Cognition, Distributed Cognition, Distributed 1. Mind in Society For many people, distributed cognition means cog- Like all other branches of cognitive science, distributed nitive processes that are distributed across the mem- cognition seeks to understand the organization of bers of a social group (Salomon 1993). The fun- cognitive systems. Like most of cognitive science, it damental question here is how the cognitive processes takes cognitive processes to be those that are involved we normally associate with an individual mind can be in memory, decision making, inference, reasoning, implemented in a group of individuals? A wide range learning, and so on. Also following mainstream of disciplines in the social sciences has explored this cognitive science, it characterizes cognitive processes question. in terms of the propagation and transformation of Treating memory as a socially distributed cognitive representations. function has a long history in sociology and anthro- What distinguishes distributed cognition from other pology. Durkheim, and his students, especially approaches is the commitment to two related theo- Halbwachs (1925), maintained that memory could not retical principles. The first concerns the boundaries of even be coherently discussed as a property of an the unit of analysis for cognition. While boundaries isolated individual. Roberts (1964) proposed that are often a matter of tradition in a field, there are some social organization could be read as a sort of archi- general rules one can follow. Bateson (1972) says one tecture of cognition at the community level. He should bound the unit so that things are not left characterized the cognitive properties of a society (its inexplicable. This usually means putting boundaries memory capacity and ability to manage and retrieve on units where the traffic is low. The second principle information) by looking at what information there is, concerns the range of mechanisms that may be where it is located, and how it can move in a society. assumed to participate in cognitive processes. While Schwartz (1978) proposed a distributional model of mainstream cognitive science looks for cognitive culture that emphasized the distribution of beliefs events in the manipulation of symbols (Newell et al. across the members of a society. Romney et al. (1986) 1989), or more recently, patterns of activation across created quantitative models of the patterns of cultural arrays of processing units (Rumelhart et al. 1986, consensus. The identification of patterns raised the McClelland et al. 1986) inside individual actors, question of why such patterns form. Sperber (1985) distributed cognition looks for a broader class of introduced the idea of an epidemiology of representa- cognitive events and does not expect all such events to tions. He suggested an analogy in which anthropology be encompassed by the skin or skull of an individual. is to psychology as epidemiology is to pathology. In When one applies these principles to the observation the same way that epidemiology addresses the dis- of human activity ‘in the wild,’ at least three interesting tribution of pathogens in a population, anthropology kinds of distribution of cognitive process become should treat questions about the distribution of apparent: cognitive processes may be distributed representations in a community. A similar set of across the members of a social group, cognitive developments followed from Dawkins’ (1976) dis- processes may be distributed in the sense that the cussion of ‘memes’ as the cultural analog of genes. operation of the cognitive system involves coordi- These ideas have now coalesced in the field of memetics nation between internal and external (material or (Blackmore 1999). March and Simon (1958) argued environmental) structure, and processes may be distri- that organizations can be understood as adaptive buted through time in such a way that the products of cognitive systems. Juries are an important class of earlier events can transform the nature of later events. distributed problem solving organization and they The effects of these kinds of distribution of process are have been intensely studied by social psychologists extremely important to an understanding of human (Hastie 1993). Of course, in social psychology there is cognition. a vast literature on small-group decision making, some The roots of distributed cognition are deep, but the of which discusses the properties of aggregates. field came into being under its current name in the Scientific communities have received special atten- mid-1980s. In 1978, Vygotsky’s Mind in Society was tion because the work of science is fundamentally published in English. Minsky published his Society cognitive and distributed. The phenomena that have of Mind in 1985. At the same time, Parallel Distri- been explored include how the organization of com- buted Processing was making a comeback as a munication media in a scientific community affect the model of cognition (Rumelhart et al. 1986). The kinds of things the community can learn (Thagard nearly perfect mirror symmetry of the titles of 1993), how conditions external to the individual Vygotsky’s and Minsky’s books suggests that some- scientists can affect their individual choices in ways thing special might be happening in systems of that lead to different high-level structures to emerge distributed processing, whether the processors are (Kitcher 1990), how the distribution of cognitive neurons, connectionist nodes, areas of a brain, whole activity within social networks and between people persons, groups of persons, or groups of groups of and inscriptions accounts for much of the work of persons. science (Latour 1987), and how scientific facts are 2068 Cognition, Distributed created by communities in a process that simply could of similar societies of mind. This means, of course, that not be fit into the mind of an individual (Fleck 1935). both what’s in the mind, and what the mind is in are Economists have been interested in the tension societies. Getting internal agencies into coordination between what is individually rational and what is with external structure can provide the organization of rational at the aggregate level. This theme has been the relations between the internal agencies that is explored in game theory under the rubric of the required to perform the new functional skill. Prisoner’s Dilemma, the paradox of the commons, Vygotsky developed this idea of the social origins of and other cases where individual rationality and group individual psychological functions in Society of Mind rationality diverge (Von Neumann and Morgenstern (Vygotsky 1978, Wertsch 1985). Vygotsky argued that 1964). every high-level cognitive function appears twice: first Anthropologists and sociologists studying know- as an interpsychological process and only later as an ledge and memory, social psychologists studying intrapsychological process. The new functional system small-group problem solving and jury decision inside the child is brought into existence in the making, organizational scientists studying organiz- interaction of the child with others (typically adults) ational learning, philosophers of science studying and with artifacts. As a consequence of the experience discovery processes, and economists and political of interactions with others, the child eventually may scientists exploring the relations of individual and become able to create the functional system in the group rationality, all have taken stances that lead absence of the others. This could be seen in Minsky’s them to a consideration of the cognitive properties of terms as a mechanism for the propagation of a societies of individuals. There is ample evidence that functional skill from one society of mind to another. the cognitive properties of a group can differ from the From the perspective of distributed cognition, this sort cognitive properties of the members of the group. of individual learning is seen as the propagation of a particular sort of pattern through a community. Cultural practices assemble agencies into working assemblages and put the assemblages to work. Some 2. The Society of Mind of these assemblages may be entirely contained in an individual, and some may span several individuals and The work described above looks for mind-like proper- material artifacts. The patterns of activity that are ties in social groups. This is the Mind in Society repeatedly created in cultural practices may lead to the reading. The metaphor can be run the other way as is consolidation of functional assemblages, the atrophy done in Minsky’s Society of Mind (1985). Rather than of agencies that are rarely used, and the hypertrophy using the language of mind to describe what is of agencies that are frequently employed. The result happening in a social group, the language of social can be individual learning or organizational learning, groups can be used to describe what is happening in a or both. mind. Minsky argued that to explain intelligence we need 2.1 Interaction as a Source of Noel Structure to consider a large system of experts or agencies that can be assembled together in various configurations to An important property of aggregate systems is that get things done. Minsky also allowed that a high-level they may give rise to forms of organization that agency itself could be composed of low-level agencies. cannot develop in the component parts. Freyd (1983) With Papert (Minsky and Papert 1988), he argued that argued that some of the features of language that are the low-level agencies (the ones that take on ‘toy-sized identified as linguistic universals could arise out of the problems’) could be implemented as distributed com- necessity of sharing the linguistic code. For instance, putations in connectionist nets. Minsky said, ‘… each the reason that linguistic categories tend to approxi- brain contains hundreds of different types of machines, mate discrete structures may have little to do with the interconnected in specific ways which predestine that organization of the brain, and everything to do with brain to become a large, diverse society of partially the problem of pushing a complex representation specialized agencies’ (1988).