An established, captive colony of 74 rhesus monkeys (Macapa mulatta) was group-tested on a simultaneous visual discrimination problem and three reversals of the initial discrimination. The task incorporated important aspects of rhesus foraging behavior. Although other studies of communal groups of nonhuman primates have reported rapid learning, subjects in the present study showed no evidence of one-trial acquisition of the initial problem or of rapid learning-set formation across the reversal series. Instead, mean and individual performance, on all variables measured, improved gradually, both within and across series. Subjects appeared to "learn how to learn," consistent with findings of traditional studies of individual discrimination learning. Our finding of gradual learning in group-living animals argues that the source of rapid learning in previous reports is not attributable to social influence or ecological relevance, but may reflect specific procedural or species differences between studies.The comparative study oflearning has used two broad approaches: a laboratory approach concerned with the mechanisms of learning, and an ecological approach concerned with functional significance. Studies of nonhuman primate discrimination learning have relied primarily on the former, investigating the behavior ofindividuals under the controlled and isolating conditions of a standardized apparatus. This traditional approach consistently has shown that performance gradually improves, within and across discrimination problems, as individuals either acquire problem-solving strategies or learn to successfully apply strategies to new problems (Harlow, 1944(Harlow, , 1949. This "learning how to learn," termed learning-set formation, is considered a fundamental characteristic of discrimination learning.The validity ofthis traditional approach has been questioned because it does not adequately duplicate ecological demands (