Causal status meets coherence: The explanatory role of causal models in categorization

Abstract

Research on causal-based categorization has found two competing effects: According to the causal-status hypothesis, people consider causally central features more than less central ones. In contrast, people often focus upon feature patterns that are coherent with the category’s causal model (coherence hypothesis). Following up on the proposal that categorization can be seen as inference to the best explanation (e.g., Murphy & Medin, 1985), we propose that causal models might serve different explanatory roles. First, a causal model can serve as an explanation why the prototype of a category is as it is. Second, a causal model can also serve as an explanation why an exemplar might deviate from the prototype. In an experiment, we manipulated whether typical or atypical features were linked by causal mechanism. We found a causal-status effect in the first case and a coherence effect in the latter one, suggesting both are faces of the same coin.