The use of category and similarity information in limiting hypotheses.

Previous studies have shown that adults usually select diverse evidence for generalizing hypotheses. We investigated what kind of evidence people select in another type of inductive task: limiting hypotheses. Whereas generalizing entails extending a property to all members of a category, limiting entails restricting a property to only members of a category. For example, if you know that elephants have property X, which would you test to determine whether only mammals have property X: a hippopotamus, a fox, a crocodile, or a snake? Both category and similarity information are relevant to selecting evidence to generalize or limit a hypothesis. Although 70% of participants chose diverse evidence for generalizing a hypothesis, only 25% chose the analogous evidence for limiting one. However, the percentage of participants choosing the appropriate evidence for limiting increased to 70% when they were given a rule for category membership. These results suggest that hypothesis-limiting behavior is affected by how participants establish category boundaries.