We report a supervised category learning experiment in which the training phase contains both classification and observation learning blocks. To explain the use of different categorization strategies, we propose an account in which use of a stimuli dimension depends on how well the dimension is learned. Our results show that there is an overall preference for a unidimensional categorization based on the perfectly diagnostic dimension. The preference for unidimensional categorization is negatively correlated with how well participants learn the partially diagnostic dimensions. Preference for unidimensional categorization is also negatively correlated with the mean response time. Bayesian modeling results show that participants use a partially diagnostic dimension only when it is learned with a very high level of accuracy. Different strategies are used for categorization depending on how well the perfectly and partially diagnostic dimensions are learned.

In unsupervised categorization, studies have shown that fewer stimuli dimensions are used for categorization with serialpresentation compared to concurrent presentation of stimuli. In this study, we investigate how a colour-based multidimen-sional descriptor might affect the number of dimensions used in categorization. Our results show that a fewer numberof dimensions are used when stimuli are presented serially irrespective of the presence of a colour-based descriptor. Wefound main effects for both the stimuli presentation mode and the colour-based descriptor. The stimuli has the same logicalstructure across all the conditions. Our results show that the notion of a natural and intuitive grouping of items is affectedby meta-level feature descriptors, that are not part of a feature-based representation of stimuli. We discuss the implica-tions of our findings for computational models of categorization, which make predictions based solely on feature-basedrepresentation of stimuli.