The kinds of models represented in this volume, variously referred to as neural networks, parallel distributed processing (PDP), or connectionist models, have received much attention in the past several years, but connectionism has longer history in psychology (for an entertaining summary, see Tolman 1948). The failure of simple stimulus-response (S-R) associations to account for the complex­ ity of animal learning led to the postulation of associative mediators, coding responses, stimulus analyzers, and expectancies. All of these ideas represented attempts to make connectionist models exhibit flexible information processing.