Abstract
Sparsely distributed dots in depth often elicit a percept of a 3D surface which spans the blank regions. We examined this surface interpolation phenomenon for point depths defined either by motion parallax or binocular disparity. In Experiment 1 a novel ‘brightness illusion’ technique was used to evaluate surface interpolation. The stimulus consisted of two vertically separated groups of dots (each in a frontal plane). The upper plane was closer to the observer in depth than the lower. The background was light grey, with a dark grey horizontal, depth-ambiguous strip located between the groups of dots. Interpolation between the groups of dots caused the horizontal strip to appear slanted and brighter than when perceived as frontal, consistent with the operation of lightness constancy. We manipulated the vertical distance, y, between the dots and the strip in both cue conditions. Observers set the brightness of a remote probe to match the horizontal strip. The apparent depth of the points was matched across cue conditions, permitting a useful comparison of interpolation. Results showed clear consistency across cue conditions, however, there were large individual differences in illusion magnitude, and variation with y. To determine if the inter-subject differences were attributable to variation in perceived slant, we measured perceived slant as a function of y in Experiment 2. The results again showed consistency across cue conditions, and the variation in slant judgements accounted for the individual differences in Experiment 1. Taken together, these experiments provide convincing evidence that: 1, Surface interpolation is not depth cue specific, rather it depends on the apparent depth of points in a scene. 2, Its operation varies between individuals. 3, Lightness constancy is applied to interpolated regions and is not depth cue specific.