Nativism is often understood as the view that a significant body of knowledge is "built in" to an organism, or at least innately predetermined. This characterization, however, fails to capture contemporary nativism as well as being inadequate for many older views (see NATIVISM, HISTORY OF). Few nativists argue today for the full predetermination of specific concepts, ideas, or cognitive structures such as a language's grammar; and few empiricists fail to argue for certain kinds of information processing, such as back propagation (see SUPERVISED LEARNING), as being built in. Every party to current debates about nativism in fact shares the view that there is something special and intrinsic, that is innate, to particular types of organisms that enables them to more easily come to engage in some behaviors as opposed to others. It is the nature of those intrinsic structures and processes that is the true focus of debates about whether some aspect of cognition or perception is compatible with a nativist perspective. In particular, nativist views endorse the presence of multiple learning systems each of which is especially effective at acquiring a particular kind of information and where that effectiveness arises from specializations for information that occurs at all levels in that learning system, not just in the initial stages of processing. The rise of connectionism (see CONNECTIONISM, PHILOSOPHICAL ISSUES and COGNITIVE MODELING, CONNECTIONIST) has been said to pose a fatal challenge to nativism; but as seen later in this article, it in itself in no way renders nativism obsolete.

Confusions about nativism often arise in cases where one organism can acquire a body of knowledge or an ability that another cannot. Thus, attempts to teach human language to primates are often seen as bearing directly on nativist views of an innate language even as the researchers on such topics are usually much more cautious (e.g., Savage-Rumbaugh et. al. 1993). But differential success at learning is in itself not relevant. This irrelevance is clear when more extreme comparisons are made. When a child acquires language and a pet gerbil in the same environment does not, no one argues that language is therefore innate in humans. Failure to learn can arise for many reasons, only some of which support a nativist perspective. There may be general cognitive capacity requirements necessary for learning complex knowledge that exceed the capacities of some organisms. When a gerbil fails to learn language, we might well assume that it simply could not acquire any knowledge system with the structural complexity and memory loads imposed by language. When a primate fails to learn a language, it too may fail to pass some general capacity threshold. Alternatively, a primate that is highly adept in some sorts of complex cognitions might fail at language acquisition because it does not have capacities that are specifically tailored for the pickup and learning of linguistic structure. It may fail because it does not "know" enough in advance about some specific properties in the domain of natural language, a pattern of failure that is compatible with a nativist view of knowledge and its origins. But that prior "knowledge" does not have to be in the form of an innately represented set of grammatical rules; it can be a set of powerful biases for interpreting linguistic information in highly constrained ways. Specifying those biases and constraints is where the real distinctions between contemporary nativists and empiricists reside (Keil 1981, 1998). Indeed, it has recently been argued that, even in traditional biology, constraints thought of as "canalization" are the best way of understanding innateness (Ariew 1996).

Although it is more common in philosophy to distinguish RATIONALISM VS. EMPIRICISM, in cognitive science today it is nativism that is usually pitted against empiricism, where, for a nativist, knowledge of such things as grammar or folk psychology does not arise from simply having rational thought and its logical consequences, but rather from having information-specific learning biases that go beyond the more content-neutral mechanisms of learning favored by empiricists, such as unconstrained associationism. Nativists and empiricists disagree on whether one organism achieves greater learning success than another because it is more cognitively capable in general or because it has specialized structures tuned to learn a particular kind of knowledge. This is the question of DOMAIN SPECIFICITY that has become such a pivotal issue in cognitive science today, especially in EVOLUTIONARY PSYCHOLOGY (Fodor 1983; Hirschfeld and Gelman 1994; Keil 1981; Cosmides and Tooby 1994).

Domain specificity alone, however, is not enough to characterize nativism. The specialization for information must involve a certain kind and "level" of processing. The eye is tailored for different kinds of information than the ear (VISUAL ANATOMY AND PHYSIOLOGY, AUDITORY PHYSIOLOGY), a fact well known to both nativists and empiricists for centuries. But empiricists see those specializations as soon disappearing when that information flows beyond the sensory transducers. If all of thought and all patterns of learning can be explained by general laws, such as those of association, once one goes beyond the specializations of the sense organs, then nativism founders. If, however, there are specialized systems for building up representations and processes in specific domains, whether they be language, biology, or number (LANGUAGE ACQUISITION, FOLK BIOLOGY, NAIVE MATHEMATICS), and general learning principles seem inadequate, nativism is supported.

Consider the difference between having a system that is tuned to expect certain patterns in a specific modality, such as the eye's "expectations" concerning reflected light patterns, and a system that has expectations that transcend modalities, such as that two physical bodies cannot interpenetrate (Spelke 1994). The second expectation can be borne out tactilely, visually, and possibly even auditorily. This expectation is still domain specific in that it applies only to bounded physical objects and not fluids, gases, or aggregates. Systems that are tuned to patterns that transcend modalities would therefore be more likely to fit with a nativist stance.

Connectionist architectures can favor either empiricist or nativist orientations depending on their implementation. A system with preset weights and compression algorithms that seem to be optimized for learning only certain kinds of information, such as that of spatial layout, might well support a nativist account. If such weights, however, only bias the learner toward low-level perceptual features, an empiricist approach is supported (Seidenberg 1992). In some models, a low-level bias, such as selective attention in human infants for moving triangular dot patterns, has been argued to result in a "face processing area of the brain," even though there were no initial biases in that region for faces initially (Johnson and Morton 1991). Thus, an end state of domain-specific processing of a particular kind of information that is localized in a specific region of the brain is not by itself nativist. The way in which that specialized processing was set up is critical. Similarly, a recent interest in "emergentism" in connectionist systems, namely ways in which general learning systems can yield unpredictable emergent higher-order properties (MacWhinney forthcoming), does not displace nativism as much as it makes apparent the subtlety needed for arguments about the origins of various types of knowledge.

In short, nativists and empiricists primarily disagree on the extent to which pre-existing biases for specific domains of information go beyond those in effect at the levels of sensory transducers. The sense of domain also shifts, with domains at the sensory levels being patterns of information such as "light" or "sound waves" and domains at higher levels being patterns corresponding to such things as bounded physical objects, intentional agents, number, or spatial layout. All of these domains of the second sort clearly are amodal and more cognitive than perceptual.

Biases on high-level cognition that work in domain-general ways do not need to support nativism. For example, the base rate fallacy (Kahneman, Slovic, and Tversky 1982) would seem to apply to any kind of experienced information, regardless of its domain. As such, it would seem to be a further modification of general laws of learning, such as those on association, all of which fits with empiricism. If, however, this bias were to be much more prominent in cases of social attribution, and seemed to help learning about social situations, it would be considered domain specific.

Some have argued that the nativism/empiricism controversy is seriously misguided because of the intrinsically interactional nature of development (Lehrman 1953). This notion has been raised again more recently (Elman et al. 1996) in attempts to argue that it makes no sense to ask what is innate in dynamic learning systems. These objections, however, attack a caricature of "innate structures" and not the current debate between nativists and empiricists. Part of the confusion is between specifying particular behaviors or pieces of knowledge as innate, as opposed to being products of the learning function itself and the cognitive biases it engenders. When learning is considered as a function from sets of environments to sets of mental representations (Chomsky 1980), the intrinsically interactional nature of learning is part of the formulation.

See also

Additional links

-- Frank Keil


Ariew, A. (1996). Innateness and canalization. Philosophy of Science 63:19-27.

Chomsky, N. (1980). Rules and Representations. New York: Columbia University Press.

Cosmides, L., and J. Tooby. (1994). The evolution of domain specificity: The evolution of functional organization. In L. A. Hirsch feld and S. A. Gelman, Eds., Mapping the Mind: Domain Specificity in Cognition and Culture. Cambridge: Cambridge University Press.

Elman, J. L., E. A. Bates, M. H. Johnson, A. Karmiloff-Smith, D. Parisi, and K. Plunkett. (1996). Rethinking Innateness. Cambridge, MA: MIT Press.

Fodor, J. A. (1983). Modularity of Mind. Cambridge, MA: MIT Press.

