ABSTRACT
Most cognitive psychologists agree that they are trying to characterize how the brain functions is the course of processing information. The "cognitive" level of analysis, then, is at a sufficiently abs•ract level that it allows one to compare the mind to computer programs. Both mental states in the brain and program states in a computer can be described using a functional vocabulary. This vocabulary specifies the nature of structures and processes and specifies how sequences of processes can be coordinated to accomplish some goal. It is natural, then, that many cognitive psychologists have begun to use the computer as a way of constructing models of mental processes. The usual goal is to program the computer in such a way that its behavior emulates that of the human, presumably because both the mind and the program have the same functional states (see Anderson & Bower, 1979/1973; Newell & Simon, 1972). In this paper we describe a somewhat different use of computer models of cognitive processes, one aimed at guiding an empirical research program rather than simply providing the bases for plausible accounts of data.
