Some Remarks about Connectionism in Psychological Simulation

This paper considers the question of whether the connectionist approach is an appropriate p a r a d . for the construction of simuiation modeis of psychologicai processes. The strengest connectionist dah is that psychological properties simply emerge from interactions among the units of a netwark; meaningfd features are distributed in these units, which therefore are not representions of anything. The problern is then: how such a system can descriie a psychological system and be relevant to explain it ? It is suggested that the psychological relevante of comectionist descriptions should not be rejected on the grounds that they are limited to a hardware-iike or implementational level. Connectionist descxiptions would be best considered as different scientific "objects", cunstructed from the same pre-scientific "facts*. As different objects they need not be at different levels of generality or abstraction and the translation of one representation into another is unnecessary. As to the explanatory power of connectionist modeis, it is claimed that it wnnot come from the interaction among units, if this interaction is invoked at the same time to give the model a psychologicai nature. Neither can the expianatory power come from the fact that comectionist modeis produce phenomena similar to human psychologicd processes, since ehis sixnüarity should refer explicitly to features that have already been accepted independentiy as psychological The claim that co~ectionist anaiysis is psychologically relevant at a subsymbolic level is eventuaily considered. There is a possible confusion in this claim. sub-symbolic models may be non-symbolic as to their object but they must be symbolic as models. A non-symbolic model simply wodd not be a model, since by dejinition it must represent something. Sub-symboiic models, however, may be non-symbolic in the sense that there are no symbols in the modelled reality. In thic regard, comectionist Systems seem to help in overcoming the idea encouraged by current Ai and Cognitive Science that any psychologicai process should be considered as made out of symbols. Revised version of a paper prcsented at the 7th Workshop of the ESSCS, St. Maximin-laSainte-Baume, Prwence, Franct, 19-22 june 1989. Cogzitive Systems 2-4, 359-375 May 1990 @ 1990 ESSCS