is doing (‘practicing a burlesque of a friend faking a wink to deceive an innocent into thinking a conspiracy is in motion') lies the object of ethnography:  a stratified hierarchy of meaningful structures in terms of which twitches, winks, fake-winks, parodies, rehearsals of parodies are produced, perceived, and interpreted, and without which they would not ..... in fact exist, no matter what anyone did or didn't do with his eyelids."  No modern ethnography fails to describe such acts (whether of speech or gesture) in communicating the particular picture of society the author has experienced.  In this sense I am arguing that all cultural anthropologists engage in the activity of describing interpretive processes.

Cognitive anthropology, however, has primarily been engaged in describing what members of a culture know (including how the organization of what they know, and internal constraints upon knowing, effect what they know) rather than describing interpretive processes more generally.  Methodologically, this has entailed much more explicit descriptions of relationships between culture and cognition.  Theoretically (and perhaps, partly, as a result of the methodology) the unexamined assumptions of cognitive anthropology have always been that: knowledge is the more or less exclusive determinant of interpretative processes; knowledge is usually composed of (or, at least, is optimally described in terms of) propositional structures; knowledge is transmitted through socialization in a group; and knowledge comes to rest in the heads of competent individuals of the group.[1]

Hutchins (1980) summarized a piece of this intellectual history in a comparison of the problems faced by earlier cognitive anthropology and cognitive psychology.

It seems that the search for representations of what people know, as exemplified by the ethnoscience tradition in anthropology, got into trouble largely because it ignored the processes of how people go about knowing.  On the other hand, the search for characterizations of how people make inferences has suffered from a failure to consider what it is that people are thinking about.  Surely this results because representation and process, while analytically separable, are intimately intertwined in any real instance of cognition (ibid:11).

Hutchins' interest in the interplay between representation and cognitive process (specifically, inference) leads him to a formal account of the negotiation of Trobriand land tenure agreements.  This account accommodated both the cultural content of what was being discussed and the (internal) mechanisms by which understanding in the discourse was accomplished.  Hutchins employed analytic constructs (schemata) which modelled the structured knowledge of participants observed in local court arguing their claims of land ownership.  In this same piece of work, Hutchins (1980) articulated a theory of culture which, following earlier proposals by Kay (1966) and D'Andrade et al. (1972), likened the relationship between cultural knowledge and social behavior to that posited to exist between language grammar and speech behavior.[2]  A cultural grammar, in this theory, is a system of knowledge which (due to structural organization and generativity) is capable of both internally representing courses of action and making available to natives the means for operating upon those representations.[3]  The analytic use of "schemata"—the formalization of cultural grammars—which followed from Hutchins' description of Trobriand land litigation continues to be the primary theoretical and methodological tool of cognitive anthropology (cf. Casson, 1983; D'Andrade, 1990; D'Andrade & Strauss, 1992; Holland & Quinn, 1987).

There are many reasons, aside from the productivity of this paradigm during the 1980's, for the primary position held by "schema theory" in cognitive anthropology today.  One factor contributing to this state of affairs has been the dominant theory of language since 1960—generative grammar—and the influence it has had upon anthropology, linguistics, psychology, philosophy, computer science, and cognitive science.  As a result of this influence, cognitive theory during the past three decades has largely been focused upon abstract mental concepts as the contents of "deep structure," where explanatory principles of cognition are posited to reside.  Explanations of cognitive phenomena are seen to lie, in this theory, in the articulation of syntactic operations which act upon mental symbols or thought tokens—uninterpreted, abstract, atomic elements whose grounding (both material/biological and semantic/social) in a real world is of secondary (and, to varying extents, peripheral) concern (cf. Fodor, 1976; Fodor & Pylyshyn, 1988; Newell & Simon, 1990[1976]; Pylyshyn, 1980).

