Abstract
The human visual system has specialised mechanisms for encoding mirror-symmetry. Recent studies suggest that temporal dynamics rather than symmetrical motion direction contribute to mirror-symmetry perception (Sharman & Gheorghiu (2016), Perception, 45). Here we investigate temporal aspects of symmetry perception by examining how pattern elements are combined over time. Stimuli were dynamic random-dot patterns consisting of an on-going alternation of two images containing different amounts of mirror-symmetry about the vertical axis. We used five different stimulus configurations: 1) a symmetric pattern alternated with a noise pattern; 2) the left and right halves of the symmetric and noise patterns presented with temporal delay (i.e. delayed halves condition); 3) two symmetric patterns alternated over time, with each pattern containing 50% of the [RS1] symmetric matched-pairs; 4) the same as configuration 3, but with matched-pairs presented with temporal delay (i.e. delayed matched-pairs condition); 5) symmetric and noise patterns presented simultaneously as one static pattern. We varied the presentation duration of the two images between 23.5ms and 294ms and the proportion of symmetrical dots. We measured symmetry detection using a 1AFC procedure in which participants indicated whether or not a stimulus was symmetric. Analyses of the slopes of the psychometric functions fitted to the data for each condition showed that symmetry is (a) detected in delayed halves and delayed matched-pairs conditions up to a temporal delay of about 120ms; (b) easily perceived when the symmetric halves or matched-pairs are simultaneously presented, irrespective of the duration of the noise pattern. We conclude that cross-correlation across the symmetry axis can be integrated over time and symmetry mechanisms can tolerate delays of up to about 120ms.
Meeting abstract presented at VSS 2017