Our knowledge of the physical objects of our world includes the categories exemplified by different objects and the relations among categories of objects. This knowledge permits us to make inferences about novel objects of a familiar category (a previously unknown dog, Rover) as well as novel kinds of object categories that include the object at issue (Rover is a dog, a vertebrate, etc.); and it permits us to deduce properties of the object at issue (Rover barks, has a backbone, etc.). Knowledge of categories related to the category at issue (topis are a kind of animal) permits one to deduce types of characteristic properties of members of the category at issue (topis have a characteristic diet) and to induce the specific values of these properties from a limited sample (topis are herbivores). The analysis of valid inductive inferences by the philosopher Nelson Goodman provides and account of the development of the ability to make inductive inferences. One premise of this position is that the more a person has attributed properties to a class, be it a class of individuals or a class of kinds (such as kinds of animals), the more readily this class acts as the range of an inductive inference. This position also predicts a facilitation of induction over one class as the result of making inductive inferences about contrasting classes. This account is to be tested in experiments with children. The studies are concerned with preconditions for induction, the relative effectiveness of different categories in constraining inductive inferences, and the experiences that facilitate induction. While the proposed research deals with physical objects such as non-human animals, plants, and artifacts, the findings are relevant to the development of children's reasoning about other types of entities.
