Abstract
We have examined the effects of chromatic surface-variegation on the identification of shadows. We employed a 6-luminance display comprising a simulated shadow overlaid on a tripartite background. The three background luminances were drawn randomly from a specified distribution that was identical for all conditions. The three conditions were a) achromatic — the stimulus was a monochromatic grey; b) chromatic — background and overlaid shadow sectors were assigned the same hue, with hues drawn randomly from the gamut available on the monitor; c) chromatic-all-border — the same as the chromatic condition except that the background and shadow sectors were assigned different hues. Using a 2IFC task, subjects had to decide which of two displays contained the shadow that looked most “natural”. The natural shadow stimuli were created by setting all three shadow sectors to 0.5 × the luminance of the background sectors upon which they were overlaid, whereas the unnatural stimuli were created by setting two sectors to 0.5 × background luminance and the third to a value that deviated from 0.5 by various specified amounts. Results showed the following order of performance: chromatic > achromatic > chromatic-all-border. This finding supports the hypothesis that the visual system has an in-built assumption that chromatic variations arise from surfaces, while uncorrelated, pure-luminance variations result from differential illumination. The results also reveal a new role for colour vision in identifying the spatial layout of the scene, specifically in helping discriminate shadow from surface.