Secondary Generalisation in Categorisation: An Exemplar-based Account

The parallel rule activation and rule synthesis (PRAS) model is a computational model for generalisation in category learning, proposed by Vandierendonck (1995). An important concept underlying the PRAS model is the distinction between primary and secondary generalisation. In Vandierendonck (1995), an empirical study is reported that provides support for the concept of secondary generalisation. In this paper, we re-analyse the data reported by Vandierendonck (1995) by fitting three different variants of the Generalised Context Model (GCM) which do not rely on secondary generalisation. Although some of the GCM variants outperformed the PRAS model in terms of global fit, they all have difficulty in providing a qualitatively good fit of a specific critical pattern.