The Structure of Cognitive Representations

Abstract

The most significant area of debate separating Classical and Connectionist theories of cognition has been the family of systematicity and productivity arguments best known from Jerry Fodor and Zenon Pylyshyn’s paper, “Connectionism and Cognitive Architecture: A Critical Analysis.” These arguments are not the sole basis upon which to try to determine the nature of human cognitive architecture, but they are currently among the most prominent. In roughest outline, the arguments are quite simple. Thought displays certain features, namely, it is systematic, compositional, inferentially coherent, and productive. These features of thought can be better explained by a Classical theory of cognition than they can be by a Connectionist theory, hence there is some defeasible reason to prefer Classicism to Connectionism. Beyond this, however, there are ambiguities, vagaries, and confusions at all levels of detail. At the most general level, therefore, the aim of the present work is simply to clarify and refine the systematicity and productivity arguments, showing how the central ideas of these arguments may be applied to a range of theories of cognitive architecture.