A connectionist model, which simulates the operation of prefrontal circuits during Stroop task is proposed. The Stroop test has traditionally been used as a measure of cognitive inhibition. The task is to inhibit an over-learned, habitual response (i.e., reading color words) in favor of an unusual, novel requirement (i.e., naming incongruously printed colors of color words). The longer durations in completing the task indicate an inability to inhibit habitual but contextually inappropriate response tendencies, which is suggestive of a prefrontal dysfunction. The connectionist model is designed adapting artificial neural networks (ANNs) in such a way that each ANN corresponds to a particular neuroanatomic component of the prefrontal circuit which is likely to take part in the execution of the Stroop task. The ability of the proposed model to simulate the normal and the abnormal performance on the Stroop task is tested. The simulation results show that the model is consistent with the clinical data. (c) 2006 Elsevier B.V. All rights reserved.