Non-optimal behavior versus cognitive illusion

Abstract

The issue of choice behavior under uncertainty is prone to controversy. Whereas in the cognitive sciences, the errors an individual can make are supposed to be product of an irrepressible illusion, research in the experimental analysis of behavior approaches non optimal behavior in terms of its acquisition. The latter perspective requires a thorough analysis of the conditions which support non optimal behavior by examining the history of verbal reinforcement of an individual and the concrete contingencies to which he or she is subjected. The present experiment stresses empirical contingencies and allows to evaluate the influence of the consequences of behavior on the subjects performance. The experiment appears as a game in which the subject must find a coin hidden under one of three cups. After a first choice, an empty and non selected cup is withdrawn in order to leave only two cups (one selected by the subject and another). The subject can then modify or maintain his initial choice. The great majority of the subjects, however, consider the two strategies to be equivalent, which constitutes non optimal behavior.  In order to identify the variables which maintain this behavior and test if it can be modified, the reinforcement rate for optimal behavior was increased by varying the number of cups at the beginning of the game: the profit probability of modifying one´s choice increases with the number of cups. Three independent groups were distributed among three experimental conditions (3.5 or 10 cups) and were confronted with experience by observation (the subject took part in the game 20 times). The results show that the more the 20 trials of free choice, which supports the hypothesis that non optimal behavior is the product of learning.