An opponent-process theory of color vision.

The two major theoretical accounts of color vision are those classified as the Young-Helmholtz and the Hering types of theories. For many years the former has been judged by most workers in the field to provide the simplest explanation of the way in which light stimuli give rise to color sensations. The advantages that appear to favor the Young-Helmholtz three-component hypothesis are two: it is parsimonious, and its postulates are easily quantifiable and hence subject to precise experimental test. In its parsimonious and easily quantifiable form, the theory is simple: in addition to the rods which subserve twilight vision, the eye contains three kinds of cone photoreceptors; each type of cone contains a differently selective photochemical substance; each is associated with its own specific nerve fiber; and each cone-photochemical-nerve fiber system is correlated with one of the three specific "fundamental" color sensations, namely, red, green, and blue (or violet). All sensations are considered as compounded of varying amounts of these three excitatory systems, with white arising from equal and simultaneous excitation of all three, and yellow from equal red and green excitations. The Young-Helmholtz three-cone, three-nerve, three-sensation theory derives directly from the basic fact of color mixture, namely, that all visible hues can be matched by the mixture, in proper proportions, of only three physical light stimuli. Based squarely on this fact, the theory is readily quantified in terms of the three measurable variables of color mixture experiments. But the three measured variables, it 1 Now at New York University. must be emphasized, are the three physical light stimuli used in the color mixture experiments; they are not the postulated three "fundamental" color sensations, for with each different stimulus triad used for color matching a different and equally valid triad of color mixture functions is obtained. Consequently, throughout some hundred years since the original formulation of the idea, a continued series of attempts has been made to find the proper transformation of the three measured colormixture curves that will bridge the gap and yield the unique spectral distribution curves of the desired physiological correlates of the three postulated "fundamental" color sensations. An infinity of such transformations is available for trial, and almost every serious adherent of the theory has proposed at least one new set of "fundamental sensation curves" (48, pp. 368-372). The search, however, continues, because serious defects have been found in every proposal made thus far. When the explanatory or predictive power of the theory in any given quantified form is tested it cannot handle more than a limited number of facts satisfactorily (11, p. 805). Moreover, some facts of color experience seem unassimilable into the framework of the simple Young-Helmholtz theory with its three independent, fundamental, process-sensation systems. How can this system of three independent processes be made to account, for example, for the apparent linkages that seem to occur between specific pairs of colors as either the stimulus conditions or the conditions of the human observer are varied? Why should the red and green hues in the spectrum predominate