We use computer simulation to identify a process by which cooperation
evolves without iteration, and evolves better in large than in small s
ocieties. It is based on an empirically supported heuristic for decidi
ng whether to enter noniterated prisoner's dilemma games, namely, Expe
ct others to have the same dispositions as yourself. Players are assig
ned a probability of cooperating that also defines their expectations
about others' behavior and thus their willingness to play. The carryin
g capacity of the ecology is 10,000. Players multiply by 2 if their ag
gregate payoff in a given round (2) places them among the more success
ful 5,000 and (2) is move than zero. We find that the most adaptive di
sposition is toward the mean of the population. That is where individu
als have the optimal mix of consummated plays with more cooperative pl
ayers and unconsummated plays with less cooperative ones. When encount
ers occur by proximity, fortuitous clusters toward the cooperative tai
l will grow and dominate the society. Such clusters are more likely in
large societies.