The Role of Labels in Learning Statistically Dense and Statistically Sparse Categories

Abstract

Subjects were given classic category formation tasks with feedback. We used two types of categories—statistically dense and statistically sparse. We conducted four experiments to assess the influence of sign type (experiment 1) and the interference of redundant actions performed with the sign (experiment 2) on the performance of learning different types of categories. We found that in the case of dense category formation, the visual distinction of the sign from other object features is more important. In the case of sparse category formation, easy verbalization is more important. Additionally we showed that verbal interference, directed at the actions with the sign, improves sparse category formation, but worsens dense category formation. The results of our experiments are discussed in accordance with the Competition Between Verbal and Implicit Systems (COVIS) model of multiple systems of categorization.