Local luminance factors that determine the maximum disparity for seeing cyclopean surface shape

We measured the maximum disparity grating amplitude (d(max)) for seeing cyclopean surface shape, using stereograms made from dense arrays of micropatterns, whose luminance characteristics were manipulated. In Experiment 1, we used disparity gratings made from Gabor micropatterns. D(max) was found to vary inversely both with luminance spatial frequency and with Gabor size, but was constant for a constant bandwidth (frequency times size). To test whether this was due to changes in bandwidth per se or to changes in the number of local features, in Experiment 2 we manipulated the local feature content with a range of micropatterns that we termed 'edgels'. The results supported neither hypothesis. In Experiment 3 we varied the phases of the Fourier components of square wave edgels, thereby introducing more features, and we found that this did not change d(max). Taken together, our results show that d(max) decreases with an increase in the number of local luminance cycles at each luminance scale. D(max) is mainly limited by false target matching between similar components of the micropatterns. Stereopis, in terms of surface shape perception, is served only by first order mechanisms, and only by luminance filters that are broadband.