Abstract
There are two conflicting views about the nature of thought: it is invariably rational or invariably irrational. Bartlett argued that thinking is a high level skill, and this idea suggests an obvious third possibility: thought is sometimes rational and sometimes irrational. This view is defended in the present paper, which argues that the doctrine of logical infallibility is either falsified by the results of some experiments on syllogistic reasoning or else empirically vacuous. There is no need to postulate a mental logic of the sort that Piaget and others have proposed. The rapid implicit inferences of daily life depend on the ability to interpret sentences by constructing mental models of the states of affairs that they describe. Deliberate deductions depend on the further ability to search for alternative models that violate putative conclusions. All that you need to know to assess validity is the fundamental semantic principle of deduction: an inference is valid if, and only if, its conclusion is true in every situation in which its premises are true and there is no way of interpreting the premises so as to render the conclusion false. This principle guides the construction of all logics though it is not explicitly stated in any of them. The paper concludes by examining the ways in which people differ in their ability to reason, the practical need to improve this ability, and some of the pedagogical implications of the present studies.