Abstract
We have previously described a perceptual asymmetry that occurs when viewing pseudo-transparent random element stereograms. That is, the minimum separation in depth needed to segregate two overlaid surfaces in a random-element stereogram depends on the distribution of elements across the surfaces. With the total element density fixed, significantly larger inter-plane disparities are required for perceptual segregation of overlaid surfaces when the front surface has fewer elements than the back surface than vice versa. In the experiments described here we test the hypothesis that this perceptual asymmetry reflects a fundamental difference in signal strength for the front and back surfaces which results from disparity interpolation. That is, we propose that the blank regions between elements are assigned to the back plane, making it appear opaque. We tested this hypothesis in a series of experiments and find that:
- i)
the total element density in the stimulus does not affect the asymmetry
- ii)
the perceived relative density of the two surfaces shows a similar asymmetry
- iii)
manipulations favouring perceptual assignment of the spaces into surfaces other than the two overlaid element surfaces reduces the asymmetry.
We propose that the interpolation of the spaces between the elements defining the surfaces is mediated by a network of inter-neural connections; excitatory within-disparity, and inhibitory across disparity. Our data suggest that the strength of the inhibitory connections is modulated according to mid-level figure ground assignment. We are using our psychophysical results to inform the development of a computational model of this network.
Natural Sciences and Engineering Research Council of Canada.