ABSTRACT
This article presents a process model of category learning that incorporates representations of both exemplars and rules. Unlike most models of category learning, this model, using an ACT-R inspired activation and decay mechanism, incorporates memory effects such as forgetting. The learning model is combined with an existing EPIC-Soar performance model and applied in a concurrent learning and perceptual-motor performance task. The model's fits and predictions are presented. We attribute the model's success, in part, to a) the architectural integration of elements of Soar, EPIC, and ACT-R, b) inheriting the validation of these systems, and c) accepting the modeling constraints of these mechanisms.
