Affiliations: Department of Computer Science, Southern Illinois University, Carbondale, IL 62901, USA | Department of Computer Science and Computer Engineering, University of Arkansas, Fayetteville, AR 72701, USA
Abstract: This paper explores delegation decisions predicated on models of trust and autonomy among agents. In socially rich environments, trust and autonomy of artificial agents are key attributes for rational delegation decisions. Social agents are affected by many social attributes such as benevolence, social exchanges, power, and norms. We present cognitively inspired working models of trust and autonomy for delegation among agents that accounts for these social attributes. These models are validated in a simulation of a group of agents who delegate tasks in order to locate the most socially well-suited individuals for performing that task. We define autonomy in such an objective manner that agents who have the highest autonomy with respect to a task will generally be the most socially well suited. Benevolence, social reciprocation, and norms most directly contribute to autonomy and proportionally to social suitability. Derived social power directly contributes to social suitability, whereas independently assigned social power ranking overrides the suitability. We show that trust and autonomy in an agent group changes under various social settings.
Keywords: Social norms, delegation theory, autonomy, trust, social influence, artificial societies, social simulation