How to fit transfer models to learning data: a segmentation/clustering approach

Abstract

Models of category transfer do not have the ability to evolve over time. This feature constrains them to only account for participants' generalization patterns. Although they can model fewer processes, transfer models have repeatedly shown to be a useful tool for testing categorization theories and for precisely predicting participants' performance. In this study, we propose a statistical framework that allows transfer models to be applied to learning data. This framework is based on a segmentation/clustering technique, that is here specifically tailored for suiting category learning data. The adjusted technique is then applied to a well-known transfer model (the Generalized Context Model) on three novel experiments. More specifically, these experiments manipulate ordinal effects in category learning by contrasting rule-based vs. similarity-based orders in three contexts. The difference in performance across the three contexts, as well as the benefit of the rule-based order observed in two out of three experiments was almost entirely detected by the segmentation/clustering method. We conclude that our adjusted segmentation/clustering framework allows one to fit transfer models to learning, while apprehending essential information in categorization.