Modularity versus Interactive Processing Research Paper

The model of interaction that has framed research on modularity is a reduction of the model generally assumed by interactionists. In the initial modularity debate (Piattelli-Palmarini 1980), interactionism was decoded as a ‘sensor-motor interaction.’ The other part of interactionism is that the study of cognition irreducibly depends on an analysis of a social interaction (e.g., Vygotsky 1978, Mead 1934). The ongoing social interaction within a community of actors, the continuous joint sensemaking of the world, also produces structure. To what degree the function of mind depends on the structures produced by specialpurpose neurologically hardwired ‘modules,’ on structures created as a part of a physical and developmental interaction with the world, or on those structures that are produced by means of a social interaction, is an open research question.

1. Modularity

Structuralism developed the idea that the structure of mind (i.e., cognitive structure) is the result of evolutionary and genetically determined biological forces and that the products of human effort reflect the biological basis of structure. An issue for structuralism is to identify where the biology of structure ends and structures created by humankind begin: To what degree is the structure of language, behavior, and society determined by the biological? To what degree is the structure a product of human endeavor? Theories of modularity tie the structural elements of language (Chomsky 1980) and thought (Fodor 1983) to special purpose, neurologically hardwired, faculties of mind. An important issue is to determine to what degree modules result from a gradual constructive and developmental interaction within specific domains of behavior (as opposed to being entirely prespecified and innately determined).

In a debate with Piaget (Piattelli-Palmarini 1980), Chomsky (1980) developed modularity as an innateness argument for a language faculty inherent in the mind. A grammar is a system that specifies the phonetic, syntactic, and semantic properties of an infinite class of potential sentences. Chomsky’s position was that there was a special purpose genetically determined language faculty—a mental organ—that specifies a certain class of ‘humanly accessible grammars.’ The language module is comparable to other organs in the body, e.g., the heart or the liver. It has both general properties that exist across the species (as is the case for any other organ, e.g., the heart) yet has unique characteristics within the individual. It has a place in a system of (mental) structures, and its development within the course of the individual’s lifetime has a basis that is genetically determined.

Chomsky’s chief argument for the existence of such a language module concerned the learnibility of language. ‘Within a given speech-community, children with varying experience acquire comparable grammars, vastly underdetermined by the available evidence’ (Chomsky 1980, p. 35). The set of initial assumptions concerning the nature of language—i.e., the grammar—the child brings to language learning is the genetically determined knowledge of the language module. The structure that is attributed to the language faculty ‘must be sufficiently rich and specific to account for the attainment of linguistic competence in particular languages, but it must not be so rich and specific that it excludes attested cases’ (p. 49).

Fodor (1983) developed the idea of modularity in terms of input modules. An input module produces structure rather than having innate knowledge about the structure of (i.e., grammar for) language. Input modules convert transduced sensory (visual or acoustic) input to a logical form over which the central processors work. ‘Whereas transducer outputs are most naturally interpreted as specifying the distribution of stimulations at the ‘surfaces’ (as it were) of the organism, the input systems deliver representations that are most naturally interpreted as characterizing the arrangement of things in the world ’ (p. 42). An input module is self-contained, informationally encapsulated, and computationally autonomous. It is a special purpose faculty of mind that is neurologically hardwired, genetically determined, and evolutionarily developed. Its operation is mandatory, fast, and domain specific, and it produces ‘shallow’ outputs in a logical form.

Karmiloff-Smith (1992) combines Piaget’s constructivism with Fodor-style modularity of the mind. One key idea is a distinction between the prespecified modules and modularization as a process. Karmiloff-Smith contends that the mind may become modularized as development proceeds; moreover, cognitive development can be domain specific without being strictly modular. The initial hardwiring of an infant is not a genetic blueprint of input module, but rather a domain-specific predisposition, an initial bias, with regard to certain kinds of input. ‘The view that I adopt throughout the book is that Nature specifies initial biases or predispositions that channel attention to relevant environmental inputs, which in turn affect subsequent brain development’ (p. 5). The language of thought (its structure) results from a process of representational redescription (RR). When first acquired, knowledge is constrained by initial biases and predispositions, and it is implicitly represented. By a process of developmental redescription, that knowledge is gradually transformed into representations that are ‘progressively more manipulable and flexible’ and ultimately result in the ‘emergence of conscious access to knowledge’ and explanatory behavior.

