Categorizing with overlapping categories

Publisher Website
Google Scholar
Abstract
An important aspect of the categorization process is that an item can be assigned membership in more than one semantic category. Previous work examining how subjects decide an item’s membership in one of several alternative categories has most often used categories having a strict hierarchical relationship (e.g., bird-canary). Four experiments are reported that examine how subjects decide membership of simple pictorial stimuli in partially overlapping categories (e.g., high and very high). Experiment 1 was a rating task designed to identify items as members, nonmembers, or as falling on the fringes for several overlapping categories. In Experiments 2–4, this information was used to predict subjects’ mean reaction time in speeded categorization tasks using the same pictorial stimuli. Subjects interpreted the categories in one of two very different ways. According to the first interpretation, there was a strict set-subset relationship between categories such as “high” and “very high.” According to the second, the entailment relationship did not hold; membership in the category “very high” did not imply membership in “high.” Even when subjects used a set-subset interpretation of the category labels, their reaction times were affected by a form of semantic response competition. Subjects took longer to verify an item’s membership in a category when there was another more appropriate category descriptor for that item included in the experiment.

This publication has 12 references indexed in Scilit: