Attempts to Predict Success on an Insight Problem

Abstract

18 predictor variables were related to a pass-fail criterion of success on Maiet's Hatrack Problem (Maier, 1945), a problem requiring insight. The correlations of all 18 predictors with the criterion were uniformly low and insignificant, pointing up our inability to predict success on an insight problem. It is suggested that insight problems may require different abilities for their solution than “other” problems. The fact that consistent relationships fail to appear when insight problems are used as the experimental task further supports this suggestion. An explanation based on the assumption that blind trial and error is the type of mental process engaged in by a person when trying to solve an insight problem, was evaluated. The behavioral data indicated that this explanation was inadequate in explaining performance of Ss on this problem because Ss tended to approach the problem in a selective manner (persisted in directions), and increased time for search did not increase the number of correct solutions produced.