Stimulus information supporting bilateral symmetry perception

Abstract

A classification image (a psychophysical reverse-correlation) method was used to investigate what stimulus regions and information the visual system uses for bilateral symmetry perception. The stimuli were symmetric random-dot patterns with either low or high dot density. First, the spatial integration region supporting symmetry perception was estimated, by analyzing the trial-to-trial correlation between the spatial location of symmetric dots and the corresponding response. It was observed that the integration region was rather compact (3 deg2 with dense stimulus), vertically elongated and located near to the axis of symmetry. The size of the area was dependent on the pattern density, being larger with low-density stimulus. Next, the resolution of the symmetry matching was probed by estimating how close to the perfect symmetry the dots in two stimulus parts must be to be perceived as symmetric (classification image for symmetry tolerance). Dot pairings up to 6 arc min off from the mirror symmetry correlated with symmetry response, suggesting that the process underlying symmetry matching has large tolerance and low resolution. Outside the integration region, the symmetry tolerance classification image weights were essentially zero, suggesting that the lack of symmetry integration there is not a byproduct of high tolerance.