The development of cooperative relationships: an experiment.

Pairs of individuals frequently face situations in which they could do well if they cooperated, but each risks being exploited. The Prisoner's Dilemma is widely used for investigating such scenarios, but it is framed in terms of cooperating and defecting, whereas in reality cooperation is rarely "all or nothing". Recent models allowing for variable investment in cooperation indicated the success of a strategy of "raising-the-stakes" (RTS), which invests minimally at first and then increases its investment if its partner matches it. We tested whether this strategy was adopted by subjects participating in an experiment in which they could choose how much money to give to a partner, reciprocity being encouraged by doubling donations. Subjects did increase their donations over successive rounds, both when playing against a stooge who reciprocated with the same investment, and when playing with a partner who was free to choose their investment. Subjects showed a strong tendency to match variations in their partner's investments. Cooperation was therefore achieved through a combination of initial escalation (RTS strategy) and quantitative responsiveness ("give-as-good-as-you-get" strategy). Although initial offers were higher than predicted, our results were broadly consistent with theoretical expectations.