Six experiments investigated the effects of background knowledge on category construction. In category construction tasks, subjects are commonly given a set of instances and asked to sort the instances into two groups. Although natural categories seem to have a family resemblance (FR) structure, subjects rarely construct FR categories in the sorting task. There is some evidence that subjects will construct FR categories when the features of the instances can be integrated into a coherent, meaningful category, if the subjects are provided with the theme that unifies the category. In the current experiments, the instances comprised two categories such that all members of a category either shared a common theme or did not share a common theme. All the categories shared the same formal structure, whether they were thematic or nonthematic. The experiments showed that the subjects were able to use their background knowledge to identify the themes underlying the categories. That is, the subjects did not need to be told the themes. When subjects were given thematic materials, FR category construction was the norm. However, when the subjects were given nonthematic materials, FR category construction was extremely rare. The discussion focused on the relation of these experiments to other work in category construction and the use of background knowledge in general.
Stored only when both papers are in the corpus — 116 citations exist upstream.