Rethinking innateness is a timely volume which forcefully demonstrates the importance of modelling in understanding development, ‘innateness’, and the nature of change. It provides an inspiring vision of what developmental psychology could one day be like, linking behaviour and biology via connectionist models. However, Rispoli's worry about the book's potential for detrimental polarization does not seem unfounded. One aspect of the book that deserves comment in this respect is the focus on connectionism to the exclusion of other types of model. It is unclear from Rethinking innateness itself whether this exclusion of other approaches merely stems from the legitimate desire to write a focused book, or whether it possibly reflects an actively held view that connectionism is the one true approach to modelling development. Regardless of the authors' intentions, the role of connectionism in relation to other computational approaches is an issue which is particularly pertinent to the study of language acquisition.

To clarify straightaway, I not only concur with Elman et al. on the central role of modelling, but also strongly believe that the task of the cognitive scientist is not complete until one has an account of how a particular process is realised in a neural architecture. These two commitments give connectionist models a central role. But they do not make connectionism the exclusive modelling tool for the study of language acquisition, nor necessarily the best path to currently pursue. The reasons for this are twofold.

The first is a matter of research strategy. It is possible that more rapid progress and greater success might be made if one starts with high-level models which give less immediate regard to matters of implementation.

The debate between nativism and empiricism has been the central issue in developmental psychology for well over three centuries. Nowhere has the opposition between nature and nurture been felt more keenly than in the study of child language development. At times the debate has proven useful, or at least entertaining. But there are those, like Matthew Rispoli, who worry that a fixation on this issue may reinforce a polarization that will eventually ‘run developmental psycholinguistics into the ground’.

Rethinking innateness (RI) is a quixotic book, enthusiastic and visionary. It is hard not to be provoked as each author races full tilt towards their chosen combat. Some readers, like Fodor (1997), respond in kind. Rispoli is more measured. Nonetheless, every reader chooses their own windmills to elevate into giants, and each believes their battle righteous. Are Elman et al. ‘building castles in the air’, or is Rispoli ‘taking the wrong sow by the ear’?

This commentary concerns two issues: the contribution of connectionism and theories of learning to developmental psychology; and the status of connectionist models of syntax acquisition.

In his review, Rispoli's main concern is that Elman et al.'s book will aggravate the degree of polarization in developmental psycholinguistics. I cannot really comment on this worry, as developmental psycholinguistics is not my field. Instead, I will discuss some questions more related to my background – the role of computational modelling in Elman et al.'s approach.

Elman et al.'s ambitious goal is to propose theories of cognitive development that are grounded in our knowledge of biology. This is of course what the great Jean Piaget tried to achieve during his lifetime – unsuccessfully, as we know. Elman et al.'s advantage over Piaget is that they have a set of computational tools, connectionism, which both allows them to specify theories precisely and to study complex behaviours (such as epigenesis, where innate and environmental factors interact to create new levels of complexity) that are just beyond the (unaided) human mind. Even though I will highlight some of the weaknesses of their approach below, I should emphasize that reading their book was an exciting and enjoyable experience.

As noted by Rispoli, there are important problems with the simulations reported by Elman et al. Rispoli focuses on simulations of past tense acquisition and syntax acquisition, but the problems are by no means limited to these areas. I will briefly consider two recent developments in neural net research, one taken from the field of language acquisition, and one from elsewhere, which underscore some of the difficulties of the simulations discussed in the book.

There are a number of ways to react to the whole problem of polarization that Rispoli's review brings up. As a first reaction, like Rispoli, personally, I deplore the polarization. I find neither side likely to be right enough about everything to be entitled to require the kind of all-or-nothing allegiance each side seems to demand. My complaints about the two sides are more basic than Rispoli's rather specific criticisms of Elman et al. Unfortunately, going on to outline what seems generally useful and less useful in each approach in a short space like this basically consists of saying ‘I like this’ or claiming ‘this is obviously wrong’ with a little elaboration. This does not seem useful for such complicated problems, so I will leave my general complaint about the all-or-none polarization simply stated outright.

We enjoyed reading Matthew Rispoli's review of Rethinking innateness (henceforth RI-Elman et al. 1996). First, we respond to what we see as the major issues that he raises. Finally, we respond to the commentaries contributed by other authors in this issue.

Rispoli has two major concerns. First, he fears that RI will do the field a great disservice, increasing polarization and sowing discontent. To some extent this is an ‘eye of the beholder’ criticism that is difficult to counter: if Rispoli feels more polarized upon reading this book, who are we to argue?

On the whole, I quite like both the target book (RI) and the review by Matthew Rispoli, but there are some additional observations that might be of interest to the readers of this journal. The whole concept of ‘truth through disputation’ is alien to my scientific tradition and I agree with Rispoli that RI is not helped by the polemical tone.

There is a companion volume and software suite by Kim Plunkett & Jeff Elman (1997), Exercises in rethinking innateness, which we have used in an undergraduate cognitive science course. This is, in my opinion, the best source for understanding the main point of RI, which I summarized for the class as: the book tries to show that PDP learning techniques have advanced to the point where we need not assume that tabula rasa learning of language must be ruled out. The Exercises in RI book and particularly doing the on-line examples give the students direct intuition about this claim. Our students were impressed by the Tlearn system, but understood its limitations and were not convinced either of the main claim of RI or of the radical nativist alternative.