Abstract
Both attractive and repulsive biases have been reported in delayed matching paradigms across perceptual domains. Here we ask whether the task structure influences biases in delayed color matching. Methods. Five observers compared the color of stimuli shown briefly in succession across a 1s delay. In a block of trials, five equally spaced reference stimuli covered a 24-degree-wide hue range in CIELAB space. Two different hue ranges were used in separate blocks. Two tasks were employed to measure short-term memory biases for color. In the 2IFC task, observers responded whether the second stimulus in a trial (the test) was bluer or yellower than the first stimulus (the reference). 1-up-1-down staircases controlled the test hue. In the modified, delayed match-to-sample task, the first interval contained the reference and the second interval contained two test stimuli. Participants responded which test stimulus appeared more similar in hue to the reference. The hue difference between the two tests was fixed at 6 CIELAB units and both were controlled with a single 1-1 staircase. In both tasks, point of subjective equality for each reference was defined as the average of last five reversals of the staircases. Results. The 2IFC task produced a robust central tendency bias: the reference hues tended to shift towards the mean reference in memory. Surprisingly, there was generally no central tendency bias for the delayed match-to-sample task; in fact, some observers showed a repulsion away from the mean color. Conclusion. Whether observers show attractive or repulsive effects in delayed comparison tasks systematically depends on the task. This difference may arise from the task wording affecting how the stimulus is encoded into memory, any encoding differences being exacerbated by the memory delay.
Meeting abstract presented at VSS 2018