Prisoner’s Dilemma Definition
Beyond any doubt, Prisoner’s Dilemma is the best-known situation in which self-interest and collective interest are at odds. The situation derives its name from the classic anecdote about two prisoners who were accused of robbing a bank. In this anecdote, the district attorney, unable to prove that the prisoners were guilty, created a dilemma in an attempt to motivate the prisoners to confess to the crime. The prisoners were put in separate rooms, where each prisoner was to make a choice: to confess or not to confess.
The attorney sought to make confessing tempting to the prisoners by creating a situation in which the sentence was determined not only by their own confessing or not but also by the fellow prisoner’s confessing or not. Yet irrespective of the fellow prisoner’s choice, the choice to confess yielded a better outcome (or less worse outcome) than did the choice not to confess. Specifically, when the other confessed, confessing yielded only an 8-year sentence, whereas not confessing yielded a 10-year sentence. And when the other did not confess, confessing yielded only a 3-month sentence, whereas not confessing yielded a 1-year sentence. So, from this perspective, it seems rational for each prisoner to confess to the crime. However, the crux of the dilemma is that the outcome following from both confessing (an 8-year sentence) is worse than the outcome following from both not confessing (a 1-year sentence). Thus, if both prisoners were completely trusting of each other, and strongly committed to supporting or helping each other, neither would confess, despite the attorney’s attempt to make confessing attractive. (The four possible sentences following from both prisoners’ choices are derived from R. Duncan Luce and Howard Raiffa; some other sources report slightly different sentences.)
The Single-Trial Prisoner’s Dilemma
This classic Prisoner’s Dilemma describes a situation in which the prisoners were to make their choices simultaneously, irrevocably (i.e., they could not undo or take back their choices), and therefore independently of one another. The independence of their choices was also ensured by putting the prisoners in separate cells, thereby excluding any possibility for communication relevant to the choices that they were going to make. In doing so, the attorney created a rather uncommon situation because people are usually able to interact in ways that permit them to respond to each other’s behavior or communicate about their choices.
Nevertheless, some situations that people encounter in real life resemble aspects of the classic Prisoner’s Dilemma. For example, it occasionally may be tempting to prepare less than fully for a working meeting with a partner to save time and energy for another activity that is more pressing or interesting. Yet, the meeting would be more fruitful if both partners invest time and effort and prepare well for the meeting. More generally, the Prisoner’s Dilemma represents exchange situations, which in the real world often occur under more flexible conditions, where both partners make choices in turn and every now and then can undo their choices. An example is the exchange of baseball cards, or cards of well-known soccer players, where two children can, at a little cost, provide each other with the card the other desires very much (e.g., to own the last card that completes one’s set of cards). In that sense, the Prisoner’s Dilemma represents a situation in which people “do business,” exchanging money, products, or services, that is more desirable to the other than to the self.
Researchers often use the single-trial Prisoner’s Dilemma when they want to study how people approach one another in the absence of a history of interaction and in the absence of a future of interaction. Hence, these choices are not influenced by considerations regarding the past (e.g., retaliation) or the future (e.g., adopting a strategy so as to obtain mutual cooperation). In these situations, impressions of the other play a very important role. In particular, any information that is relevant to one’s expectations regarding the other’s probable choice is useful, at least when one’s own choice depends on what the other is going to choose. For example, people expect much more cooperation from another perceived as honest than from another perceived as dishonest. Also, people may also derive expectations from stereotypical information. People expect more cooperation from a theology student than from student in economics or public administration.
More recently, it has been shown that choices can also be influenced in very subtle ways. In this research, participants typically first engage in a different task in which they unscramble sentences (putting scrambled sentences together in the correct order) that contain words having to do either with morality (e.g., honest, dishonest) or might (e.g., strong, weak). This task, or related task, seeks to activate morality-related concepts or might-related concepts—rather unconsciously. As it turns out, people are more likely to make a cooperative choice when morality was activated in such a task than when might was activated.
The Iterated Prisoner’s Dilemma
The Prisoner’s Dilemma has often been used to study repeated choices by which people respond to one another’s choices, a situation that captures interaction. Actually, most of the examples discussed so far illustrate the Prisoner’s Dilemma but do not perfectly match the features of a single-trial Prisoner’s Dilemma because there usually is a history or future of interaction that accompanies working meetings or exchanges of products (e.g., baseball cards). As such, single-trial interactions are more common in dealings with relative strangers rather than with partners, friends, or acquaintances. In contrast, the iterated Prisoner’s Dilemmas, characterized by repeated interaction, is more relevant to processes that shape people’s interactions with partners, friends, or acquaintances.
This research has focused on a variety of processes. One such process is the role of verbal communication. Often, cooperation can be enhanced if people are able to communicate before their choices in a Prisoner’s Dilemma. The more important question is, of course, how one can persuade the other to cooperate. Some research has compared the effectiveness of four messages: (1) “I will cooperate.” (2) “I would like you to cooperate.” (3) “If you don’t cooperate, then I will choose so that you can’t win.” (4) “If you now decide to cooperate and make a cooperative choice, I will cooperate.” This research has shown that a message that communicates conditional cooperation involving threats and promises tend to be somewhat more effective than those that do not incorporate such messages. These principles were subsequently used in designing strategies for building trust and resolving conflict, as well as further theorizing on these topics.
