Game Theoretic Investigation of Decision-Making and Theory of Mind in Neurotypical Individuals with Differential Levels of Autistic Traits

Theory of mind (ToM) is the cognitive ability to imagine the thoughts, beliefs, goals, and motivations of another person. This ability is utilized both consciously and subconsciously in the majority of social interaction in order to conduct rational, appropriate decision-making and form cooperative or competitive relationships to achieve one’s goals. Automatic, or implicit, ToM is a trait that has been shown to be impaired in individuals who suffer from autism spectrum disorders (ASD), which is one of the factors leading to the social deficits experienced by these individuals. The objectives of this dissertation are to 1) develop a non-verbal social task in which implicit ToM is evoked; 2) develop a computer agent that is capable of acting as a social partner within the task to evoke ToM response; 3) use this task in behavioral research to collect data from neurotypical individuals who engage socially with the computer agent. The first chapter introduces the Stag Hunt, a game theoretic task with a payoff matrix biased toward cooperation in which two players must decide whether to hunt a low payoff hare individually or attempt to cooperate to catch a high payoff stag. For the purposes of conducting the task with subjects, an adaptive agent was developed that initially enters the task naïve and develops in real time, based on the actions of the other player and the outcomes of repeated trials, a strategy suited to the subject it is interacting with. The second chapter discusses a behavioral study using a probabilistic variant of the Wisconsin Card Sorting Test in order to investigate cognitive biases inherent in decision-making in an uncertain environment. Understanding common coping mechanisms for uncertainty provides a baseline for comparing the decision-making strategies of atypical individuals. The third chapter expands upon the efforts of the first with the inclusion of variants of the preexisting adaptive agent in order to differentially evoke complex ToM responses correlated to the level of autistic traits in subjects drawn from the general population. This was accomplished through use of forward planning and simulated ToM. The fourth chapter discusses a pilot fMRI study to collect data on the correlation between levels of autistic traits and brain activation related to ToM and decision-making. Taken together, this body of work provides a foundation for utilizing non-verbal social tasks to evoke ToM response with non-human agents, a format that lends itself well to autism research. The overarching goal of this line of work is to aid in the identification of differences in neural processing of individuals affected with varying levels of ASD, both clinically and subclinically, to provide information that can be traced back to the locus of development in the brain leading to autistic expression