It has been argued that some categories are privileged in that the correlational slructure of features in the environment creates natural chunks that are compelling to learners. If so, then differences in goals should not affect the organization and use of such so-called privileged categories. This paper reports research where the amount and type of expertise is varied for the same domain. Goals exert a major effect on category organization as well as on category-based induction. Even for the case where some categories seemed to be relatively privileged, the basis for this privilege may derive from linguistic cues rather than strictly empirical structure.