Abstract
Idiosyncratic directional preferences have been reported in several exemplars of ambiguous perception, such as perceived depth order in transparent motion (Mamassian & Wallace, 2010), perceived tilt in structure-from-motion (Wexler et al., 2015) or perceived motion direction in apparent motion (Schütz, 2014). Although these preferences are widespread across individuals and stimuli, their functional relevance remains elusive. Here we tested the hypothesis that the function of these biases is to minimize ambiguity and perceptual decision costs. In this case, presenting stimuli more frequently along the most ambiguous axis should alter the directional preference to reduce ambiguity along this axis. In two separate experiments, observers had to report the perceived motion direction in an apparent motion stimulus that was consistent with two opposite motion directions or the motion direction they perceived in front in a transparent motion stimulus. In a pre-phase, all stimulus axes (0° to 170° in 10° steps) were equally likely and we measured observers' individual directional preferences. In a test-phase, either the least ambiguous axis, parallel to the individual preference, or the most ambiguous axis, orthogonal to the individual preference was shown in 40% of trials. In the remaining trials, all other axes were equally likely to measure individual preferences. A post-phase was identical to the pre-phase to detect potential aftereffects. Replicating previous findings, our results showed pronounced directional biases in both experiments. Individual preferences changed slightly between pre-, test and post-phases, but the amount of change was not modulated by the axis that was presented more frequently, parallel or orthogonal. Altogether, our results indicate that the reduction of ambiguity is not the primary aim of directional preferences. Since directional preferences can be adapted according to their usefulness for an additional task (Chopin & Mamassian, 2011), this suggests that perceptual ambiguity is not avoided by the visual system.
Meeting abstract presented at VSS 2018