The Influence of Cast Shadows on the Detection of Three-Dimensional Curved Contour Structure

Abstract

Cast shadows have been shown to provide an effective ordinal cue to the depth position of objects. In the present study, two experiments investigated the effectiveness of cast shadows in facilitating the detection of spatial contours embedded in a field of randomly placed elements. In Experiment 1, the separation between the cast shadow and the contour was systematically increased to effectively signal different contour depth positions (relative to background elements), and this was repeated for patterns in which the lighting direction was above and from below. Increasing the shadow separation improved contour detection performance, but the degree to which sensitivity changed was dependent on the lighting direction. Patterns in which the light was from above were better detected than patterns in which the lighting direction was from below. This finding is consistent with the visual system assuming a “light-from-above rule” when processing cast shadows. In Experiment 2, we examined the degree to which changing the shape of the cast shadow (by randomly jittering the position of local cast shadow elements) affected the ability of the visual system to rely on the cast shadow to cue the depth position of the contour. Consistent with a coarse scale analysis, we find that cast shadows remained an effective depth cue even at large degrees of element jitter. Our findings demonstrate that cast shadows provide an effective means of signaling depth, which aids the process of contour integration, and this process is largely tolerant of local variations in lighting direction.