Recognition of common object-based categories found in toddler’s everyday object naming contexts

Abstract

Previously, we investigated the distribution of instances of early-learned object-based categories in toddler’s realistic everyday learning episodes; we found important differences in terms of frequency and variability (3D vs. 2D; real object vs. realistic toy vs. simple shape). Using a picture book task we tested 24-36 month olds’ recognition of these categories in four conditions: Realistic; Features (only parts of the photo visible); Silhouettes; and Geons (a shape caricature version made with only 3-4 parts and no color or texture). Results show similar recognition for all Realistic and Silhouette versions; Geons were lower than the first two; and Features had the lowest recognition rate. Critically, categories with the highest variability in our previous study were readily recognized by Features but difficult to recognize in Geon version. These results suggest that abstracting global shape is influenced by the specific trajectory of experienced exemplars.