The Development of Induction within Natural Kind and Artifact Categories SUSANA.GELMAN

membership as the basis of their inductive inferences. The present studies ex-amine how children determine which category-based inductive inferences are warranted and which are unwarranted. Preschool and elementary school children learned various facts (e.g., “This apple has pectin inside”) and reported whether they thought the facts generalized to other items varying in similarity to the target (e.g., other apples, a banana, and a stereo). Categories included both natural kinds and artifacts and varied as to how similar category members were to one another (category homogeneity, as rated by adults). Results indicate that even the youngest children placed certain constraints on their inferences. However, the preschoolers made few principled distinctions among categories, basing their in-ferences primarily on category homogeneity. In contrast, older children made several distinctions that seemed based on domain-specific knowledge. Most im-portantly, they drew more inferences within natural kinds than within artifact categories, at times even overextending the distinction. Comparison with other research suggests that increasing scientific knowledge exerts powerful effects on patterns of induction within basic-level categories. 6 1988 Academic PESS, IIIC. This paper focuses on children’s inductive inferences and presents evi-dence that constraints on category-based inductions emerge early in de-velopment and undergo marked changes with age. Much of human rea-soning involves induction, or the process of making inferences that ex-This paper is based on a Ph.D. dissertation submitted to Stanford University. The re-search was supported by a National Science Foundation predoctoral fellowship and NICH