Abstract
Purpose: Studies on binocular contrast sensitivity have predominantly focused on flat, two-dimensional gratings. The underlying hypothesis of such studies is that contrast sensitivity is determined at the early stages of visual processing and is not influenced by the process of 3D shape recovery. However, it can be argued that contrast detection involves identifying changes in albedo of a 3D surface rather that strictly determining the presence of 2D luminance changes (Knill & Kersten, 1991). Therefore, if a sinusoidal grating aligns with a stereo-specified sinusoidal corrugation, it should be harder to detect than an identical grating that does not align with a corrugation. In effect, in the first case the grating would be interpreted as a shading pattern produced by illuminating a Lambertian corrugated surface, whereas in the second case the grating would be interpreted as a spatial modulation of the reflectance properties of the surface.
Method: Subjects were presented with random-dot stereograms representing 3D sinusoidal corrugations superimposed on low frequency luminance gratings. Gratings and corrugations were frequency-matched. The corrugation was either aligned with the grating, orthogonal to the grating, or flat. The task was to discriminate the presence of the grating from a uniform gray field.
Results: We analyzed the average difference between the measured d′ for detecting the grating superimposed on the corrugated surfaces and the d′ for detecting the grating on the flat surfaces. Consistent with our hypothesis, this difference was positive in the orthogonal case and negative in the aligned case.
Conclusion: Our results suggest that contrast detection does not exclusively rely on early detection of 2D luminance changes but that it involves later stages of visual processing concerned with 3D surface reconstruction.
Supported by National Science Foundation grant 78441