Prisoner's Dilemma

The prisoner's dilemma game specific to game theory is widely used to study human interactions from market exchanges and armament decisions to collective action problems. In this game, two players each have a single choice between two symmetrical actions: to cooperate or to defect. There are four possible outcomes: both players cooperate, both players defect, player A cooperates and B defects, or player B cooperates and A defects. The players' preference rankings are symmetrically inverse, with player A preferring, first, A's unilateral defection and B's unilateral cooperation; second, that both cooperate; third, that both defect; and fourth, that A unilaterally cooperates and B unilaterally defects.

The rudimentary logical structure of the prisoner's dilemma is usually embedded in a narrative explaining [Page 747]of what the various outcomes represent. Sometimes the two players are cast as co-conspirators caught by a jailor who wishes that each prisoner confesses to a crime. In this scenario, the payoff matrix is explained to the prisoners such that if both confess, they both get a moderate sentence; if neither confesses, then they both receive a light sentence; if only one confesses, then the other will walk free while the co-conspirator who confesses will stay in jail for life. This game structure demonstrates that, regardless of what the other person chooses to do, it is rational for the agent to choose to confess as walking free is superior to receiving a light sentence and receiving a moderate sentence is superior to receiving a life sentence. Thus, both rational agents select confessing rather than not confessing: In this case the two each achieve a mutually inferior outcome of receiving a moderate sentence instead of a light sentence.

This single-play version of the prisoner's dilemma game exhibits strict dominance for both players because, regardless of what course of action the other adopts, each player gains by the strategy of defecting from cooperation. Even though other outcomes are possible if the prisoner's dilemma was to be played repeatedly, the basic game form has received abundant attention among rational choice researchers for being paradigmatic of many types of human interactions, wherein motivations of coordinating actions to achieve a better mutual outcome reside side-by-side with motivations to better one's own condition at the price of potential collective impoverishment. Even the market, once viewed as reflecting individuals' mutual interest in exchange, is now cast as a prisoner's dilemma in which each would prefer to cheat the other. The free rider and collective action problems rely on the logic of the prisoner's dilemma to demonstrate that in group situations relying on voluntary contributions, each has the ever-present incentive to withdraw support, regardless of what others choose to do.

It is widely thought by rational choice researchers that the prisoner's dilemma is a ubiquitous feature of human society that cannot be resolved through voluntary agreement among community members. Instead it is proposed that external sanctions must be imposed to enforce voluntary compliance so that communities of individuals are able to achieve the rewards of cooperation instead of paying the price of mutual defection. These external sanctions resolve the prisoner's dilemma by altering the game's payoff structure, thereby transforming it into a different game.

...