Two groups of squirrel monkeys were trained on discrimination learning-set problems in Phase I. The experimental group discriminated between differently colored edible stimuli which were also reinforcers. In each problem, one colored stimulus tasted sweet and one colored stimulus tasted bitter. Control subjects discriminated between differently colored stimulus blocks upon which identically colored edible reinforcers were placed. In Phase II, both groups were trained on the control task. During Phase I, experimental subjects evidenced significant learning while the control group did not. In Phase II, the experimental group performed significantly above chance, indicating positive transfer between the experimental and control tasks. The results, besides showing the efficiency of the ingested stimulus procedure, indicated a close association between correct and random responding during learning-set acquisition in primates.