MULTIPLICATION NUMBER FACTS - MODELING HUMAN-PERFORMANCE WITH CONNECTIONIST NETWORKS

Three connectionist models of human performance on simple multiplicati on number facts, commonly called ''times tables,'' are reviewed. Also, human data from normal subjects and brain-damaged patients, which con strain these models, are presented. These human data include the probl em size effect, error effects, priming effects, use of strategies and rules, and number representation. The connectionist models presented a re: a simple auto-associator (J. A. Anderson's Brain-Stare-in-a-Box), a standard back-propagation model, and McCloskey and Lindemann's MATHN ET. The review of human data and connectionist models of memory retrie val provides some insight into the strengths of, differences between, and challenges for, this approach to computational modeling. Particula r attention is paid to the representation of number used by these mode ls, and a related ability to generalize learning.