These scholars (for the sake of convenience, call them "classicists") all generally argue as follows: a) cognition is "systematic" (i.e., basic concepts and relations combine to yield complex expressions yet retain their independent constituent effects upon the cognitive system); b) cognition is "productive" (i.e., an infinite set of behaviors can be generated by a finite system); therefore, c) cognitive architectures manipulate context-independent symbols by means of "deep" syntactic rules—that is, logical operations insulated from the fuzziness of semantic interpretation and real world performance.  The digital computer, at the abstract level of symbol manipulation, has provided the prototype upon which this model of cognition has flourished (cf. Newell, 1980).  Of course the methodology of ethnography has forced a more "semantically grounded" version of this theory upon cognitive anthropologists than is true of other disciplines.  That is, the methodological requirement that ethnographies incorporate real world (as opposed to laboratory or introspective) data, together with an epistemological position which privileges the native's point of view as the object of study (as opposed to one which renders meaning transparent and unproblematic) have mitigated the influence classical cognitive theory has had in anthropology to an extent which is not true of philosophy, psychology, and artificial intelligence.

At the same time, cultural anthropology's own history suggests a bias towards abstract-level representations of social and historical phenomena which predisposed cognitive anthropology to partake in the classical paradigm.  Following a tradition that goes back at least as far as Durkheim's (1915) notion of "collective representations," most anthropological formulations of culture have entailed the idealized mental contents of people as the backbone of theory meant to bridge history, behavior, and social constitution.  These formulations, whether focused upon individuals or societies, were attempts to account for behavioral regularities—within a more or less bounded group of directly or indirectly interacting individuals—as a consequence of shared contents of mind (cf. Benedict, 1934; Kluckhohn, 1941; Kroeber, 1917; Linton, 1936; Sapir 1931, 1934).  The modern central dogma of cultural anthropology, "shared understandings," is a product of these accounts.[4]

Werner and Schoepfle (1987), citing the extent to which data and explicit formalisms in analysis have been left out of published ethnographies, attribute the central dogma to anthropology's traditional methodology of participant observation which, they say, "forced ethnographers to learn culture as any human being immersed in it would... who then wrote about their personal synthesis of what the natives' culture was about" (ibid:82).

While participant observation employed as an unexplicit methodology undoubtedly accounts for part of the central dogma, an unappreciated aspect of this issue lies in anthropology's explicit emphasis upon interviews which do not contain the data necessary to account for contextualized performance in cognition and social behavior.  As a result of (or even, as theoretical grounding for) this methodological commitment, cognition is equated with "knowledge," culture with "shared knowledge," and intracultural variation becomes a statistic of consensus rather than an aspect of the way cognition and culture work.  For instance, interviews have long been taken by cultural (including cognitive) anthropology to be the premiere "epistemological window" through which understanding of social life is to be attained (cf. Werner & Schoepfle, 1987, Chapter 1).  But interview data, results of social behavior in their own rights, are seldom explicitly contextualized for the performance properties entailed in the production of those data.  Instead, interview data (including the protocols of formal elicitation techniques) are usually analyzed for the context-free content and structure of the propositions expressed or derived.[5]  By imputing mental contents in one-to-one correspondence with these data (or the derived propositions) this kind of analysis has been unable to account for the dynamics among and relationships between mental structure, behavior, and the material, historical, and social contexts of human activity.

Current theory in cognitive anthropology, it is being suggested, is a result of both multi-disciplinary ("classical") formulations of cognition derived from a formal theory of language—a theory emphasizing the disembodied, uninterpreted, context-free, structural organization and syntactic manipulation of thought tokens or concepts at the level of deep structure—and the central dogma of cultural anthropology which equates "culture" (a sum of the defining characteristics of an historically situated group of people) with idealized mental structure, or "shared understandings."  For the current argument, it is most important simply to recognize that theory in cognitive anthropology has reduced the description of processes of interpretation to the description of knowledge structures with the consequence that many of the important contributors to acts of interpretation get left out.  There is little attention paid to situations or the contexts which structure interpretive acts; to the learning of appropriate behaviors or courses of action; to the ways history, institutions, materials, and other actors constrain and facilitate cognitive and social performance.  In short, an exclusive focus upon conceptual structure in cognitive theory has neglected many of the processes and structures involved in individuals interpreting and acting in the world and through such acts contributing to its contents, form, and meaning.[6]

