Classification of color dimensions in multiple contexts

Hue, saturation, and brightness were tested in pairs, with one dimension creating a classification context and the other serving as the basis of a set of speeded classification tasks. In Experiment 1, performance suffered when the context-setting dimension varied within a response category (intraclass context) and benefited when this variation occurred between response categories (redundant context). In Experiment 2, participants showed no interference from irrelevant variation in tasks that combined intraclass context with redundant context. Experiment 3 opened the interpretative window by varying 5 levels of intraclass context factorially with 5 levels of redundant context. The dimensions were found to differ in hardness--the degree of resistance to intraclass context across levels--with hue showing the greatest resistance and brightness the least. Hardness may reflect the efficacy of a dimension for distinguishing real-world categories.