Symbolic and sub-symbolic knowledge organization in the computational theory of mind
Michael Huth · KNOWLEDGE ORGANIZATION · 1995
We sketch the historic transformation of culturally grown techniques of symbol manipulation, such as basic arithmetic in the decimal number system, to the full-fledged version of the Computational Theory of Mind. Symbol manipulation systems had been considered by Leibniz as a methodology of inferring knowledge in a secure and purely mechanical fashion. Such inference calculi were considered as mere artefacts which could not possibly encompass all human knowledge acquisition. In Alan Turing's work one notices a crucial shift of perspective. The abstract mathematical states of a Turing machine (a kind of calculus universalis that Leibniz was looking for) are claimed to correspond to equivalent psychological states. Artefacts are turned into faithful models of human cognition. A further step toward the Computational Theory of Mind was the physical symbol system hypothesis, contending to have found a necessary and sufficient criterion for the presence of intelligence in operative mediums. This, together with Chomsky's foundational work on linguistics, led naturally to the Computational Theory of Mind as set out by Jerry Fodor and Zenon Pylyshyn. We discuss problematic aspects of this theory. Then we deal with another paradigm of the Computational Theory of Mind based on network automata. This sub-symbolic paradigm seems to avoid problems occuring in symbolic computations, like the proble frame problem and graceful degradation