Abstract:
Humans can categorize objects at different levels of abstraction: basic (for example, cars and chairs) and superordinate (for example, vehicles and furniture). Usually, basic-level categorization is acquired
faster than superordinate-level categorization, but the opposite results have been reported as well. One possible reason is that similarity between different basic-level categories influences speed of learning of these categories. We investigated how between-category similarity affects speed of
learning of basic-level discrimination and superordinate-level discrimination. In Experiment 1, we used four families of abstract shapes called Attneave's shapes to create four basic-level categories, each containing 8 exemplars of shapes. Participants were asked to rate similarities between the shape families shown on a computer monitor. We then analyzed the ratings using multidimensional scaling to evaluate similarities between four categories. For Experiment 2, we combined two similar basic-level categories in one superordinate-level category. Two dissimilar basic-level categories formed the second superordinate-level category. Participants then had to learn to make a correct response to a member of each category shown on computer monitor. The response had to be made either at the basic level (four choice alternatives) or at the superordinate level (two choice alternatives). We expected participants to learn basic-level discrimination faster when two categories are similar. When two categories
are dissimilar, we expected them to learn superordinate-level discrimination faster. Preliminary results of Experiment 2 will be presented.
