Abstract
In many studies, human reasoning has been depicted as “biased” or deviating from normative models in both areas of deductive and inductive reasoning. Two criteria for evaluation of reasoning studies are proposed in this paper. The first criterion concerns the selection and application of normative models against which human performance is assessed. The second criterion concerns the role of conversational factors in the differential selection of information used in the subsequent judgment. These two criteria were applied to studies on two tasks from the fields of deductive and inductive reasoning, the Wason selection task and the base rate fallacy. In view of alternative normative systems, control of the application of the normative models used and the perceived conversational relevance of the information presented in the experimental tasks, the alleged biases have been demonstrated to be largely unsupported. The analysis is consistent with the notion that human reasoning is optimal and rational. The implications of the present account for reasoning experiments are discussed