Abstract
The disparity energy model suggested that disparity is coded by neurons whose receptive fields for the left and right eye inputs have similar spatial frequency (SF) selectivity but with a p/2 phase-shift. The phase shift between the two RFs determines the preferred disparity of the channel. Thus, the disparity selectivity of a channel depends on its preferred SF (Fleet et al., 1996; Ohzawa et al., 1990; Sanger, 1988). Here, we investigate whether the SF of band-pass noise stereograms affect the perceived depth from disparity. The test stimuli were rectangular band-pass noise stereograms (1.27 x 3.44 degree), in which the binocular disparity at each location was determined by a raised cosine function (0.29 cycle/deg) to give a bulge or depression percept. The band-pass filter is a Gaussian distribution centered at 1.5, 3 or 5 cycle/deg with a bandwidth of one octave. The maximum test disparity ranged from 0 to 24 arc min while the luminance contrast ranged from 5% to 80%. The observers' task was to adjust the length of a horizontal bar to match the perceived depth in the test stimuli. Regardless the central SF of the band-pass filter, the perceived depth was an inverted-U disparity matching function. Furthermore, for each SF condition, both peak and peak position of the disparity matching function increased with luminance contrast. As the peak SF increased, the peak of the disparity matching function increased while the peak-position decreased. Thus, the perceived depth depends on both luminance contrast and SF of the test pattern. Our results cannot be explained by a phase-encoding disparity energy model, which would predict that the perceived depth to be a constant within each SF band. Instead, our result suggests a disparity averaging across channels that subject to contrast gain control, which can be affected by SF filtering.
Meeting abstract presented at VSS 2018