Making causal inferences about illness, compared to making causal inferences about mechanical breakdown and reading causally unconnected sentences, activates a semantic brain network implicated in the conceptual representation of animate entities (e.g. people, animals).