Characterization of human lightness discrimination thresholds for independent spectral variations

Abstract

The lightness of an object is an intrinsic property that depends on its surface reflectance spectrum. The visual system estimates an objects lightness from the light reflected off its surface. The light reflected also depends on object extrinsic properties of the scene. For stable perception, the visual system needs to discount variations due to extrinsic properties. We characterize this perceptual stability for variation in two spectral properties of the scene: the reflectance spectra of background objects and the intensity of light sources. We use a two-alternative forced-choice task to measure human observers thresholds of discriminating computer-generated images of 3D scenes based on the lightness of a spherical target object in the scene. We measured how the discrimination thresholds changed as we varied the reflectance spectra of the objects and the intensity of the light sources in the scene, both individually and simultaneously. For small amounts of extrinsic variations, the thresholds of discrimination remained constant indicating that the thresholds were dominated by observers intrinsic representation of lightness. As extrinsic variation increased, it started affecting observers lightness judgment and the thresholds increased. We estimated that the effects of extrinsic variations were comparable to observers intrinsic variation in the representation of object lightness. Moreover, for simultaneous variation of these spectral properties, the increase in threshold square compared to no variation condition was a linear sum of the corresponding increase in threshold squares for the individual properties, indicating that the variation from these independent sources combines linearly.