Color Vision: A New Understanding
John A. Medeiros
What Does a Color Vision Model Have to Explain?
Actually the three-cone model explains well almost nothing about how color vision works. Explanations of color vision phenomena in terms of the three-cone model tend to be rather contrived and involve assumptions about the existence of improbably large pigment densities, or non-linear processes occurring in the cones or their subsequent processing circuitry in the retina or the brain. A short list of what it does it not explain very well would include the following:
- The exact shape and details of the color (hue) discrimination curve
- The colorimetric purity (saturation) function
- Color defective vision in which the common forms of (red-green) color blindness are poorly explained by either the missing-cone type of explanation or by the substituted wrong pigment approach and where, even in dichromats with only one X-chromosome opsin, consistent color-names are used for the correct colors in otherwise “color-blind” subjects. Underscoring this problem (for the three-cone models of vision) is the careful work of Crognale, et. al. (1998) who examined dichromatic subjects that had only a single X-linked pigment gene but who could none-the-less make chromatic discriminations by Rayleigh matching (a standard technique to test red-green vision). After ruling out involvement by other possible receptors (rods or other types of cones) they concluded: "The mechanism of chromatic discrimination in the presence of a single photopigment therefore remains unknown."
- Subjective colors, where colors are seen by all observers in a consistent and universal way with intermittent black and white illumination as in the Benham’s Top phenomenon
- The Stiles-Crawford Effect of the second kind (SC-II), the change in the apparent color of a light with its direction of incidence on the retina (Stiles, 1937; Enoch and Stiles 1963). The change is predominantly a red shift. Within the context of the three-cone model, attempts have been made to explain the effect in terms of "pigment self-screening" (Walraven & Bouman, 1960) without notable success (Fuld, Wooten & Katz 1979; Alpern, 1986).
- The similar appearance of short-wavelength violet light and of purple, a mixture of red and blue
- That color vision function is only three dimensional to a first-order approximation. There is ample evidence that human color vision is substantially better than an absolute partitioning into only three dimensions at the first stage of detection would suggest. For example, Nascimento, Foster, and Amano (2005) conducted a principal component analysis of the psychophysical perception of natural scenes and stated as a result that: "The combination of the spectral diversity of the natural world and the observed levels of color discrimination suggest that estimates of the minimum number of basis functions necessary to reproduce natural scenes may need to be revised upward." They concluded, in fact, that the "...original images were visually indistinguishable from their approximations only if there were at least eight basis functions."
- That any one cone can report any color, including white (see above)
- And most importantly – the cone shape!
So, is that it? Are we left with nothing, with no idea about how human color vision works?
Well, no. The key is the cone shape.
Back to the Beginning