Identifying grounds for a new theory of culture and cognition

The motivation for changing this state in the theory of cognitive anthropology is easily communicated by reviewing a simple experiment conducted by John Gatewood (1983).  Gatewood asked students in his class to list all of the trees they could name, and then to list all of the trees they could recognize.  He found that, on average, subjects could only recognize (according to their own introspections) 50% of the trees they could name.  Furthermore, he says, "informants may be unable to recognize a kind of tree in their lists, and yet they can arrange their labels into taxonomies, use them in descriptions of landscapes, and wield them in figurative language" (ibid:382)  Clearly, there is a large degree of "slippage" between knowledge of language and knowledge of trees (in terms of recognition ability).  The two kinds of knowing entail different cognitive phenomena which would clearly be conflated if one simply asked subjects to build a taxonomy of tree names and then went on to draw conclusions about what these taxonomies mean for the individuals or, worse yet, for the society as a whole.

Thus far, the result merely reproduces an earlier concern for the "psychological validity" of underconstrained analytic models of cognitive phenomena (cf. Burling, 1964; Wallace, 1965; D'Andrade, 1976; Holy & Stuchlik, 1983).  Of more interest, is Gatewood's discussion about why the observed slippage (in his experiment) may exist.  He says, "[a]n obvious proximate cause for the persistence of unrecognizable category labels is the prevalence of written language in modern society.  People learn names for things and even a considerable amount about them without ever encountering a concrete example" (ibid:384).

One way to take Gatewood's observation here is in terms of how people come to "know" things from performing in specific contexts with those things as resources for acting–in this case with trees and with tree labels, respectively.  Furthermore, institutions and value systems arise and evolve through individual performances which take place in these socio-historically constructed, and materially instantiated, situations.  Knowing about trees is not the same as knowing about "trees" because of the differences in embodied experiences and social practices which structure the properties of these two kinds of cognitive events.  The point is deceptively obvious, and there is a risk it might be taken to be trivial.  For example, a response might be, "this is just the difference between knowing about trees as a college student in anthropology would, and knowing about trees as a logger, or botanist, or gardener would—a difference all anthropologists are sensitive to in their elucidations of cultural diversity."  The goal is precisely to make the processes which lie behind this kind of difference more explicit, not simply as a phenomenon which appears across "cultures"—i.e., tautologically defined by the partition of possible mental contents into sub-groups of shared understandings—but one prevalent within all interacting human groups as a consequence of the nature of situated human activity.[7]

Finally, Gatewood finds it plausible to speculate about the origins of this kind of phenomenon (that is, the "slippage" between the two kinds of knowing in his experiment) in human evolution.  The notion here is that social organization and symbol manipulation have extended the capacity of a strictly genetic system to reproduce ordered heterogeneity.  That is, solutions to problems confronting goal-directed agents (many entailing matters of survival) are now distributed beyond the immediate access of any individual (Hutchins & Hazlehurst, 1991).  Accessing those solutions requires intermediaries (e.g., symbols, tools, and social organization) and expertise in the use of each of these.  These intermediaries, then, come to set up their own constraints upon the interpretive processes of individuals (that is, they mediate cognitive performance) and yet they only do so via the consequences of those processes (that is, the learning and institutionalization of these intermediaries in social practice).[8]

The problems for cognitive science posed by classicists have not gone away—the "systematic" aspects of human cognition remain in need of attention.  But research entailing both computational modelling and empirical study of language and cognition has shifted the focus of attention to more inclusive properties, making classical concerns appear selective and unrepresentative of cognitive phenomena.  For example, supposed "noise" in language behavior, once taken to be artifacts of production from deep structure by classical accounts—the residues of competence minus resource limitations—is now recognized to be a productive source of information for language processing.  Systematic analyses of language use reveal cognitive mastery of complex conventions for: turn-taking (Sacks et al., 1974); utterance repair (Jefferson, 1974); signaling appropriateness or meta-communicative frames for interpretation (Goodwin, 1981; Gumperz, 1982); employing gesture and shared visual representations to ground intersubjectivity (Goodwin, 1993; Hutchins & Palen, 1993; Ochs et al., n.d.); and the construal of meaning via both referential and syntactic construction (Lakoff, 1987; Lakoff & Johnson, 1980; Langacker, 1986, 1987; Ochs, 1988).  Clearly, the lesson of this research is that language performance (the embodied, communicative, and context-sensitive aspects of language) should not be partitioned off as an orthogonal domain of study, but rather must centrally contribute to the constraints that are placed upon theories of language and cognition.

