Bringing Empathy to Game Theory

IGNACIO VIGLIZZO; FERNANDO TOHMÉ

Tandil

Congreso; LX Reunión Anual de la Unión Matemática Argentina; 2010

Unión Matemática Argentina

A frequent criticism that game theory faces is the blind egotism that seems to drive the rational agents that the theory purportedly models. We will introduce here a new (as far as we know) tool that will allow us to model agents that care about the payoffs that the other players will receive. We will proceed by applying the new method to a well known and problematic example: the prisoner´s dilemma. Then we will draw some conclusions and will try next to generalize the method, while preserving its desirable features. Classically, a player only regards the information they have about the other player´s payoffs as a way of assessing their most likely actions, under the hypothesis that they will behave rationally: they will take the actions most likely to maximize their utility. We will keep this notion of rational agents but we will distinguish between payoffs for an outcome in a game and their (subjective) utility. This will in fact let us describe a space of types of players, one we hope is wide enough to explain a range of behaviors based on the simple single assumption that players care about each other´s payoff (either in a positive or negative fashion).