|M.Sc Student||Arshavsky Olga|
|Subject||Choice Convergence in Decision Making: Studies of Healthy|
Adolescents and Adolescents with Pervasive
|Department||Department of Industrial Engineering and Management||Supervisor||Professor Eldad Yechiam|
|Full Thesis text|
Recent studies have recorded a tendency of individuals with Autism Spectrum Disorders (ASD) to continually change their choices in repeated-choice tasks, displaying low choice convergence - the magnitude of repeated choices from an alternative which maximizes personal benefit. The present research examines this construct as a manifestation of the cognitive style of ASD, its possible sources, boundaries, and relation to other constructs such as indecisiveness and trait anxiety in an explorative manner. Study 1 examined the stability of choice convergence across different decision tasks (differing in general task and payoff structure, and number of alternatives), in order to establish it as a measure of cognitive style. Sixty-two undergraduate students completed seven repeated-choice tasks: the Iowa Gambling Task (IGT), the Bangor Gambling Task (BGT) and five Two-Buttons (TB) tasks. Choice convergence showed sufficient stability across tasks, yet it also showed meaningful variations: significant positive correlations were found between the tasks, some which among the easier tasks were weakened by controlling for performance (the proportion of choice from the higher yielding alternative). The BGT was the most weakly related to the others tasks. In Study 2, these tasks and intellectual aptitude measures were administered to 17 adolescents with ASD and 28 matched controls. Two competing hypotheses were examined regarding the origins of ASD's choice pattern: 1) Inflexible processing of loss information in tasks which combine gain and loss in the outcomes of all alternatives (e.g. IGT) and possibly increased regret over experienced losses, which leads to attempts to avoid losses altogether and thus to low convergence; 2) erratic choice process regardless the valence of payoffs. These hypotheses were not supported, as although ASD have shown lower convergence in IGT, the cognitive modeling analysis indicated that their choices were guided not so much by the values of alternatives as by other features, such as their novelty. ASD showed significantly higher convergence in BGT, yet on the risky and potentially more lucrative alternative, in which losses were increasingly more frequent. Therefore, their behavior in this task reflected inflexible ‘chasing’ of gains (supported by cognitive modeling analysis). Finally, as there were no group differences in the TB tasks, the decision making pattern of ASD cannot be attributed to an uncontrollable tendency to shift choice, and therefore more complex tasks such as IGT constitute the boundary of this phenomenon. The implications of these results to cognitive theories of autism as well as treatment approaches are discussed.