Prisoner's Dilemma
Prisoner's Dilemma: Two rational actors each have incentives to defect (betray), even though mutual cooperation would produce better outcomes for both. Individual rationality produces collective irrationality. The paradigm case for why self-interest undermines cooperation β and why institutions, reputation, repeated interaction, and binding agreements are necessary to achieve cooperative outcomes.
What Is the Prisoner's Dilemma?β
The Prisoner's Dilemma was formulated by Merrill Flood and Melvin Dresher at RAND Corporation in 1950 and given its famous name by Albert Tucker. The scenario: two suspects are held separately and cannot communicate. Each can cooperate (stay silent) or defect (betray the other).
The payoff matrix:
- Both cooperate: each gets 1 year in prison
- Both defect: each gets 3 years
- One defects, one cooperates: defector goes free (0 years), cooperator gets 5 years
Regardless of what the other prisoner does, each is individually better off defecting. If the other cooperates: defecting gives 0 years vs. 1 year for cooperating. If the other defects: defecting gives 3 years vs. 5 years for cooperating. Defection dominates β it's the better choice in every scenario. But when both defect (3 years each), the outcome is worse for both than if both had cooperated (1 year each).
This is the dilemma: individual rationality produces collective irrationality.
How It Worksβ
The payoff matrix (years in prison):
Other cooperates Other defects
You cooperate: 1 year 5 years
You defect: 0 years 3 years
The dominant strategy:
β If other cooperates: defect (0 < 1 year)
β If other defects: defect (3 < 5 years)
β Therefore: defect regardless of what the other does
The dilemma:
β Both defect β (3, 3) β the Nash Equilibrium
β Both cooperate β (1, 1) β better for both, but unstable
β Neither party has incentive to unilaterally switch to cooperation
Mechanisms that produce cooperation:
β Repeated interaction: defection today risks retaliation in future rounds
β Communication: parties can coordinate on cooperative strategy
β Reputation: defection is known to future potential partners
β Binding agreements: external enforcement makes defection costly
β Social norms: social sanction for defection
Three Real-World Examplesβ
Corporate Price Warsβ
Two airlines on the same route both benefit more from maintaining prices than undercutting each other. But each has an individual incentive to undercut: if the competitor maintains prices, you capture market share; if they undercut, you must match to stay competitive. Both undercutting (mutual defection) leaves both worse off β with lower margins and the same relative market position. The Prisoner's Dilemma structure explains why price wars are difficult to exit even when both parties recognise they're destructive.
International Climate Agreementsβ
Each country benefits from global emissions reduction but has an individual incentive to free-ride: if others reduce emissions, you benefit without the cost; if others don't reduce, your reduction alone is insufficient to prevent harm. The result: countries underinvest in emissions reduction relative to the globally optimal level. The Paris Agreement attempts to escape the dilemma through mechanisms β public commitments, transparency requirements, diplomatic reputation costs β that make defection more costly.
Patent Races and R&D Duplicationβ
Two pharmaceutical companies race to develop the same drug, each investing heavily despite knowing the other is doing the same. If they cooperated (shared research), they could achieve the same result at half the cost. But sharing risks giving the other party a competitive advantage. Both investing independently (mutual defection) produces duplicated effort and wasted resources β worse for both than coordination would be.
When to Apply Itβ
β Recognise Prisoner's Dilemma structure when:
- Each party has an individually rational incentive to act in ways that harm collective outcomes
- The outcome would be better for everyone if all cooperated but each party has reason not to
- You're designing incentive systems, policies, or institutions that need to produce cooperation
β The dilemma doesn't apply when:
- One party's interest genuinely conflicts with the other's (true zero-sum)
- The interaction is purely competitive with no mutually better cooperative outcome
- There are no strategic interdependencies between the parties
| Pairs well with | Why |
|---|---|
| Nash Equilibrium | Nash Equilibrium is the formal description of the dilemma's stable outcome |
| Tragedy of Commons | Multi-player version of the Prisoner's Dilemma |
| Incentive Theory | Changing incentives changes the Prisoner's Dilemma structure |
| Reciprocity | Reciprocity norms are a natural mechanism for sustaining cooperation |
Common Misuses and Limitationsβ
Treating all conflict as a Prisoner's Dilemma. The dilemma's specific structure requires that mutual cooperation is better for both parties than mutual defection. Some situations are genuinely zero-sum β one party's gain is exactly another's loss β and the Prisoner's Dilemma framing doesn't apply.
Assuming the dilemma is inescapable. Real-world repeated interactions, reputation mechanisms, communication, and institutional design regularly produce cooperation in situations with Prisoner's Dilemma structure. The dilemma shows why cooperation is difficult without these mechanisms, not that cooperation is impossible.
Ignoring asymmetric payoffs. The standard formulation assumes symmetric payoffs. Real situations often have asymmetric payoffs where the gains and costs of cooperation and defection differ significantly between parties β which changes the cooperative stability analysis considerably.
Related Modelsβ
| Model | Relationship |
|---|---|
| Nash Equilibrium | Nash Equilibrium describes the dilemma's stable (but suboptimal) outcome |
| Tragedy of Commons | Multi-player Prisoner's Dilemma with shared resources |
| Incentive Theory | Changing incentive structure can transform the dilemma |
Frequently Asked Questionsβ
What is "tit for tat" and why is it effective?
Tit for tat is a strategy for repeated Prisoner's Dilemmas: cooperate on the first round, then mirror whatever the other player did in the last round. In Robert Axelrod's famous tournaments (1980s), tit for tat consistently outperformed more sophisticated strategies because it: (1) starts cooperatively (nice); (2) immediately punishes defection (retaliatory); (3) immediately returns to cooperation when the other does (forgiving); (4) is simple and legible (clear). These properties β niceness, retaliation, forgiveness, clarity β are the foundations of sustainable cooperation in repeated interactions.
How does communication change the Prisoner's Dilemma?
In the classic one-shot dilemma, communication doesn't help β both parties can promise to cooperate, but each still has an incentive to defect after the other has committed. However, in repeated dilemmas, communication enables coordination on cooperative strategies, shared norm-setting, and the establishment of trust. This is why international agreements, business contracts, and institutional frameworks β all forms of structured communication and commitment β are so valuable in enabling cooperation in situations with Prisoner's Dilemma structure.
Is the Prisoner's Dilemma played in real life exactly as described?
Experiments consistently find that people cooperate more than pure game theory predicts β roughly 40β60% of people cooperate in one-shot anonymous Prisoner's Dilemma games, compared to the 0% prediction. This is because people have social preferences (altruism, fairness concerns, reciprocity norms) that modify pure self-interest. The dilemma is a useful idealisation, not a literal description of how people behave. Understanding both the model and the deviations from it produces more accurate analysis.
Further Readingβ
- Axelrod, R. (1984). The Evolution of Cooperation β how cooperation emerges from self-interest in repeated dilemmas
- Rapoport, A. & Chammah, A. (1965). Prisoner's Dilemma β the original academic treatment
- Poundstone, W. (1992). Prisoner's Dilemma β accessible history and applications
Apply with AIβ
π Analyse a Prisoner's Dilemma situation with MindMax β
This page is part of the MindMax Mental Models Knowledge Base.