Altered States and Virtual Beliefs
Jon Oberlander, Peter Dayan · 1996
Abstract Functionalism derives its explanatory force from its firm grounding in the certainties of Turing machines (TMs) and their practical approximations, the von Neumann machines (vNMs). One branch of cognitive psychology, and its philosophical stablemate, has taken this doctrine seriously, and coupled it with some of the basic notions of folk psychology such as beliefs, desires, hopes, etc. Most of the theories are wedded rather firmly to these architectures, with all their connotations of logicality. They therefore face problems explaining the apparent illogicality and obvious tractability of human cognition. In desperation, some functionalists, whilst naturally retaining the Turing metaphor at the heart of their theories, have been forced to abandon these folk psychological constructs. The switch is either to instrumentalism, proposing that these constructs are not causal in producing behaviour but are still useful for explaining or predicting it [4], or else to a more radical position, holding that they are not even satisfactory for that [23]. Connectionism, for all its many faults, gives the first hint that something may be wrong with this retreat. It demonstrates that, despite essential computational equivalence, the strict analogy with TMs that has survived into vNMs is not the only way of conceiving of the functional mechanisms of thought. Furthermore, on broadening the range of possible mechanisms, it becomes apparent that too simplistic a notion of human competence is potentially dangerous. Unfortunately, connectionism seems rather remote architecturally and representationally from high-level psychological entities, so some independent motivation behind alternative functional mechanisms is required.