CUE COMPETITION IN HUMAN CATEGORIZATION - CONTINGENCY OR THE RESCORLA-WAGNER LEARNING RULE - COMMENT

Shanks (1991) reported experiments that show selective-learning effect s in a categorization task, and presented simulations of his data usin g a connectionist network model implementing the Rescorla-Wagner (R-W) theory of animal conditioning. He concluded that his results (a) supp ort the application of the R-W theory to account for human categorizat ion, and (b) contradict a particular variant of contingency-based theo ries of categorization. We examine these conclusions. We show that the asymptotic weights produced by the R-W model actually predict systema tic deviations from the observed human learning data. Shanks claimed t hat his simulations provided good qualitative fits to the observed dat a when the weights in the networks were allowed to reach their asympto tic values. However, analytic derivations of the asymptotic weights re veal that the final weights obtained in Shanks' Simulations 1 and 2 do not correspond to the actual asymptotic weights, apparently because t he networks were not in fact run to asymptote. We show that a continge ncy-based theory. that-incorporates the notion of focal sets can provi de a more adequate explanation of cue competition than does the R-W mo del.