Abstract
Stimuli have been constructed to separate simultaneous contrast into local retinal (i.e., opponent) processes and global cortical level processes (i.e., inferred illumination gradients). Using a haploscope to display binocular disparity on a CRT screen, a number of small circles were randomly assigned to one of two depth planes. One circle contained a test patch, while a circle in the other depth plane contained a comparison patch. The subject's task was to vary the test patch in intensity to match the comparison patch in brightness. The circles in each depth plane differed by a 5:1 luminance ratio. Within a depth plane the luminances of the circles could either be (1) constant, (2) graded along a spatial continuum (with the upper-right region of the scene having a higher luminance than in the lower-left region), or (3) graded in a spatially random fashion. These luminance distributions created the perception of either (1) a light falling directly on the circles from behind the subject's head (conditions 1 & 3), or a light coming from behind the upper-right of the subject's head (condition 2). Simultaneous contrast was 33% in the constant luminance condition, and 26% in the graded spatial continuum. However, spatially randomizing the luminances of the circles (condition 3) reduced simultaneous contrast to a mere 10%. This remaining contrast is hypothesized to be the full extent of local, retinal opponent processes, while in conditions 1 and 3 global cortical processes that alter the ability to infer the illuminant produce the additional simultaneous contrast.