Different mechanisms may underlie the performance in relative localization tasks.

Two different groups of subjects had to adjust two-dimensional stimuli, differing in size, shape and type (dot patterns or irregular contour figures), within a reference circle. The two groups performed under two different instructions. The first instruction stressed matching the centres of the stimulus and the circle, while the second required simply positioning the test stimulus in the middle of the reference circle. In two control experiments the subjects had to determine the position of the centres of each stimulus and of the reference circle. Under the first instruction the accuracy of performance, estimated by the variance of the responses, depended on the stimulus size, shape and type in agreement with previous results and models of relative localization. Under the second instruction, however, accuracy remained invariant. Possible mechanisms of relative localization that might differ at their first stages of localization of the separate stimuli are considered.