Copyright © 1999 Massachusetts Institute of Technology
Two influential theoretical positions have permeated cognitive science: (1) that the mind/brain is a general-purpose problem solver (NEWELL and Simon 1972; PIAGET 1971); and (2) that it is made up of special-purpose modules (Chomsky 1980; Fodor 1983; Gardner 1985). The concept of modular organization dates back to KANT (1781/1953) and to Gall's faculty theory (see Hollander 1920). But it was the publication of Fodor's Modularity of Mind (1983) that set the stage for recent modularity theorizing and which provided a precise set of criteria about what constitutes a module.
Fodor holds that the mind is made up of genetically specified, independently functioning modules. Information from the external environment passes first through a system of sensory transducers that transform the data into formats each special-purpose module can process. Each module, in turn, outputs data in a common format suitable for central, domain-general processing. The modules are deemed to be hardwired (not assembled from more primitive processes), of fixed neural architecture (specified genetically), domain-specific (a module computes a constrained class of specific inputs bottom-up, focusing on entities relevant only to its particular processing capacities), fast, autonomous, mandatory (a module's processing is set in motion whenever relevant data present themselves), automatic, stimulus-driven, and insensitive to central cognitive goals. A further characteristic of modules is that they are informationally encapsulated. In other words, other parts of the mind can neither influence nor have access to the internal workings of a module, only to its outputs. Modules only have access to information from stages of processing at lower levels, not from top-down processes. Take, for example, the Muller-Lyer illusion, where, even if a subject explicitly knows that two lines are of equal length, the perceptual system cannot see them as equal. Explicit knowledge about equal line length, available in what Fodor calls the "central system," cannot infiltrate the perceptual system's automatic, mandatory computation of relative lengths.
For Fodor, it is the co-occurrence of all the properties discussed above that defines a module. Alone, particular properties do not necessarily entail modularity. For instance, automatic, rapid processing can also take place outside input systems such as in skill learning (Anderson 1980). Task-specific EXPERTISE should not be confounded with the Fodorian concept of a module. Rather, each module is like a special-purpose computer with a proprietary database. A Fodorian module can only process certain types of data; it automatically ignores other, potentially competing input. This enhances automaticity and speed of computation by ensuring that the organism is insensitive to many potential classes of information from other input systems and to top-down expectations from central processing. In other words, Fodor divides the mind/brain into two very different parts: innately specified modules and the nonmodular central processes responsible for deductive reasoning and the like.
Fodor's modularity theory had a strong impact on researchers in cognitive development. Until the 1980s BEHAVIORISM and Piaget's constructivism had been dominant forces in development. Both these theories maintain that the infant and child learn about all domains -- SYNTAX, SEMANTICS, number, space, THEORY OF MIND, physics, and so forth -- via a single set of domain-general mechanisms (the actual types of mechanism invoked are very different in the two theories). By contrast, with Chomskyan linguistics and Fodorian modularity, a sizable number of developmentalists opted for an innately specified, modular view of the infant mind. Not only did Chomskyan psycholinguists argue for the innately specified modularity of syntax (e.g., Smith and Tsimpli 1995; see Garfield 1987; but see also Marslen-Wilson and Tyler 1987 for a different view), but developmentalists also supported a modular view of semantics (Pinker 1994), of theory of mind (Anderson 1992; Baron-Cohen 1995; Leslie 1988), of certain aspects of the infant's knowledge of physics (Spelke et al. 1992; but see Baillargéon 1994 for a different view), and of number in the form of a set of special-purpose, number-relevant principles (Gelman and Gallistel 1978).
Data from normal adults whose brains become damaged from stroke or accident seem to support the modular view (Butterworth, Cipolotti, and Warrington 1996; Caramazza, Berndt, and Basili 1983). Indeed, brain-damaged adults often display dissociations where, say, face processing is impaired, while other aspects of visual-spatial processing are spared, or where semantics is spared in the face of impaired syntax, and so forth. On the other hand, several authors have now challenged these seemingly clear-cut dissociations, demonstrating, for instance, that supposedly damaged syntax can turn out to be intact if one uses on-line tasks tapping automatic processes rather than off-line, metalinguistic tasks (e.g., Tyler 1992), and that a single underlying deficit can give rise to behavioral dissociations (Farah and McClelland 1991; Plaut 1995) .
