THE EVOLUTION OF TRUST AND COOPERATION BETWEEN STRANGERS - A COMPUTATIONAL MODEL

Social and economic exchanges often occur between strangers who cannot rely on past behavior or the prospect of future interactions to estab lish mutual trust. Game theorists formalize this problem as a ''one-sh ot prisoner's dilemma'' and predict mutual noncooperation. Recent stud ies, however challenge this conclusion. If the game provides an option to emit (or to refuse to play), strategies bused on ''projection'' (o f a player's intentions) and ''detection'' (of the intentions of a str anger) can confer a ''cooperator's advantage.'' Yet previous research has nor found a way for these strategies to evolve from a random start or to recover from invasion by aggressive strategies that feign trust worthiness. We use computer simulation to show how trust and cooperati on between strangers can evolve without formal or informal social cont rols. The outcome decisively depends, however; on two structural condi tions: the payoff for refusing to play, and the embeddedness of intera ction. Effective norms for trusting strangers emerge locally, in excha nges between neighbors, and then diffuse through ''weak ties'' to outs iders.