Abstract
We measured motion coherence and contrast thresholds for determining the direction of motion of a random dot pattern. Dots within a “target” region falling in one of 8 positions equally spaced around a circle moved either leftwards or rightwards (limited life 2–4 frames) for 150 ms. Subjects were required to identify the direction of motion, not the position of the patch. The target was cued by a partial cueing technique, where 1, 2, 4 or 8 spokes pointed to the target, as well as to a variable number of “distractors”. The 8 potential-target regions could either be contiguous with the background random dots or be windowed as distinct stimuli behind a grey background. Coherence thresholds were lowest when the target was cued by a single spoke, and increased with the square-root of the number of spokes. We also measured contrast sensitivity for motion direction and found no set-size effect; but if we added a field of high-contrast noise dots only on the target and distractor regions, there was a square-root set-size effect for contrast thresholds. Interestingly, in conditions where a set-size effect was observed, the slope of the psychometric function also increased with set-size; when set-size had no effect on threshold, the slope of the psychometric function was always steep. A model based on the largest response of the pool of monitored detectors predicts that the slope of psychometric function increases with set-size (uncertainty), and that the presence or absence of a set size effect depends on stimulus and intrinsic uncertainty. The high intrinsic uncertainty for contrast thresholds may reflect an underlying large neuronal population (such as V1), while the low uncertainty associated with coherence thresholds may reflect a smaller population with large receptive fields (such as MT).