2. Contrasting Modularity and Interactionism

A significant part of interactionist accounts of cognition concerns the sensor-motor interactions of the individual with her environment; see, for example, Brooks (1991). However, the other part of interactionism is that the study of cognition irreducibly depends on an analysis of a social interaction. A crucial idea is that structure is created as a product of a social interaction, and is modified and accumulates within a community of actors across generations. Thus special purpose, hardwired, evolutionarily designed, input modules can be seen as producing one set of structures and social interaction and cultural history another set of structures (Cole 1996, p. 198–214).

Modularity accounts assume, as a basic unit of analysis, a reduction of mind to what goes on in the head; for example, language can be analyzed independent of communication. Interactionist accounts assume interaction, especially social interaction, as the basic unit of analysis. ‘Mind arises through communication by a conversation of gestures in a social process or context of experience—not communication through mind’ (Mead 1934, p. 50). The sources (roots) of intelligence do not only move upwards, from the biological, but also downwards, from social to cognitive (Vygotsky 1978). The study of cognition must be anchored into larger units of analysis than the individual; units that include the sociotechnical (Hutchins 1995a). Cognition is inextricably tied to activity (Suchman 1987); it emerges from an interaction (Agre 1997). It includes the context, and not just its internalization (Lave 1988).

A second point of contrast concerns the difference between structure as biologically determined and external structures that emerge as a product of human activity. The focus of interactionist accounts is best exemplified by the analyses of the structure of behavior. For an interactionist account, much of the structure of behavior emerges from an interaction, e.g., Garfinkel (1967), Suchman (1987), and Agre (1997). Context, knowledge, and ‘seeing’ depend on interaction with structures in the world (Lave 1988, Hutchins 1995a, 1995b, Goodwin and Goodwin 1996). The final contrasting point concerns the historical aspects of cognition. Modularity proponents ground cognition in genetics and physical evolution, and interactionists focus on the aspects of cognition that are the product of a social, cultural, and historic interaction within a community of actors. The full ambit of activities is culturally and historically conditioned, ranging from simple activities like greeting a friend on the telephone to complex cognitive performances like reading a newspaper to physical activities like skiing. In each of these cases, cognitive performance is tied to the performance of this activity by prior generations of individuals within the community. Acquisition of the structure of a given activity depends on the individual internalizing a structure for the activity from somebody else who already knows how to perform the activity (Vygotsky 1978). Examples of models of cultural learning that depend on interaction include Hutchins (1995a), Lave and Wenger (1991), Alterman et al. (1998), and Tomasello et al. (1993). These differences in assumption lead to some critical differences in viewpoint. Consider the case of language learning, which is critical to Chomsky’s (1980) argument for a language module. Chomsky uniformly dismisses the ‘performance’ aspects of language learning, arguing that children acquire the grammar for a language even though it is ‘vastly undetermined by the available evidence.’ For interactionists, semantics, reference, and conversation are all jointly engineered among participants in a social interaction (see, e.g., Clark 1996, Hanks 1996, Clark and Wilkes-Gibbs 1986, Sacks et al. 1974). Whether language and its structure can be reduced to an analysis independent of semantics and the social interaction in which language (and language learning) occurs is a point of debate.

With regard to Fodor’s version of modularity, the Structure as biologically determined External Structure issue pivots over the difference between a ‘language of thought’ and the structure of thinking. In Fodor’s view, central processes work from internal representations. An input module produces a logical form, which is the language of thought. Contrast that to Vygotsky’s view that thought, memory, and planning all begin with a social interaction. The structure of thinking depends on the history of such an activity within a community of actors. Consider the example of literacy. The reasoning processes of literacy are dependent on internalization of the structure of the ‘reading’ activity. To become literate is to tap into the culture of the activity of ‘reading’ within a given community (Scribner and Cole 1978). How this practice emerges may differ between communities. Additionally, many interactionists argue against the notion of internal representations of the sort a ‘language of thought’ argument supports; see a special issue of Cognitive Science (1993) for a debate on this issue.

