Human color constancy in cast shadows

Abstract

Most natural scenes contain cast shadows to a varying extent. Illuminant conditions inside and outside the shadow typically differ largely both in intensity and in chromaticity. Nevertheless, our daily experiences suggest that colored materials appear to have the same color in shadows even though the reflected light from the material might be very different. Two experiments were conducted to reveal mechanisms that support human color constancy in cast shadows. In all experiments we built a real scene that consisted of colored hexagons illuminated by two independent liquid crystal projectors simulating "sunlight" and "skylight", respectively. A part of the scene included a cast shadow under which observers were instructed to change the luminance and the chromaticity of a test field so that it appeared as a full-white paper under the shadow. The color of the skylight was manipulated, testing if our visual system uses a prior that the skylight is typically bluish or yellowish to achieve color constancy. We also created a condition where a cast shadow is not recognized as a shadow. Results showed that color constancy generally holds well in shadows and changing skylight color had little effect. Recognizing a cast shadow as a shadow also had no effect. Overall, these results are consistent with our daily experiences that we stably judge objects color even in shadows, providing a key step to reveal mechanisms of color perception in real-world scenes where lighting conditions spatially vary.