The iterated Prisoner’s Dilemma has also been used to examine the effectiveness of behavioral strategies. How should one behave if one seeks to obtain stable patterns of mutual cooperation? Or how can a person motivate, through his or her own behavior, the other person to make cooperative choices? Consider, for example, the tit-for-tat strategy that begins with a cooperative choice and subsequently imitates the partner’s previous choice. This strategy has been shown to yield greater outcomes than a 100% cooperative or 100% noncooperative strategy. Following early experiments examining this strategy, Robert Axelrod in 1984 organized a computer tournament in which several social and behavioral experts submitted programmed strategies that they believed would, when pitted against other possible programs, produce the highest outcomes. Each strategy then played against (or with) each other strategy. The interesting result was that tit for tat yielded far better outcomes for itself than did any of the other strategies.
An important feature accounting for tit for tat’s effectiveness is its niceness, in that the self is never first to make a noncooperative choice, and therefore cannot be perceived as exploitative or aggressive. Tit for tat is also effective because it is retaliatory: Non-cooperative behavior is responded to with a reciprocal noncooperative action. Furthermore, tit for tat is forgiving, in that noncooperative choices by the other in one situation are easily remedied in subsequent situations. Finally, tit for tat is also a clear strategy, readily understood by others, and indeed it tends to be experienced as directed toward establishing cooperation.
At the same time, tit for tat fails to initiate cooperation after there has been a lapse in it. Hence, a limitation of tit for tat is that it may give rise to the so-called echo effect (or negative reciprocity), that is, interaction patterns whereby the two persons are “trapped” in cycles of noncooperative responses. This limitation is especially important in situations characterized by noise—when there are discrepancies between intended and actual outcomes for an interaction partner because of unintended errors (e.g., not being able to respond to an e-mail because of a local network breakdown). In such situations, an unintended error may lead to misunderstanding (“why hasn’t he responded to my e-mail”) and eventually a noncooperative response (“I will make him wait as well”), which may instigate the echo effect. Indeed, some recent research indicates that some level of generosity might be important in overcoming the detrimental effects of such unintended errors. That is, when unintended errors are likely to occur with some regularity, strict forms of reciprocity will give rise to the echo effect, which can be prevented or overcome by adding a little generosity to reciprocity: that is, by consistently behaving a little more cooperatively than the other did in the previous interaction.
Moreover, the Prisoner’s Dilemma also often operates in situations involving more than two individuals (the so-called N-person Prisoner’s Dilemma; also referred to as social dilemma). For example, everyone enjoys clean public places, such as clean parks or sports stadiums. Yet people often find litter in such places, indicating that it is somewhat tempting to litter. As another example, whether or not to exercise restraint in the use of energy represents such a dilemma because overuse eventually leads to depletion of natural resources. In N-person Prisoner’s Dilemmas, threats or promises, or tit for tat, are generally less effective because there are so many people involved so that they are harder or even impossible to implement (to whom should I give tit for tat?). Typically, the level of cooperation is much lower in N-person Prisoner’s Dilemmas than in two-person Prisoner’s Dilemmas. Also different mechanisms tend to underlie behavior in N-person situations. For example, feelings of perceived efficacy, the feeling that one can make a difference and affect collective outcomes in a positive manner, feelings of personal responsibility (feeling responsible for a positive collective outcome), and feelings of identifiability (whether to feel anonymous or identifiable such that others can tell who cooperated and who did not) are all important ingredients of cooperation. These and other findings may be effectively used in public campaigns, which emphasize that people can make a difference (“all pieces help” to enhance perceived efficacy), or that people need to do so out of moral obligation or concern with the group (to enhance feelings of responsibility).
And finally, there are Prisoner’s Dilemmas between groups, or between representatives of two groups. Two companies may compete for the same clients, even though they both enjoy the public attention for their new products. Nations also often face such conflicts between their own group’s interest and both groups’ interests. Frequently, interactions between groups (or their representatives) are often less cooperative than are interactions between individuals. One reason is that groups do not tend to trust each other as much as individuals do, in that groups often rely on a scheme of distrust—which is not too surprising because groups often do compete in everyday life. A second reason is that group members tend to support their representative, and one another, for actions that serve the interest of their group (and themselves), but not that of the two groups together. The Prisoner’s Dilemmas of this sort have received relatively little attention, but may well be one of the most challenging to manage.
- Dawes, R. M. (1980). Social dilemmas. Annual Review of Psychology, 31, 169-193.
- Hamburger, H. (1979). Games as models of social phenomena. San Francisco: Freeman.
- Kelley, H. H., Holmes, J. W., Kerr, N. L., Reis, H. T., Rusbult, C. E., & Van Lange, P. A. M. (2003). An atlas of interpersonal situations. New York: Cambridge University Press.
- Pruitt, D. G., & Kimmel, M. J. (1977). Twenty years of experimental gaming: Critique, synthesis, and suggestions for the future. Annual Review of Psychology, 28, 363-392.
- Van Lange, P. A. M., Ouwerkerk, J., & Tazelaar, M. (2002). How to overcome the detrimental effects of noise in social interaction: The benefits of generosity. Journal of Personality and Social Psychology, 82, 768-780.