Evidence from idiots savants (Smith and Tsimpli 1995) and from persons having certain developmental disorders (e.g., Baron-Cohen 1995; Leslie 1988; Pinker 1994) has also been used to lend support to the modularity view. There are, for instance, developmental disorders where theory of mind is impaired in otherwise high functioning people with AUTISM (Frith 1989), or where face processing scores are in the normal range but visuo-spatial cognition is seriously impaired, as in the case of people with Williams syndrome (Bellugi, Wang and Jernigan 1994). These data have led some theorists to claim that such modules must be innately specified because they are left intact or impaired in genetic disorders of development. Yet this claim has also been recently challenged. In almost every case of islets of so-called intact modular functioning, serious impairments within the "intact" domain have subsequently been identified (e.g., Karmiloff-Smith 1998; Karmiloff-Smith et al. 1997), and in cases of purported singular modular deficits, more general impairments have frequently been brought to light (e.g., Bishop 1997; Frith 1989; Pennington and Welsh 1995). In other words, abnormal development does not point to isolated, prespecified modules divorced from the rest of the cognitive, motor, and emotional systems. Genetic impairments affect various aspects of the developmental process, in some domains very subtly and in others more seriously.
In normal development, too, new research is also pointing to gradual specialization rather than prespecification. Take the case of syntax, a particularly popular domain for claimants of modularity. Brain imaging studies of infants and toddlers have shown a changing pattern of HEMISPHERIC SPECIALIZATION (Mills, Coffey-Corina, and Neville 1993, 1994). Initially, the infant processes syntax in various parts of the brain across both hemispheres. It is only with time that parts of the left hemisphere become increasingly specialized. This also obtains for other aspects of language and for face processing in which infant imaging studies using high-density ERPs show progressive localization and specialization (Johnson 1997). The human cortex takes time to structure itself as a function of complex interactions at multiple levels: differential timing of the development of parts of cortex, the predispositions each part has for different types of computation, and the structure of the inputs it receives (for detailed discussion, see for example Elman et al. 1996; Johnson 1997; Quartz and Sejnowsky forthcoming). While there may be prespecification at the cellular level, this does not seem to hold for synaptogenesis at the cortical level. Specialized circuitry and the rich network of connections between cells appear to develop as a function of experience, which challenges the notion of prespecified modules.
Although the fully developed adult brain may include a number of module-like structures, it does not follow that these must be innately specified. Given the lengthy period of human postnatal brain development and what we know about the necessary and complex interaction of the genome with environmental influences (e.g., Elman et al. 1996; Johnson, 1997; Quartz and Sejnowsky 1997; Rose 1997), modules could be the product in adulthood of a gradual developmental process (Karmiloff-Smith 1992), rather than being fully prespecified, as Fodorians maintain. This is not a return to a general-purpose, equipotential view of the infant brain. On the contrary, an alternative to representational nativism (the innate knowledge position on which modularity theory is based) has been proposed by several theorists who have formulated hypotheses about what might be innately specified in terms of computational and timing constraints, while leaving ample room for epigenetic processes (Elman et al. 1996; Quartz and Sejnowsky 1997).
While the concept of prespecified modules has been challenged on a number of fronts, it has also become increasingly clear that the general-purpose view of the brain is inadequate. The human mind/brain is not a single, domain-general processing system, either in infancy or in adulthood. Nor is the alternative a return to simple behaviorism. The genome and sociophysical environment both place constraints on development. A different way to conceive of modularity might therefore be to adopt a truly developmental perspective and acknowledge that the structure of minds could emerge from dynamically developing brains, whether normal or abnormal, in interaction with the environment. The long period of human postnatal cortical development and the considerable plasticity it displays suggest that progressive modularization may arise simply as a consequence of the developmental process. Variations in developmental timing and the brain's capacity to carry out subtly different kinds of computation, together with differential structures in the environmental input, could suffice to structure the brain (Elman et al. 1996; Karmiloff-Smith 1992, 1995; Quartz and Sejnowsky 1997; Rose 1997). Nativists of course recognize that environmental input is essential to trigger developmental processes, but the environment only plays a very secondary role to the genome in such theories. In the alternative framework suggested above, there is no need to invoke innate knowledge or representations to account for resulting specialization, because of variations in developmental timing, different learning algorithms together with information inherent in different environmental inputs would together play a central role in the dynamics of development and in the gradual formation of module-like structures.
Anderson, J. R. (1980). Cognitive Psychology and its Implications. San Francisco: Freeman.
Anderson, M. (1992). Intelligence and Development: A Cognitive Theory. Oxford: Blackwell.
Baillargéon, R. (1994). How do infants reason about the physical world? Current Directions in Psychological Science 3:133-140.
Baron-Cohen, S. (1995). Mindblindness: An Essay on Autism and Theory of Mind. Cambridge, MA: MIT Press.
Bellugi, U., P. P. Wang, and T. L. Jernigan. (1994). Williams syndrome: An unusual neuropsychological profile. In S. H. Broman and J. Grafman, Eds., Atypical Cognitive Deficits in Developmental Disorders: Implications for Brain Function. Hillsdale, NJ: Erlbaum, pp. 23-56.
Bishop, D. V. M. (1997). Uncommon Understanding: Development and Disorders of Language Comprehension in Children. Hove, England: Psychology Press.
