Integration of Category Induction and Hierarchical Classification: One Paradigm at Two Levels of Complexity

Abstract

Hierarchical classification (HC) and category induction (CI) were tested by a common property inference procedure to facilitate comparison and enable relative complexities to be assessed. Relational complexity theory predicts that HC is more complex because it entails a ternary relation between categories B, A, and A' such that A and A' are included in B, whereas CI entails a simpler binary relation between a category and its complement. Experiment 1 tested inferences about familiar categories with plausible but unfamiliar attributes, and Experiment 2 assessed inferences about fictitious categories with familiar attributes. As predicted, HC was more difficult than CI. Children over 5 years succeeded on both, but 3-year-olds succeeded on CI only. Tasks of the same level of complexity predicted 68% (Experiment 1) and 80% (Experiment 2) of age-related variance. The results suggest that HC and CI may be regarded as 1 paradigm with 2 levels of structural complexity.