Individual differences in exemplar-based interference during instructed category learning

Instructed category learning studies have shown that categorization practice on a fixed set of labeled training exemplars can cause learners to violate explicitly provided categorization instructions. We have previously proposed a connectionist account of this exemplar-based interference effect - an account which predicts that individuals who display initial difficulty in the application of a categorization rule will exhibit greater exemplar-based interference than good rule-followers. In this paper, we report on a study of human instructed category learning performance intended to test this prediction of the model, and we provide the results of additional connectionist simulations which are fit to the human experimental data.