Subjects played strategically similar 4 x 4 and 6 x 6 constant sum gam
es under varying payoff scales. Substantial divergences from equilibri
um predictions were exhibited. The dynamic pattern of play is best exp
lained by a stimulus learning model whereby players allocate weight to
different actions according to their relative (time average) payoff e
xperience in past plays. The results do not provide much support for t
he hypothesis that players select best responses to beliefs about oppo
nent play based on observed choice frequencies in past plays, modified
by random errors or preference shocks. (C) 1997 Academic Press.