What’s Within? Nativism Reconsidered

By Fiona Cowie

Oxford University Press, 1998. xvii + 334 pp. £ 35.00 cloth.

This is in many ways a good book: careful, knowledgeable, intelligent, brave and often insightful. But it by no means lives up to the rave descriptions provided on the dust-jacket. To begin with, the book is actually much narrower in scope than its title and endorsements would suggest. It is not a consideration and critique of contemporary nativism as such. Rather, it focuses on Fodor’s arguments for the innateness of almost all of our concepts, on the one hand, and on Chomsky’s ‘poverty of stimulus’ argument for the innateness of a structured language-learning mechanism on the other. Since Fodor’s position on concepts is highly idiosyncratic and is (so far as I am aware) endorsed by absolutely no one else in the world, an attack on his views is of limited interest. No one else who believes in innate concepts will find anything in this book to bother or concern them. Moreover, only in a rather brief final chapter does Cowie consider (and tentatively endorse) some of the many other arguments for the innateness of language; let alone consider the many and varied arguments which have been proposed for believing in innately structured learning mechanisms and domain-specific cognitive faculties in other areas of cognition. So even if all Cowie’s criticisms of nativist arguments were successful and accepted, the impact on cognitive science should actually be small.

The book falls into three parts. In the first, Cowie considers the history of the empiricist–nativist debate, concentrating mostly on Locke, Hume, Descartes and Leibniz. She argues that the debate divides into two main distinct strands. On the one hand, nativists like Leibniz were inclined to argue for the existence of what we would now call domain-specific learning mechanisms; rejecting the empiricist picture that all learning employs the same general principles. And on the other hand, nativists were also motivated by anti-naturalism, claiming that the various processes involved in knowledge-acquisition must fall outside of the purview of any natural science; this being opposed by empiricists who insisted that knowledge-acquisition must itself be a natural process, governed by natural laws. This part of the book is valuable and historically sensitive; but it is unlikely to be of much interest to those who want to understand, or to know what they should believe in relation to, contemporary debates concerning nativism.

In the second part of the book Cowie examines Fodor’s view that (almost) all concepts are innate. This view is said to align with the anti-naturalist strand in classical nativism, since Fodor is shown as insisting that concept-acquisition cannot receive any cognitive explanation. But this is no more than a not-very-illuminating metaphor. For there is no question of Fodor (or any other contemporary nativist) being motivated by anti-naturalism, in the way that the early nativists were. (Indeed, many of us think that the boot is now on the other foot, and that it is now anti-nativism which is the last refuge of anti-scientific attitudes in some areas of the humanities and social sciences.)

More importantly, however, Cowie’s concentration on Fodor leads her to mis-represent the nature of the debate between empiricists and nativists about concepts. She takes over from Fodor the view that empiricists allow that simple concepts are innate (= unlearned), but that most concepts are complex and learned on the basis of simple ones. The debate then becomes whether many concepts are innate (empiricism) or almost all are (nativism). But this is a mistake. Locke and Hume thought that all concepts were learned from experience – the simple ones by copying from previous impressions (Hume) or by abstraction from experience (Locke: there is a case for saying that at least some concepts must be presupposed in this process, namely concepts of sameness and difference); and then complex ideas are acquired by compounding out of simple ones. (The quotations which Cowie provides on pp. 107-8 are misinterpreted. They are actually about the mysteriousness of world–mind interactions, not about the mysteriousness of the generation of concepts in experience. On the contrary, both Locke and Hume thought that the learning of simple concepts was remarkably easy.)

One can readily imagine a contemporary version of empiricism about concepts. It might be claimed that innately-structured perceptual mechanisms initially deliver a set of non-conceptual experiential contents, say 3-D representations of the ways in which surrounding space is filled. These form the basis for laying down a set of perceptual templates in memory, which serve as the foci for prototype-classifications into kinds, which are in turn elaborated by accumulating further information by cross-classification, and so on. The question which needs to be addressed is whether this sort of picture is basically correct, or whether, as many claim, we need to postulate innate concepts in various domains. The question should not be whether almost all concepts are innate as opposed to just many; rather, it is whether any significant number are innate as opposed to none or very few. Unfortunately, Cowie has nothing to say about this question, and conducts none of the necessary analysis and examination of the kinds of argument for innate concepts which might be deployed.

Many readers would be well-advised to skip the first two parts of the book, and get straight into the third. This is an extended examination of ‘poverty of stimulus’ arguments for an innate language faculty, as deployed by Chomsky and others. It presents a powerful and well worked-out challenge to the nativist about language (albeit somewhat limited in scope), which deserves to be studied and responded to in detail. This section of the book vindicates, at least partially, the rave endorsements on the back cover. Here I have space to raise just two swift points.

First, Cowie’s case depends upon distinguishing between three different claims about language acquisition: that it proceeds by principles which are domain-specific; that it proceeds by means of principles which are innate; and that it makes use of UG (or Chomsky’s ‘Universal Grammar’). She allows that the case for domain-specificity is very powerful, but claims that this need not entail either innateness or UG; rather, the domain-specific principles may themselves have been learned at an earlier stage in language acquisition. But it is very hard to see how this could work. For what has to be explained is that all children proceed through the same learning sequence, but on the basis of varied and idiosyncratic inputs. Merely postulating an initial domain-general tendency to look for underlying regularities, as Cowie does, will not carry us very far. What needs to be explained is how all children converge, at a very early age (2+ years), on the search for the quite specific sorts of syntactic structure-dependencies that they do.

Second, Cowie complains that the data concerning the poverty of the child’s linguistic stimuli have never been gathered, at least in any detail. In which case an argument based on such alleged poverty simply cannot be evaluated. But this seems disingenuous. For as Cowie herself notes in her closing sections of the book, there exists a powerful set of data to make precisely this point, namely the evidence provided by various forms of Creolisation. Children who are known to receive very degraded linguistic inputs (for example, being brought up in an environment in which all those around them speak an impoverished form of pidgin language) will nevertheless regularise and extend the language learned into a fully-structured linguistic system (a Creole) in the course of a single generation.

University of Sheffield Peter Carruthers