Can children generalise newly acquired object categories based on visual and haptic features?

Abstract

Although children constantly interact with the objects in their environment using vision and touch, our understanding of how these senses contribute to category formation is relatively poor. Children may find the complementary information about the relevant object features across vision and touch helpful for categorisation. Alternatively, cross-sensory information may increase the search space within which the relevant category features are encoded. The aim of the current experiment was to investigate children’s ability to learn and generalise object categories based on unimodal or bimodal information. Different groups of children aged from 4 to 13 years (N = 246) first learned to categorise novel objects using either vision, haptics or both modalities. Category learning was worst in bimodal conditions in younger children only, but modality had no effect on learning in older children. We then assessed categorisation performance to learned objects and generalisation to novel exemplars defined by a change of object features selected from either within or across categories. Relative to categorisation, generalisation was unaffected by within-category feature changes but reduced by cross-category feature changes. Our results suggest that children’s ability to generalise object categories is best for visual information. Moreover, we found no specific benefit for multisensory information on category learning or generalisation at any age. These findings provide insights into children’s ability to selectively attend and encode object features and suggest an independent influence of vision and touch in the formation of object categories during childhood.