Some of these implications have been productively explored in a new paradigm of cognitive modeling called (generally) "connectionism"  (Rumelhart et al., 1986; McClelland et al., 1986).  Building upon earlier information processing models of cognition (including schema theory), this research has provided a refreshing alternative to classical accounts of cognition.  Connectionist modeling has emphasized learning as a process involving the satisfaction of diverse constraints on performance.  Cognitive performance is seen in terms of computational processes that integrate representations at different levels internally, and from different sources externally.  Connectionist models implementing this theory have demonstrated the ability to account for a variety of cognitive and language phenomena (cf. Elman, 1990; McClelland & Rumelhart, 1981; Plunket & Marchman, 1991; Seidenberg & McClelland, 1989; St. John & McClelland, 1990).  These models are capable of abstracting relevant information from the structure of the environment.  This abstraction process yields internal coding of competing constraints that represent experience with these particular environments and which become internal resources for performing in these contexts and generalizing in novel situations.  By focusing on the process of abstracting environmental regularities in order to build internal structure which is a reliable predictor of (i.e., affords the organization of behavior using) external structure, this paradigm shows promise for effectively addressing many of the inadequacies of the dominant view in language and cognition discussed above.

An outline of a theory of distributed cognition

The organization of behavior in this re-envisioned cognitive system can be profitably viewed as some function of current and past behaviors, as well as the nature of media and processes (e.g., internal representations, social organization, and material artifacts) which modulate the flow of information to (and within) the system and actions taken by the system (cf. Hazlehurst, 1990; Hutchins, 1990, 1991(a), 1993).  In such a framework, the "systematic" cognition of individual behavior results from the use of symbols—themselves the material products of social and cognitive action—and other resources available in the environment for achieving individual and collective goals.  Cognition, then, is to be studied through descriptions of the relations that obtain and evolve between resources and constraints (both internal and external to the individual) in the course of acting.  Culture, here, is a term that characterizes those processes which determine the intergenerational stability (or change) in the constraints and resources by which individuals organize their behavior.  Social organization, here, emerges through the institutionalization (for example, the elaboration through explicit, salient, redundant, or exclusive coding and instantiating) of means for interaction between individuals which, in turn, functions to constrain individual action (cf. Barth, 1966).

Both culture and social structure, that is, are most profitably understood through articulation of those mechanisms of situated human activity which are essential to the organization of individuals' behaviors.  The broad outline of these mechanisms is best articulated via three claims about situated human activity.

(1) Humans are continually engaged in the ongoing process of acting (i.e., behaving intentionally, or with purpose as in a goal-driven manner) in structured environments at the same time that they are—through those actions—constructing those environments.

By structured environment I mean to identify the physical and material settings in which action takes place as well as the historical and social context of the activity.  That is,

(2) Structured environments include two different populations of (ordered, information-bearing) material elements:  artifactual structures, and natural structures.

By artifactual structure I mean ordered material elements (including behaviors) in the environment created by other social actors or their artifacts.  By natural structure I mean objects in the environment created by processes of nature.

"Structure" is a term which must be used be used with care.  It only makes sense to talk about "structure" in conjunction with an interpreter which (or who) acts on it.  The earth's magnetic field is "structure" only insofar as it creates differences which make a difference to some interpreter(s).  In this case, the natural structure is a force which can be registered on a device (a magnetic compass), thereby reliably indicating a consistent direction in two dimensional space (say, along the surface of the earth) which is of use to an interpreter trying to locate something in that space.  Surely aspects of the earth's magnetic field are "structure" to some interpreters and not to others, depending upon those interpreters' uses of the entailed properties of this structure.

