Abstract
Mirror symmetry is believed to be encoded by specialised visual mechanisms. Recent evidence suggests that symmetrical motion-direction does not contribute to symmetry perception, but limiting the lifetime of pattern elements improves performance (Sharman & Gheorghiu, 2017). Here we examine whether symmetry detection mechanisms are selective for speed and lifetime duration of symmetrical pattern elements. Stimuli were dynamic dot patterns containing different amounts of positional-symmetry about the vertical axis. Symmetric patterns contained both position and motion-direction symmetry with matched-pairs moving in symmetrical directions, but with different pairs having randomly allocated directions. Noise dots moved in the same directions as signal dots, but did not have positional symmetry. We used stimuli in which symmetry signal and noise dots drifted at either the same or different speeds. We manipulated the amount of positional symmetry by varying the proportion of symmetrical dots and measured symmetry detection thresholds using a 2IFC procedure. Foil stimuli were noise patterns containing the same speed distribution as the symmetric stimuli. These symmetry detection thresholds were compared with those obtained with (i) static patterns and (ii) dynamic-flicker patterns in which the symmetry signal and noise dots had either the same or different lifetime durations and were relocated without local motion. We found that (a) symmetry detection thresholds were lower for moving and dynamic-flicker patterns than for static patterns in all conditions; (b) thresholds were higher when the symmetry signal and noise dots had the same speed and decreased gradually with increasing speed differences between symmetry signal and noise dots; (c) thresholds for dynamic-flicker patterns increased gradually with the ratio of symmetry-to-noise lifetime duration. We conclude that symmetry detection mechanisms are tuned to the speed of symmetrical motion and this is not explained by increasing the number of symmetrical element-locations as elements move from one location to the next.
Meeting abstract presented at VSS 2018