Abstract

A variant of psychological essentialism referred to as the causal-essentialism hypothesis was tested. According to the causal-essentialism hypothesis, categories with a single attribute causally responsible for most other category attributes will exhibit essentialist classification, stronger generalization of novel properties, and ease of learning. Common-cause categories exhibited neither stronger generalization nor ease of learning, but did exhibit essentialist classification. However, common-effect categories in which a single attribute was caused by most other category attributes also exhibited essentialist classification, supporting the conclusion that attributes become essential to category membership to the extent they participate in many causal relationships. Equally important to category memberships was the confirmation and violation of pairwise causal relationships, and a substantial minority of experimental participants engaged in complex causal reasoning strategies while categorizing. Accordingly, the role of essential attributes in category research is questioned, and a new emphasis on the role of configurations of attributes that collectively confirm or disconfirm causal knowledge is called for