Notice that the presence of natural structures in no way precludes the use or importance of artifactual structures (and vice versa).  Throughout human history natural structures have provided a rich domain upon which artifactual structures have been modelled in order to capture relevant properties of the natural world.  Sun dials, magnetic compasses, the formula "F=ma," and words uttered as simple referential nouns denoting aspects of the physical world are all examples of using artifactual structure to model natural structure.  Of course, artifactual structure is also regularly generated from processes involving other artifactual structure.  Thus "F=ma," is quite arguably mediated by many layers of artifactual structure (e.g., a theory of gravitation, algebraic notation, institutions for learning, etc.) which stand between an event in nature and someone knowing what to do with or employing this string of symbols.  In any case, regarding human interpreters,

(3) The organization of behavior is made possible by, and simultaneously yields, internal structures.

By internal structures I mean those resources internal to individuals which enable organized behavior in structured environments.  Internal structure comes into being—that is, becomes organized—via mechanisms which save information generated by embodied action in structured environments.  Internal structure results from both phylogenetic learning (evolution) and individual learning of the constraints imposed by structured environments on the processes entailed by embodied action.  That is, in a manner analogous to the way evolution has brought about genetically encoded (and ontogenetically developed) structures for effective behaviors in ecological niches of survival value, learning affords building of internal structures which enable coordinated action in socio-historically, and materially, structured environments.

The interpreters which act upon internal structure are taken to be physiological systems within the individual.  The interpreters at work here are engaged in processes as diverse as parsing sequences of acoustic signals, generating motor commands, and manipulating the relations which constitute organized experience.[9]  It may be the case that we have very impoverished data on the nature of many of these internal interpreters and how they interact.[10]  However, many of the important properties can be described by careful attention to the situations in which individuals act and which, when repeated often, in time lead to the deposit of residua—or saved solutions or working pieces of an activity cycle—which enable coordination between parts of the system (Hutchins, 1986).

We are also free to "probe" the depths of internal processes with, for example, psychological tests in order to try and see what these internal structures are like directly.  When this is done, however, attention must be paid to the ways our probes may be unable to "see" the internal workings without the supporting role of the structured environment from which internal structure was derived and in which it normally operates.[11]  Another way to state the epistemological dilemma here is, again, with regard to the need to observe interpreters at work in order to see the structures and processes involved.

Finally, notice that "action in structured environments" can entail a complex set of relationships between artifactual, natural, and internal structure.  Each may be engaged in mediation between the others via the imposition of constraints on the functional possibilities of the system.  (See Figures 1 and 58.)

This set of ideas about the ways interpretive processes function at the level of the individual (mediated by internal structure as a result of embodied experience with—and resulting in—the structure and processes of social practices and cultural histories), I call a theory of "distributed cognition."  The name and many of the ideas come from the work of, and with, Edwin Hutchins and colleagues (Flor & Hutchins, 1990; Hazlehurst, 1990; Hutchins 1990, 1991(a), 1991(b), 1993, 1995; Hutchins & Hazlehurst, 1991, 1995; Hutchins & Klausen, in press).  The theory entails an information processing account of the cognitive, material, social, and historical means for organizing behavior.  The term "distributed cognition" comes from an explicit recognition of the way interpretive acts are carried out in structured environments—shaped by structure, both internal and external to actors, as well as the processes which bring structures into interaction—resulting in changes to both internal and external structures and the processes which involve them.  The theory requires moving the traditional boundaries of cognition outward from the individual, without losing sight of the fact that there is still one kind of boundary at the skin of the individual and others "inside," which have their own psychological and information processing properties.  Cognition, in this sense, is distributed across the participants and mediating objects of socio-historically constructed situations and develops in real time—both the micro time of situated performers' actions, and the macro time of cultural history.

