Sz~mmary.-Two groups of squirrel monkeys, Saimiri sciureris, were trained in learning set (LS) ro determine the effect of making the stimulus objects' rightleft position assignments contingent upon Ss' responses. In the first group, objects were assigned according to a random sequence, but in the second group the position of objects was conticgent upon the nature of the preceding response. For this latter group, the right-left position assignments changed on each trial that followed a correct and reinforced trial; on all other trials the assignments remained unchanged. In subsequert test problems in which position assignments for both groups were on a random basis, the group that had its initial training in the response-contingent LS condition was significantly superior ( F = 16.7, 9 < .01).Results were interpreted in accord with the logic that the response-contingent method served to efficiently suppress the differential cue error factor.In discrimination learning set (LS) experimencs, the right-lefc position assignments for stimulus objects typically follow some predetermined sequence unrelated to Ss' responses. The purpose of che present experiment was co study the effects of making such assignmencs systematically contingent upon S's response on the preceding trial, excludicg, of course, Trial 1.According to Harlow ( 1959), LS is acquired through the ordered suppression of certain inappropriate reaction tendencies or error factors (EFs), identified as stirn~~lus perseveration, differential cue, response shift, and position preference. Differential cue error, of particular relevance to the present srudy, refers co the concurrent, hence ambiguous, reinforcement of boch the position and stimulus chosen on any given trial (Spence, 1936). Braun, Patton, and Barnes ( 1952), Davis, McDowell, andThorson ( 1953), andHarlow ( 1950) have shown that EF is a function of the number of trials preceding the differential cue trial ( D C T ) , the first trial on which che stimuli change their positions. Ostensibly, then, assigning objecc positions in such a way as to minimize the number of s~~ccessive trials on which ambiguous reinforcement occurs should lead to a faster rate of error suppression and, consequently, to Inore rapid LS formation.T h e following experiment was conducted to test this hypothesis. T w o groups of Ss were compare;. One group received conventional training in which the objeccs' right-left positioning followed a random sequence. The second group received response-contingent training defined as the systematic variation of object position from crial to trial in a manner contingent upon the response made by S