The Induction-Deduction Asymmetry: Fact or artifact?

Five studies are reported that test the hypothesis that inductive inferences from behaviors to traits are performed more frequently than deductive inferences from traits to behaviors--a phenomenon referred to as the Induction-Deduction Asymmetry (I.D.A., Maass et al., 2001). In Exp. 1 (n = 320) participants read a recommendation letter that contained verb and adjective information about a job candidate. A free recall test shows that information originally presented in verb form was falsely remembered as adjectives whereas the opposite kind of memory distortion occurred much less frequently. In Exp. 2 (n = 52) participants received information about a target person, either presented in trait form (e.g. dominant) or in behavior form (e.g. dominates), with adjectives and verbs sharing the same word-stem. A free recall test confirmed the asymmetrical memory pattern of study 1. In Exp. 3 (n = 52), using the same-word stem paradigm, the verb tense (past vs. present) of verbs was varied. A recognition test showed that traits implied by a behavior were more frequently misidentified as already seen than behaviors implied by a trait, and this was true regardless of verb tense. The last two experiments demonstrate that I.D.A. occurs only when the information refers to a target individual but not when the exact same information is provided as a simple word list (Exp. 4, n = 40) or as referring to a non-human target (Exp. 5, n = 40). Together, these studies show that I.D.A. is a very robust phenomenon, but that it only occurs for information embedded in a social context.