Research on the discrimination of planometric (pattern) stimuli by monkeys has shown that changes which serve to bring the relevant cues closer to the edges of the stimuli facilitate the discrimination of those stimuli. Increasing the relative size of the relevant cue, for example, either by decreasing the total stimulus size while holding cue size constant (Riopelle, Wunderlich, & Francisco, 1958), or by increasing the size of the relevant cue while holding stimulus size constant ( Blazek & Harlow, 1955; Harlow, 1945; Warren, 1953b; Warren, 1 9 5 3 ~ ) has resulted in improved performance on both color and form discriminations. Reduction in errors has also been obtained by shifting the relevant cue from the center toward the periphery of the stimulus (Riopelle, et al., 1958) and by increasing the extent of differentially colored border (Warren, 1953c) with stimulus and cue size held constant. Schrier and Harlow (1957) have proposed that results such as these may be explained on the basis of contiguity of the relevant cue and reward, which is maximal when the edges of the stimuli are differentiated, and decreases as the width of undifferentiable border surrounding the relevant cues is increased. Their hypothesis, based on the results of smdies employing symmetrically placed cues, should predict that, in the case of asymmetrically placed cues, discrimination will be facilitated when the cue is placed nearer to the lower (toward S) edge of the stimulus, and will increase in difficulty as the cue is placed farther from that edge. Further, there is some evidence that performance improves with increases in absolute size when relative size is controlled (Harlow, 1951; Schrier & Harlow, 1957), so that, under the contiguity hypothesis, if distance from the edge is held constant, large cues should be discriminated more easily than smaller cues. The testing of rhese predictions formed one of the primary objectives of the present experiment. The reinforcement variable has been studied by comparing performance following selection of the correct (rewarded) stimulus with performance following selection of the incorrect (nonrewarded) stimulus. Two methods of stimulus presentation have been used in these studies. One method (single-stimulus technique) has been to present one stimulus alone for one or two trials, with both stimuli presented on subsequent trials. The second method (paired-stimulus technique) has been to present both discriminanda on all trials. With objects, initial nonreward is, in general, followed by fewer errors whichever