Testing hypotheses about mechanical devices

We use the literature on mechanical reasoning to derive predictions about how people will test a mechanical rule. In the presence of a single rule we predict significantly more selections of tests in which the hypothesized cause is manipulated than in the presence of two rules: the original and one casting doubt on the sufficiency of the hypothesized cause for the effect. We describe an experiment using Wason's selection task that confirms our predictions and go on to discuss the implications of our results for recent work on causal cognition.6 page(s