Hirschfeld, L. A., and S. A. Gelman, Eds. (1994). Mapping the Mind: Domain Specificity in Cognition and Culture. Cambridge: Cambridge University Press.

Johnson, M. H., and J. Morton. (1991). Biology and Cognitive Development: The Case of Face Recognition. Cambridge, MA: Blackwell.

Kahneman, D., P. Slovic, and A. Tversky. (1982). Judgement under Uncertainty: Heuristics and Biases. New York: Cambridge University Press.

Keil, F. C. (1981). Constraints on knowledge and cognitive development. Psychological Review 88:197-227.

Keil, F. C. (1998). Cognitive science and the origins of thought and knowledge. In R. M. Lerner, Ed., Theoretical Models of Human Development. Vol. 1, Handbook of Child Psychology. New York: Wiley.

Lehrman, D. (1953). A critique of Konrad Lorenz's theory of instinctive behavior. Quarterly Review of Biology 28:337-363.

MacWhinney, B. J., Ed. (Forthcoming). Emergentist Approaches to Language. Hillsdale, NJ: Erlbaum.

Savage-Rumbaugh, E. S., J. Murphy, R. A. Sevick, K. E. Brakke, S. L. Williams, and D. Rumbaugh. (1993). Language Comprehension in Ape and Child. Monographs of the Society for Research in Child Development, 233. Chicago: University of Chicago Press.

Seidenberg, M. S. (1992). Connectionism without tears. In S. Davis, Ed., Connectionism: Theory and Practice. Oxford: Oxford University Press.

Spelke, E. (1994). Initial knowledge: Six suggestions. Cognition 50:431-445.

Further Readings

Chomsky, N. (1988). Language and Problems of Knowledge: The Managua Lectures. Cambridge, MA: Bradford Books/MIT Press.

Fischer, K. W., and T. Bidell. (1991). Constraining nativist inferences about cognitive capacities. In S. C. A. R. Gelman, Ed., The Epigenesis of Mind: Essays on Biology and Knowledge. Hillsdale, NJ: Erlbaum, pp. 199-235.

Fodor, J. (1981). Representations: Philosophical Essays on the Foundations of Cognitive Science. Cambridge, MA: MIT Press.

Keil, F. C. (1990). Constraints on constraints: Surveying the epigenetic landscape. Cognitive Science 14:135-168.

Keil, F., C. Smith, D. Simons, and D. Levin. (1998). Two dogmas of conceptual empiricism. Cognition 65(2).

Lerner, R. M. (1984). On the Nature of Human Plasticity. New York: Cambridge University Press.

Leslie, A. (1995). A theory of agency. In A. L. Premack, D. Premack, and D. Sperber, Eds., Causal Cognition: A Multi-Disciplinary Debate. New York: Oxford, pp. 121-141.

Lightfoot, D. (1982). The Language Lottery: Toward a Biology of Grammars. Cambridge, MA: MIT Press.

Mandler, J. M. (1992). How to build a baby: II. Conceptual primitives. Psychological Review 99:587-604.

McClelland, J. L., M. Bruce, L. O'Reilly, and C. Randall. (1995). Why there are complementary learning systems in the hippo-campus and neocortex: Insights from the successes and failures of connectionist models of learning and memory. Psychological Review 102(3):419-437.

Meltzoff, A. N., and M. K. Moore. (1989). Imitation in newborn infants: Exploring the range of gestures imitated and the underlying mechanisms. Developmental Psychology 25:954-962.

Piatelli-Palmarini, M. (1994). Ever since language and learning: Afterthoughts on the Piaget-Chomsky debate. Cognition 50:315-346.

Pinker, S. (1994). The Language Instinct. New York: Morrow.

Pinker, S. (1997). How the Mind Works. New York: W. W. Norton.

Prince, A., and P. Smolensky. (1997). Optimality: From neural networks to universal grammar. Science 275:1604-1610.

Spelke, E., and E. Newport. (1998). Nativism. In R. M. Lerner, Ed., Theoretical Models of Human Development. Vol. 1, Handbook of Child Psychology. New York: Wiley.

Wynn, K. (1995). Origins of numerical knowledge. Mathematical Cognition 1.