Human Factors and Behavioral Science: Retrospective Reports Reveal Differences in People's Reasoning

Reasoning for a class of transitive inference problems was studied and the following questions were experimentally investigated: (1) Can people give reliable retrospective reports about their reasoning processes? (2) Do people who report different reasoning processes actually reason in different ways? (3) Can people be trained to use different reasoning processes? In the situations studied, subjects' retrospective reports about reasoning contained sufficient information to classify the subjects reliably. Subjects classified as using different reasoning strategies made different amounts and different kinds of reasoning errors. As a result of training, subjects could use reasoning processes that they would not have used spontaneously. These results have implications for developing theories of reasoning and for assessing and modifying reasoning‐like processes in practical situations.