players have purely selfish preferences, this is the end of the story (and communication has no impact).

In the presence of reciprocal preferences, however, the common-pool resource game is no longer a prisoners’ dilemma (see Table 5-1b). The reason is that if both players are sufficiently reciprocally motivated, they don’t like to cheat on the other player. If, for example, player 1 chooses the low appropriation strategy, player 2 with reciprocal preferences is better off choosing the low instead of the high appropriation level and vice versa. Even though players forgo some material payoffs, they have a higher utility if they reciprocate the nice behavior of the other player. If player 1 chooses the high appropriation level, however, player 2 has no desire to choose the low level (neither if he is a selfish nor a reciprocal player). Instead, player 2 will in this case also play the high appropriation strategy. As a consequence, there are two (pure) equilibria now, the efficient equilibrium with low appropriations and the inefficient one with high appropriations. Put differently, the prisoners’ dilemma game in Table 5-1a with a unique and inefficient equilibrium has turned into a coordination game with one efficient and one inefficient equilibrium. Game theory does not help much in this situation. It simply predicts that some Nash equilibrium will be played, but not which one.16

In the presence of multiple equilibria, subjects face a tremendous strategic uncertainty. How shall a person know which strategy the other player will select? It is obvious that communication can have a positive impact in a situation of strategic uncertainty. In fact it has been shown experimentally that communication can help players to coordinate on better equilibria.17 As an example, take Cooper et al. (1992), who study different coordination games with and without communication. They find that, depending on the precise structure of the coordination game, communication may improve efficiency. This holds even though all announcements are nonbinding. They also show, however, that communication does not always improve coordination. The prospects for improved cooperation depend both on the nature of the game and the nature of the communication process.

In the common-pool resource experiments with communication mentioned earlier, players had intensive opportunities for communication because they could actually talk to each other and were (with some restrictions) allowed to discuss anything they wanted. As reported in Ostrom et al. (1994), subjects usually came

TABLE 5-1b A Simple CPR Game in the Presence of Reciprocal Preferences



Player 2




Low appropriation

High appropriation

Player 1

Low appropriation




High appropriation



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement