Abstract
Primate vision exhibits a "light-dark asymmetry" originating with ON and OFF channels in the retina. Neurophysiology and psychophysics show a processing speed advantage for darks (Komban et al. 2011, J Neurosci). We looked for a corresponding "dark benefit" in global motion perception. Performance during global motion perception in random-dot kinematograms (RDKs) depends on stimulus duration, so when stimulus duration is very short, a dark benefit in early vision could lead to better efficiency for dark dots as compared to light dots. We therefore measured percent coherence thresholds for dark-dot and light-dot RDKs, on a grey background, at four stimulus durations: 0.33, 0.23, 0.14, and 0.07 sec. Viewing was monocular or binocular, to test whether binocularity interacts with dot polarity. Four normally sighted observers were tested in a forced-choice net motion discrimination task using a stereoscope. Threshold was estimated using a 3-down-1-up staircase procedure. Stimuli contained 100 dots. Two-frame dot motion at 30Hz was used to minimize selective tracking. Dot polarity and stimulus duration were constant within blocks, while trials from the monocular and binocular viewing conditions were intermixed. As expected performance was better (lower percent coherence thresholds) at long stimulus durations. Surprisingly, all observers performed better with light dots than with dark dots of the same contrast. The neural mechanism behind this advantage is unknown. For light dots, performance in the monocular and binocular conditions was similar, consistent with Cai, Yuan, and Backus (VSS 2015). However, for dark dots, two highly practiced observers performed better when dots were presented binocularly; this unexpected interaction could result from left-eye/right-eye asymmetry of input to global motion mechanisms.
Meeting abstract presented at VSS 2017