Gobet, F. (1992). Learned helplessness in chess players: The importance of task similarity and the role of skill. Psychological Research, 54, 38-43.


Several variables that might influence performance within the learned helplessness context have been identified, such as importance of task, causal attributions, and the amount of experience with uncontrollable outcomes. This study investigates the effects of noncontingency between subjects' responses and outcomes with respect to two other theoretically and practically important variables: treatment/posttest similarity and skill in the domain.

It is argued that three characteristics impede the external validity of most learned-helplessness studies in humans: (a) restricted population (mainly college students or depressed people); (b) artificial tasks, and therefore lack of subjects' motivation; and (c) lack of assessment of subjects' skill in the domain. The use of chess players as subjects is proposed as one way to remedy these deficits.

The experimental design consisted of three balanced groups. The first group had to solve 14 chess problems with objective solutions and received veridical feedback; each member of the second group faced 14 problems with no objective solutions, and received the same feedback as the member of the first group he was yoked with, but without any control on it; the control group received a waiting task. Subjects' performance was then assessed in two posttests, one similar to the treatment, the other consisting of a different task, but still related to chess. Moreover, subjects had to respond to a self-report inventory at the beginning and the end of the experiment. The subjects (N=48) were Swiss adult male chess players, whose skill ranged from average club player to International Master.

It was found that the group with unsolvable problems performed worse and was more depressed at the end of the experiment than the two other groups. The mid-strength players were the most sensitive to the manipulation, and the weakest players showed little effect of learned helplessness. It was also found that the effects were proportional to the degree of similarity between the treatment and the posttest. The results suggest limits to the domain of applicability of the learned helplessness model. It is proposed that the conditions for the development of learned helplessness require (a) a high similarity between the noncontingency learning situation and the following situations and (b) average skill in the domain.