Howard and Kaneko (1994 Vision Research 34 2505 – 2517) suggested that stereoscopic slant and inclination may be encoded on the basis of relative expansion and shear, respectively. We investigated the perception of surfaces with diagonal axes of tilt, having components of both inclination and slant. Latencies for the perceptual resolution of slant can be several times those for inclination (Gillam et al, 1988 Journal of Experimental Psychology 14 163 – 175). Using a forced choice methodology, we measured the time required to discriminate surfaces with horizontal, vertical, and diagonal axes. Subjects were presented with random-dot stereograms of planar surfaces, with or without a visual reference. Stimuli were followed by a random-disparity mask. In the first experiment, subjects discriminated the direction of slant or inclination of surfaces rotated in depth by 45° from frontoparallel. A staircase procedure was used to find the shortest presentation time for which surfaces could be correctly discriminated. With a visual reference, both surface types could be discriminated with brief presentations. Without a reference, mean thresholds were 300 ms for inclined surfaces, and 2.5 s for slanted surfaces. In a second experiment, temporal thresholds were measured for surfaces with diagonal axes. In one condition, subjects discriminated between surfaces with equal inclination, but opposite slant. In the second condition, surfaces had equal slant, but opposite inclination. Surfaces differing in inclination could be discriminated with shorter presentations than surfaces differing in slant. These results support the notion that surfaces are encoded in terms of components of inclination and slant.