The methodology employs empirical methods meant to answer the question, "what are the differences that make a difference for individuals and groups of interacting individuals located and acting in (real and historical) time and (material/physical and social) space?"  The specific methods for doing this are various, and a sample of them are employed in the chapters which follow.  In general, all of these methods attempt to articulate the relations between—and processes entailing—cognitive resources (material structures both internal and external to actors) employed in socio-historically situated activity, and how these evolve in time.

Chapter 2 attempts to bring to bear diverse sources of information about social activities now and in the past in order to portray "fisherman identity," the means by which accounts of experience are organized and rendered by fishermen from this community.  The data sources of this chapter include: story-telling in the harbor and in interviews, historical accounts of the island and fishermen's activities, the nature of local institutions and their influence on attitudes and activities, materials and behaviors which were observed to signify or make salient to actors and others who they are, what they do, what they have done, where they come from, and where they plan to be in the future.

The analysis of Chapter 2 makes it clear that the organization of experience for fishermen from this community is constructed through participation in a practice which is (and has been) shaped by a wide array of structures and processes.  These are not abstract entities but, rather, are parts of the lived-in world—they include the ships fishermen design and spend many hours aboard, the containers into which fish get packed and which provide the units for measuring success, the behavioral patterns of fish which determine where and how fishermen spend their time, the linguistic formulations by which fishermen communicate on the radio and in person, the conventions for telling stories at the harbor, and the means for getting along with other residents on the island.  Casting these phenomena (as embodied experiences of individuals) in terms of distributed cognition—entailing structures whose interactions have functional consequences for social and cognitive actors—provides an insightful characterization of "identity" as those structures internal to individuals which participate in this interaction and which make possible and coherent both the subjective and objective properties of social life.

In Chapter 3, the analysis turns its attention to a structural and historical description of one specific fishing practice, pair-trawling for herring.  In this chapter, the history of practice is grounded in a structural account of the actual tasks required of fishermen—as cognitive performers and social actors—for accomplishing their objectives.  These tasks—that is, the articulation of constraints upon, and resources for, the accomplishment of work routines as they are now and have been in the past—provide the lense through which the significance of ecological, economic and political conditions, technology, and social organization is made explicit.

In Chapter 4, the investigation of pair-trawling practice continues with a "micro analysis"—a reconstruction of part of one night's fishing by a team of trawlers.  On board audio and video recording of events are translated into task-relevant representations which reveal the ways resources—internal and external structures and the processes which incorporate these—are employed by actors to solve problems and accomplish goals in the live performance of their activities.  In particular, a transcript of sequential events recorded during the team's search for where to set their trawl is employed to investigate properties of both the individual and collective organization of behavior. 

The analysis of Chapter 4 makes it clear that processes of reasoning employed by the driving captains are very dynamic, opportunistic, and heavily dependent upon the contexts and technology which structure the activity.  In their efforts to keep track of where other boats are (in order to make sense of what other teams are "seeing" for fish) much of the activity centers upon disambiguating which boat is where as reports come over the radio and echoes are displayed on radar devices local to each boat of the team.  These information gathering devices and strategies—the generators of external representations which, in coordination with internal structures, enable reasoning processes—are seen as cognitive resources which are essential to the nature of the activity.  These external structures, and the internal structures built through experience using them, thus mediate the ongoing activity at the same time actions taken which employ these resources reproduce the practice itself.

Finally, Chapter 5 presents a more formal account of some relations between internal and external structures which are employed by fishermen in their practice.  In this chapter, the results of an experiment are reported, in which fishermen's understandings and linguistic productions regarding sonar representations of fish are compared, both within and across individuals and fishing practices.  Sonar devices are the primary vehicles for "seeing" fish underneath the surface of the water, and the representations these devices create are very important components of all techniques employed for catching herring.  These techniques vary, however, with respect to the characteristics understood to define potentially "good" sightings of fish, and the means, necessity, and scope of sharing of this information with others while fishing.

