Bayesian Approaches to Color Vision

Abstract

Visual perception is difficult because image formation and sensory transduction lose information about the physical scene: Many different scenes lead to the same image data. Understanding how the brain copes with this information loss, so that our percepts provide a useful representation of the world around us, is a central problem in cognitive neuroscience. In the case of color vision, the nature of the information loss is well understood. First, the light reflected to the eye confounds illuminant properties with those of objects. Second, spectral and spatial sampling by the cone photoreceptors further reduces the available information. To provide a stable representation of object color, the brain must compensate by combining the directly available information with assumptions about which scene configurations are likely to occur. This chapter reviews how Bayesian decision theory can model how this happens and discusses two Bayesian models that have been effective in accounting for color appearance.