Abstract

This essay will look at two approaches to psychological theory. The received view, the Classical theory, has it that psychological states are relations to representations that have syntactic structure, like sentences, and that functions from one state to another are computed with sensitivity to that structure, like computation of the sort found in digital computers. In the rival approach, Connectionism, functions are computed by spreading activation between neuron-like units. Typical networks consist of a set of input units, a set of output units, and a set of hidden units that mediate the spread of activation between the input and output units. Psychological states here are not syntactically structured, and the computation promises to be brain-style. ;Fodor and Pylyshyn criticize the Connectionists denial of constituent structure and structure sensitive computations. They claim that the traditional arguments in favor of the constituent syntactic structure of mental representations cannot be met by anything the Connectionist has to offer. The first part of this essay will outline both the Classical theory and the Connectionists' alternative and conclude that Connectionism can meet the arguments in favor of syntactically structured mental representations and structure sensitive processes. In so doing, the Connectionist is committed to a causal theory of representation; this sort of theory is examined in detail in order to establish its legitimacy. ;In the second part of the essay a positive argument is put forward for the Connectionist approach that exploits the close ties between Connectionism and the neurosciences. It will be argued that a condition of adequacy on an information flow theory is that it must be reducible to a true neurophysiological theory. Connectionist theories, I will argue, will satisfy this condition while Classical theories will not. The last chapter will enter the individualism debate in order to answer a potential objection to our reductive program.