Subjects in the experiment—representing members of different teams who engage in different fishing practices (which engender different experiences with each other and with the representations which mediate their respective activities)—should, the theory holds, reflect these differences in their task performance.  In the terms of distributed cognition, internal structures are generated through the retention of information (the residua of coordinating actions) brought on by patterns of embodied experiences—in this case, performances employing visual and linguistic representations of sonar displays of fish for accomplishing goals.  Task performance then, to the extent the experimental tasks simulate the real problems of interpreting sonar representations and communicating about them, should reflect those patterns of shared activity which are defined by the boundaries of the practice—who fishes with whom, and what peculiar properties of interpreting sonar images are entailed by specific fishing techniques.  The findings are consistent with this prediction.  The experiment also provides specific evidence for the proposition that the linguistic representations employed to discuss and communicate about sonar displays of fish are essential resources for reasoning about those sonar representations.


 



[1]In each of the last two clauses, one wants to write "culture" (as a noun) in place of "group."  It seems to me this is a symptom of thinking in this tradition where the "buzzing booming confusion" of social life is crystallized into discrete, synchronic descriptions of mental contents, which lose their historical import and their grounding in the world of material and social action (cf. Boyer, 1990; Holy & Stuchlik, 1983).

[2]Interest in the analogy between culture and language grammar was quite widespread throughout the 1960's and 1970's (cf. Goodenough, 1957; Burling, 1969; Spradley, 1972).

[3]Thus, for example, inferences (including extended or "chained" reasoning) are possible due to the relations which obtain between basic elements of structured knowledge.

[4]Although recent work has suggested formal validity for the central dogma on statistical grounds (cf. Boster, 1985; Romney et al, 1986; D'Andrade, 1990), there remains a need to establish what effects variability in cognitive resources afford cognitive, cultural, and social processes.

[5]Holy and Stuchlik (1983) make a similar argument about anthropology's misuse of the problematic between knowing and acting.  They also attempt, as others have before them (for example Barth, 1966), to formulate a solution via processes which construct social phenomena through individual action.  However, Holy and Stuchlik fail to make problematic the organization of behavior at the level of the individual.  As a result, all behavior of interest is taken to be social and all knowledge of interest is taken to be intersubjectively shared and (by definition) cultural.  In failing to articulate a theory of cognition, they are left with only logical (rather than empirical) mechanisms to do the work of bridging knowing and acting.

[6]The reader will notice, in what follows, a distinction between this critique of cognitive anthropology and other critiques which start from similar observations but end up with a theory of anthropologists as interpreters.  I do not believe the current critique is inconsistent with these other formulations, in the sense that both are responding to similar failures in the established paradigm.  Where they end up, however, is another matter.  My own view is that a theory of anthropologists as interpreters must entail a theory of interpretation as a cognitive phenomenon which applies to natives and anthropologists, and so the latter is a more inclusive, epistemologically sound, and useful theory.

[7]The point is, anthropology's traditional attribution of significance to "cultures" as the object of study, has the longstanding (and still lingering) effect of deflecting our attention from addressing the more general problem of "the ways humans organize their behavior."  It should be clear that this is not a call for anthropologists to abandon their sensitivities to the ways people of different histories organize experience in different ways.  On the contrary, it is a call to recognize our own framing of the problems of describing human behavior in terms of our own history of practice—a practice were the exotica of foreign peoples were best communicated in terms of the disembodied propositions one could learn, bring home, and conveniently arrange in a monograph for publication.

[8]Neither the author, nor Gatewood, are the first to make this line of argument, which has resurfaced under the banner of "NeoVygotskian theory."  Recent publications (or translations) of relevance here include Goody (1977),Scribner and Cole (1981), Rogoff (1990), Vygotsky (1962, 1978), Wertsch (1985).

[9]See Dennett (1992) for an account of human consciousness built out of many such interacting, yet independent, processes.

[10]In the context of these theoretical constructs, cognitive psychology has an important role to play and provides us with many insights into the nature of the internal interpreters.

[11]More precisely, psychological experiments will substitute a new structured environment which probably only partially overlaps with the structured environment from which the target internal structure was derived.