Children and adults selectively generalize mechanistic knowledge.

A central component of evaluating others as sources of information involves estimating how much they know about different domains: one might be quite knowledgeable about a certain domain (e.g., clocks), but relatively ignorant about another (e.g., birds). Estimating one's domain knowledge often involves making inferences from specific instances or demonstrations, with some suggesting broader knowledge than others. For instance, an American who demonstrates knowledge of an unfamiliar country like Djibouti likely knows more about geography as a whole compared to an American who demonstrates knowledge of a more familiar country like Canada. The current studies investigate the extent to which one potentially salient kind of knowledge - mechanistic knowledge - signals greater domain knowledge as a whole. Across four developmental studies, we find that both adults and children as young as six think that those who possess mechanistic knowledge about a basic level artifact category (e.g., clocks) are more knowledgeable about its superordinate level category (e.g., machines) than those with factual non-mechanistic knowledge (Studies 1a and 2a). We also find an analogous, yet delayed pattern with biological categories (Studies 1b and 2b). Together, these studies demonstrate that even young children, who possess little mechanistic knowledge themselves, nevertheless have a sophisticated sense of how knowledge of mechanism generalizes across related categories.