ladkin@kestrel.ARPA (Peter Ladkin) (07/25/86)
(Ellis) > Homunculism is by no means dead. Dennett in "Brainstorms" speaks > of mental processes in terms of progressively stupider homunculae. Not only that, but it should even be respectable by now. Haugeland's book indicates that modularisation of a formal game-playing system can allow a system to exhibit behaviour that looks intelligent (as in chess-playing systems) from components that by themselves do basic, easily comprehensible (rule-based) tasks. (This is only the latest, more careful version of a very old argument). If one doesn't believe in the full possibility of artificial intelligence, could one at least believe in artificial homunculi? Peter Ladkin ladkin@kestrel.arpa