Butterworth, B., L. Cipolotti, and E. K. Warrington. (1996). Short-term memory impairment and arithmetical ability. Quarterly Journal of Experimental Psychology: Human Experimental Psychology 49A(1): 251-262.
Caramazza, A., R. S. Berndt, and A. G. Basili. (1983). The selective impairment of phonological processing: A case study. Brain and Language 18:128-174.
Chomsky, N. (1980). Rules and Representations. New York: Columbia University Press.
Elman, J. L., E. Bates, M. H. Johnson, A. Karmiloff-Smith, D. Parisi, and K. Plunkett. (1996). Rethinking Innateness: A Connectionist Perspective on Development. Cambridge, MA: MIT Press.
Farah, M. J., and J. L. McClelland. (1991). A computational model of semantic memory impairment: Modality-specificity and emergent category-specificity. Journal of Experimental Psychology: General 120:339-357.
Fodor, J. A. (1983). The Modularity of Mind. Cambridge, MA: MIT Press.
Frith, U. (1989). Autism: Explaining the Enigma. Oxford: Blackwell.
Gardner, H. (1985). Frames of Mind: The Theory of Multiple Intelligences. London: Heinemann.
Garfield, J. L., Ed. (1987). Modularity in Knowledge Representation and Natural-Language Understanding. Cambridge, MA: MIT Press.
Gelman, R., and C. R. Gallistel. (1978). The Child's Understanding of Number. Cambridge, MA: Harvard University Press.
Hollander, B. (1920). In Search of the Soul. New York: Dutton.
Johnson, M. H. (1997). Developmental Cognitive Neuroscience. Oxford: Blackwell.
Kant, E. (1781/1953). Critique of Pure Reason. Translated N. K. Smith. New York: Macmillan.
Karmiloff-Smith, A. (1992). Beyond Modularity: A Developmental Perspective on Cognitive Science. Cambridge, MA: MIT Press.
Karmiloff-Smith, A. (1995). Annotation: The extraordinary cognitive journey from foetus through infancy. Journal of Child Psychology and Child Psychiatry 36(8):1293-1313.
Karmiloff-Smith, A. (1998). Development itself is the key to understanding developmental disorders. Trends in Cognitive Sciences 2(10):389-398.
Karmiloff-Smith, A., J. Grant, I. Berthoud, M. Davies, P. Howlin, and O. Udwin. (1997). Language and Williams syndrome: How intact is "intact"? Child Development 68:246-262.
Leslie, A. M. (1988). The necessity of illusion: perception and thought in infancy. In L. Weiskrantz, Ed., Thought without language. Oxford: Oxford University Press.
Marslen-Wilson, W. D., and L. K. Tyler. (1987). Against modularity. In J. L. Garfield, Ed., Modularity in Knowledge Representations and Natural-Language Understanding. Cambridge, MA: MIT Press.
Mills, D. L., S. A. Coffey-Corina, and H. J. Neville. (1993). Language acquisition and cerebral specialization in 20-month-old infants. Journal of Cognitive Neuroscience 5(3):317-334.
Mills, D. L., S. A. Coffey-Corina, and H. J. Neville. (1994). Variability in cerebral organization during primary language acquisition. In G. Dawson and K. Fischer, Eds., Human Behavior and the Developing Brain. New York: Guilford Press.
Newell, A., and H. Simon. (1972). Human Problem Solving. Englewood Cliffs, NJ: Prentice Hall.
Pennington, B. F., and M. C. Welsh. (1995). Neuropsychology and developmental psychopathology. In D. Cicchetti and D. J. Cohen, Eds., Manual of Developmental Psychopathology, vol. 1. New York: Wiley, pp. 254-290.
Piaget, J. (1971). Biology and Knowledge. Chicago: University of Chicago Press. Originally published 1967.
Pinker, S. (1994). The Language Instinct. London: Lane.
Plaut, D. (1995). Double dissociation without modularity: Evidence from connectionist neuropsychology. Journal of Clinical and Experimental Neuropsychology 17:291-231.
Quartz, S. R., and T. J. Sejnowsky. (1997). A neural basis of cognitive development: A constructivist manifesto. Behavioral and Brain Sciences 20:537-556.
Rose, S. (1997). Lifelines: Biology, Freedom, Determinism. London: Lane.
Smith, N. V., and I. M. Tsimpli. (1995). The Mind of a Savant: Language Learning and Modularity. Oxford: Blackwell.
Spelke, E. S., K. Breinlinger, J. Macomber, and K. Jacobson. (1992). Origins of knowledge. Psychological Review 99:605-632.
Tyler, L. K. (1992). Spoken Language Comprehension: An Experimental Approach to Disordered and Normal Processing. Cambridge, MA: MIT Press .
Copyright © 1999 Massachusetts Institute of Technology