The role of number of categories and conditions of their learning in formation of probabilistic categorization rules

In traditional studies of categorical learning, its conditions are usually limited to only two categories for which a categorization rule needs to be found. However, in practice, when learning any new category, people often use examples from other, additional categories as well. How does this affect the learning of the main category? In the present experiment, participants performed a task to form new categories. They had to learn a probabilistic rule, a prototype. In addition to the main task (distinguishing between examples from two main categories), participants were presented with examples from a third category. The conditions of presentation of this category were varied: existence or absence for the examples included in it oftheir own categorization rule, as well as feedback. After the formation of categories, participants performed a transfer test. It was found that the accuracy of learning basic categories was influenced by the existence of a categorization rule for additional category and was not influenced by the presence of feedback for it. However, when there was feedback, participants categorized new examples more quickly. The findings are discussed in the context ofthe multiple categorization systems model (COVIS) and studies of learning with partial feedback.