The effects of theories on children's acquisition of family-resemblance categories.

Two experiments examined whether 4- to 5-year-olds' acquisition of family-resemblance categories (for fictitious animals) was benefited by giving them a theory that explained the behaviors of category members ("fighter" versus "hider") in terms of the relations between functional surface features. As gauged by immediate and 24-hour-delayed categorization tests, children who performed theory-guided learning were more successful at making feature/category associations than those who performed similarity-guided learning. The Theory group categorized individual attributes significantly better than children for whom features of the training examples were identified simply by pointing and naming (Features condition, Experiment 1) or who were taught unrelated functions for features such that they could not be united within any obvious causal schema (Features Description condition, Experiment 2). These results support claims that mere similarity is insufficient to support category acquisition (Murphy & Medin, 1985) and show that theories giving an explanation for the existence of correlated properties can assist children to learn the distribution of perceptual attributes across categories.