Learning optimal behavior through exposure to contingencies in the Monty Hall dilemma

Abstract

Judgment under uncertainty has generated many research in Cognitive Sciences where the erroneous judgment perceived as the product of mental modules unsuited to the situation and in Experimental Analysis of Behavior, where it is apprehended as a non-optimal behavior acquired by learning. The experiment consists of a game, the Monty Hall Dilemma, in which 3 goblets are presented turned over to a subject who must find a ball hidden under one of them. After a first choice, an empty and non selected goblet is withdrawn. The subject can then stick with his initial choice or switch to the other remaining goblet. The “Switch”  strategy and the “Stick” strategy have, respectively, a profit probability of 2/3 and 1/3, but, in great majority, the subject consider that the two strategies are equivalent, what constitutes a non optimal behavior. We formulate the hypothesis according to which repeated exposure to choice consequences will allow learning of the optimal behavior. Three conditions constitute the experiment: a condition consisting of a positive reinforcement procedure, a condition consisting of a negative punishment procedure, and a condition consisting of a positive reinforcement and negative punishment procedure. The addition of an aversive value when the choice of strategy does not make it possible to find the ball and the extension of the series of tries to 100 should allow an improvement of the learning effectiveness compared to the studies having previously employed series of reinforced tries in the Monty Hall Dilemma. Results show that optimal behavior rate increases with series of tests, but remains close to 50 %. No learning procedure appears more effective than another for acquisition of the choice for “Switch” strategy. The hypothesis according to which subjects choice behavior would be controlled by verbal rules making them insensitive with changing contingencies is formulated.