Classical game theory assumes that players have a coherent model of th
e game in which they are engaged, that they make optimum plans assumin
g that everyone else makes optimum plans, and that all of this is comm
on knowledge. Evolutionary game theory postulates instead that players
have limited understanding of their environment, only modest reasonin
g ability, and no common knowledge. Surprisingly, a fair amount of cla
ssical solution theory survives with this change of perspective. We sh
ow how high-rationality solutions can emerge in low-rationality enviro
nments provided the evolutionary process has sufficient time to unfold
. (C) 1998 Elsevier Science B.V. All rights reserved.