抄録
Induction is called category-based induction if its premises and conclusion are of the form “All members of a category C have property P.” Osherson, Smith, Wilkie, Lopez, and Shafir (1990) advanced similarity-coverage model. But their data from a quantitative test of specific induction did not necessarily support the model. Three experiments investigate whether their modeling of similarity and coverage was not valid or their experiments were not valid. The following results were obtained. Coverage did not affect confirmation judgement in premisevariation type induction which Osherson et al. (1990) dealt with, and coverage effect was verified in conclusionvariation type induction which is more everyday and needs less processing load. In the latter case, coverage had to be weighted by typicality of conclusion in an inclusive category. And when mixed inductions were included, confirmation judgement was based not on coverage but on a belonging relation to an upper category and similarity between premise categories and a conclusion category.
