Modification of non-optimal behavior in the Monty Hall Dilemma

Abstract

Over the past few decades, non-optimal behavior has been widely observed by researchers studying human reasoning. Cognitive psychologists have explored these phenomena and have provided many descriptions about such non optimal behavior. Most of their models conceive these maladaptive behaviors as an outcome of an unsuppressible cognitive illusion dependent upon the structure of the cognitive system. Instead of adopting a structural and static view, researchers in experimental analysis of behavior suggest that non optimal behavior can be explained in learning terms. In the same view, this article focuses on how non optimal behavior occurs (the conditions that generate it) in taking into account environmental contingencies combined with experimental history of an individual. The experiment attempt to show that behavioral approach allows to provide a best understanding about maladaptive behavior in a given situation known as a variant of the Monty Hall Dilemma. In this variant, subjects must find a coin hidden under 3 turned over goblets. After a first choice, one unselected and empty goblet is taked out. Then, subjects can preserve or modify their choice first choice. The chosen strategy “to modify” and “to preserve” permit to obtain coins according different probabilities (respectively 2/3 versus 1/3). In this case, the strategy “to modify” constitutes the optimal strategy to obtain coins. However, most subjects consider each strategy with the same profit probability (1/2 versus 1/”) that proves to be a non optimal behavior. To identify the conditions which permit to modify a such non optimal behavior, the experiment manipulate the number of goblets in 3 experimental conditions (3, 5 or 10 goblets). In more concrete terms, the experiment attempts to test if maladaptive behavior observed in the Monty Hall Dilemma can be evinced by controlling the rate of reinforcement combined with the optimal responding (which increases with the number of  goblets). The results show an increase of the optimal behavior according to the rate of reinforcement associate with the correct responding “modify” Thus, these experiments support behavioral assumption in demonstrating that non optimal behavior is dependent upon preexisting learning and consequently can be removed.