Connectionist and rule-based representations of expert knowledge

Research in computer science has led to the development of two broad classes of models of knowledge representation: rule-based and connectionist systems. Both techniques solve the same abstract problem, that is, the assignment of cases to classes. Connectionist modeling techniques have been applied to three classification situations in which one would expect rule-based models to be applicable. Two of the situations involved the diagnosis of problems in and operation of a fictitious power plant. The third situation involved the classification of misconceptions held by elementary physics students. Connectionist modeling developed adequate simulations of behavior in all three cases.