The Karmiloff-Smith (1992) model has interactionist elements, but it does not account for the social elements of the representational structures acquired during the child’s development. Consider a task like learning to play the piano. It is true that part of ‘learning to play the piano’ depends on sensor-motor interactions. But it is also true that it depends on the social interaction with a teacher and the representations provided by the musical score, which is the product of a social interaction. In other words, the child’s participation in an ongoing social interaction within a community of actors is a significant factor in the development of high-level representations of the sort that allow for conscious access to knowledge and explanation. Even the more flexible access to knowledge that would be required, say, by jazz improvisation on the piano, is dependent on structures created by a social interaction; e.g., compare bebop to modal forms of jazz improvisation.

Bibliography:

1. Agre P 1997 Computation and Human Experience. Cambridge University Press, Cambridge, UK
2. Alterman R, Zito-Wolf R, Carpenter T 1998 Pragmatic action. Cogniti e Science 22(1): 53–105
3. Brooks R 1991 Intelligence without representation. Artificial Intelligence 47: 139–59
4. Chomsky N 1980 On cognitive structures and their development. In: Piattelli-Palmarini M (ed.) Learning and Language. Harvard University Press, Cambridge, MA, pp. 35–52
5. Clark H H 1996 Using Language. Cambridge University Press, New York
6. Clark H, Wilkes-Gibbs D 1986 Referring as a collaborative process. Cognition 22: 1–39
7. Cognitive Science 1993 Situated action. Cogniti e Science 17: 87–116
8. Cole M 1996 Cultural Psychology. Harvard University Press, Cambridge, MA
9. Fodor J 1983 The Modularity of Mind. MIT Press, Cambridge, MA
10. Garfinkel H 1967 Studies in Ethnomethodology. Polity, Cambridge, UK
11. Goodwin C, Goodwin M H 1996 Seeing is a situated activity: Formulating planes. In: Engestrom Y, Middleton D (eds.) Cognition and Communication at Work. Cambridge University Press, Cambridge, UK, pp. 61–95
12. Hanks W F 1996 Language and Communicative Practices. Westview Press, Boulder, CO
13. Hutchins E 1995a Cognition in the Wild. MIT Press, Cambridge, MA
14. Hutchins E 1995b How a cockpit remembers its speed. Cognitive Science 19: 265–88
15. Karmiloff-Smith 1992 Beyond Modularity. MIT Press, Cambridge, MA
16. Lave J 1988 Cognition and Practice. Cambridge University Press, Cambridge, UK
17. Lave J, Wenger E 1991 Situated Learning: Legitimate Peripheral Participation. Cambridge University Press, Cambridge, UK
18. Mead G H 1934 Mind, Self & Society. University of Chicago Press, Chicago
19. Piattelli-Pamarini M (ed.) 1980 Language and Learning: The Debate Between Jean Piaget and Noam Chomsky. Harvard University Press, Cambridge, MA
20. Sacks H, Schegloff E, Jefferson G 1974 A simplest systematics for the organization of turn-taking in conversation. Language 50: 696–735
21. Scribner S, Cole M 1978 The Psychology of Literacy. Harvard University Press, Cambridge, MA
22. Suchman L A 1987 Plans and Situated Actions. Cambridge University Press, Cambridge, UK
23. Tomasello M, Kruger A C, Ratner H R 1993 Cultural learning. Beha ioral and Brain Sciences 16: 495–552
24. Vygotsky L S 1978 Mind in Society. Harvard University Press, Cambridge, MA
25. Vygotsky L S 1986 Thought and Language. In: Kozulin A (ed.) MIT Press, Cambridge, MA