Analysis on Trust Game by Reciprocal Agents

Abstract

In this paper, the author proposes a game-theoretical model of trust among reciprocal agents. Our model, a trust game, is a non-cooperative game in extensive form. By considering about this game, we can define clearly the concept of trust behavior in general games in extensive form. But just using ordinary equilibrium concept (e.g. subgame perfect equilibrium), we cannot explain the trust behavior in some situations. This result contradicts with some observations in real world. So, we have to adopt another solution concept, sequential reciprocity equilibrium (SRE), which is suggested by Dufwenwerg. Adopting this SRE concept, we analyze repeated trust game (RTG). As a result of analysis, I find the condition of reciprocity to trust others, and reciprocal agents can get higher payoff than non-reciprocal agents when the length of game is enough long.