Institute for Mathematical Behavioral Sciences

Perceived color has been an object of wonder and fascination for laymen, artists, philosophers, and scientists for centuries. Obviously the illuminant and surface reflectances of objects affect perceived color. In the primate, cone pigments have different sensitivities to different regions of the visible spectrum. Recordings from many locations in the primate brain also show specificity to the wavelength properties of objects. Modern brain imaging studies in awake humans are just beginning to understand the brain processes that underlie color perception, as distinguished from simple wavelength processing.

Principal components analysis of the wavelength response curves in many different color domains indicates that to a first approximation, three components can span most spectral spaces -- whether natural light illuminants, the reflectance spectra of natural objects or scenes, primate cone pigments, primate LGN cells, or the perceived colors of patches on a neutral background. Interestingly, the three components in the different domains are related by simple rotational transformations. A common model for representing simple color-related phenomena across these domains might therefore be possible.

However, there are numerous perceived color phenomena which cannot be explained by a linear model. Some of these phenomena have immediate practical consequences, such as the effect of sunlight glare on radio displays in vehicles, which is not well modeled by current colorimetric theory. Many color “illusions” also cannot be explained by a linear model – Benham’s disk (multiple colors from 1 color channel), Land-like phenomena (multiple colors from 2 color channels), or brightness illusions such as the attached (which can easily be extended to the color domain). The patches marked “A” and “B” are identical.