Neural nets as models for study of multivalued logic

Summary form only given. It is suggested that since one can now implement neural nets which can actually perform certain basic information processing functions, it is of interest to see if one can fashion nets or systems of nets which can reproduce (i.e. mimic) certain trains of actions regularly performed by humans. The particular issue addressed is that of implementing a multivalued logic system in a neural net, concentrating on one such logic system, namely fuzzy set logic. The focus is on how a network might be used to describe a membership function and how a (multivalued) fuzzy logic system might also be accommodated with such a net.<
>