Language pragmatics is applied to analyse problem statements and instructions used in a few influential experimental tasks in the psychology of reasoning. This analysis aims to determine the interpretation of the task which the participant is likely to construct. It is applied to studies of deduction (where the interpretation of quantifiers and connectives is crucial) and to studies of inclusion judgment and probabilistic judgment. It is shown that the interpretation of the problem statements or even the representation of the task as a whole often turn out to differ from the experimenter's assumptions. This has serious consequences for the validity of these experimental results and therefore for the claims about human irrationality based on them.