Expressing Emotions in the Iterated Prisoner’s Dilemma

Abstract

The mystery of cooperation has been researched from many perspectives and one explanation is that agents have fairness preferences where reciprocity matters. A key question in fairness literature is whether intentions matter and the modeling of fairness intentions turns out to require the tools of psychological game theory where payoffs depend not only on actions but also on beliefs. A problem in psychological game theory is that equilibrium analysis is difficult since beliefs are not typically observable ex post. I analyze how this affects behavior in the iterated prisoner´s dilemma, with agents that might care about fairness intentions, when one agent (Alice) cooperates and the other (Bob) defects. Alice thinks that Bob acted unfairly if he did so while believing that Alice would cooperate, but fairly if he did so while believing that she would also defect. In my analysis I define two types of players, a good one that cares about fairness intentions when choosing a strategy, and a bad one that only cares about himself and therefore always defects. Since Alice doesn’t know ex post what Bob’s belief was about her strategy, she has difficulties knowing his type and predicting his next move. This is a problem of asymmetric information and I look at the problem of signaling one’s type to the other using signaling theory where the central problem is that for signals to be believable, they must be costly to fake. Emotions have been proposed as commitment mechanisms that help with this problem in human societies. I present a way to solve this by allowing players to express emotions between iterations of prisoner’s dilemma. To do this I develop a theory of social affordance appraisal for agents, to argue that they can use emotional expressions as strategic actions in a social world.