Proceedings of the Annual Meeting of the Cognitive Science Society

Causal reasoning is inherently temporal in nature; humans can infer whether one event is a possible cause of another, even from temporal descriptions. For example, the sentence, "after the book was placed, the shelf collapsed" implies that it's possible (but not necessary) that the book's placement caused the shelf to collapse, and that it's impossible that the shelf's collapse caused the book's placement. These modal inferences challenge existing accounts of causal reasoning. Two experiments reveal how humans infer possible causes from temporal descriptions. Experiment 1 showed that their possibility judgements depend on the event structures described by temporal relations. Experiment 2 varied the consistency of event descriptions and found that people inferred possible causes even from conflicting information. These data challenge extant theories of causal reasoning but corroborate the hypothesis that people comprehend causal relations by building mental simulations of event structures.