5.32(c)]. A relationship between achromatic and chromatic ‘primary processes’, such as the one postulated by Judd and by Hurvich and Jameson, is not likely to be found at the early levels since retinal and LGN opponent cell responses cannot be directly linked to the perception of elementary hues (Valberg, 2001).

It may be of interest to ask what consequence the relative number of L and M cones in the retina may have for hue perception? Does the excitation and response amplitude of opponent cells, and hence chromatic strength, depend on the relative number of each cone type having input to the cell, or maybe to the number of synaptic contacts? In a retina where the L-cones dominate (or have more L-cone synapses), would the ‘L–M’, Increment and Decrement cells (the red ON-center cells and green OFF-center cells) be more excitable (except in the fovea where there is a one-to-one relationship), or would there be more such cells? Would this lead to a stronger input to higher-level ‘L– M’cells, and to stronger inhibition of high-level ‘M–L’ cells. The relation between cone proportions and their synaptic contacts to ganglion cells and hue perception is by no means clear. For instance, there is no clear numerical relationship between relative cone density and elementary hue.

Summary

In this chapter, we have explained why a match of any colored light can be achieved using only three base colors, and that this is readily explained by the excitations of three types of cone. A match is obtained when, and only when, each of the three cone types is excited equally by the two lights being compared. Systems for color measurement, like that of CIE, rest upon this principle of equality. Such systems, however, say little about the colors actually perceived in a match.

In accounting for perceptive color phenomena, receptor excitations and linear models of neural processing are bound to fail except for near threshold. Neurophysiological research has brought major advances in our understanding of the underlying mechanisms of color vision, but much remains to be discovered. We still have no complete physiological model of color vision, but the model we have described here has many promising features. In addition to describing color differences and color scaling in an encouraging way, it accounts well for constant hue perception, the Abney effect, and the Bezold–Bru¨cke phenomenon (considering the large individual differences that exist in the psychophysical data).

When a spectral light increases in luminance, the hue changes. Normally, longwavelength light becomes increasingly yellow, and short-wavelength light turns blue or blue-green. Less attention has been paid to the change in relative chromatic content (saturation or chromatic strength) that accompanies these hue shifts. As luminance increases from zero, chromatic strength grows to reach a maximum at a luminance that is wavelength-dependent. Short-wavelength bluish light reaches this maximum at

low relative luminance, whereas mid-spectral yellowish stimuli need several log units higher luminance. Red and green fall somewhere in between. For a luminance above this maximum, the chromatic strength usually diminishes, and most wavelengths become more whitish. Both phenomena can be accounted for by a physiological interpretation in terms of the nonlinear and non-monotonic responses of coneopponent cells in the retina and lateral geniculate nucleus of the primate. Our model, that combines the outputs of six opponent cell types, accounts fairly well for observers’ estimates of hue and chromatic strength.

However, further explanation is required. For instance, we would like a more complete model to explain:

color adaptation;

simultaneous color contrast;

color constancy, or more specifically the deviations from color constancy (being of greater practical importance);

border and area contrasts.

If one believes, like Ernst Mach did, that a unique color perception has a unique neurophysiological correlate, then one must also expect a future model to incorporate a structure of

unique hues and elementary colors; and

of white and black

The last two points refer to qualities of immediate color experiences, phenomena that relate to a different level of experience than do those mentioned earlier.

We have argued that, for the same adaptation, combined neural activities of opponent cells correlate well with the perception of constancy of hue and chroma, whereas the elementary hues qualities cannot be associated with these neural processes. This is a general statement, not restricted to the model presented here. The old idea that the elementary colors have a direct relation to activation and inhibition of early opponent cells, either in the retina or LGN, does not hold. As yet, no other physiologically plausible model has come up with good neural correlates. What conclusion should we draw from this lack of correspondence? Is the program that Crick (1994) announced to find functional parallels between neural activity and perception proving more difficult than one had anticipated at least for color vision? Perhaps neural correlates for the qualitative basis of color vision – such as the unique colors – do not belong to Chalmer’s easy problems after all (Chalmers, 1995a, b; Valberg, 2001).

When the experience of colors is related to physical stimulation (even though qualities cannot logically and causally be derived from it, since physics and perception are on different sides of the ‘explanatory gap’), it should be possible to

establish neurophysical and neuropsychological relationships at different levels in the visual process. For instance, a number of phenomena and experiments point to underlying opponent mechanisms. This applies to adaptation to the prevailing illumination, color discrimination, the organization according to attributes (dimensions) in a three-dimensional color space, and movement. In all these cases, the neuro-scientific program still seems to be adequate.

Lightness, hue and color strength (chroma) are abstract dimensions (attributes) of colors, and they have a simpler connection to known opponent neural processes than perceived color qualities. These three dimensions can be given geometric representations, and color discriminations made along these dimensions can be derived form neural activities. In blind-sight it is possible for a person to discriminate between colors without having a conscious knowledge of seeing them. If asked, the person will firmly deny that he has seen any color at all. These rare cases suggest that stimulus discrimination may depend on rather peripheral processes, since some information can be available without consciousness perception. The processes in the primary visual center, V1, for instance, seem largely to escape consciousness (Zeki, 1993; Crick, 1994; see also Chapter 7).

We have offered a physiological explanation of color discrimination in terms of equal differences in responses of cone-opponent units, but the qualitative opponency and independence of elementary colors still remains a problem. Referring to what we have just said about blind-sight, this is perhaps not so strange, considering that elementary colors are qualities associated with conscious experience, whereas electrophysiological recordings are done on anesthetized monkeys. Unique colors may be regarded as an inner, subjective, but nevertheless common reference system for color perception for all people. However, we are not primarily seeking an explanation of these qualities, only for the particularities of some unique directions and structures in psychophysical color space. We had hoped – and maybe expected – that they would be revealed in the firing of visual neurons.

We are free to interpret colors as expressions of physical and chemical processes, and also as a product of the reaction of the brain and the neural system to stimulation of the eye by light. Colors reveal themselves as one of many attributes of our visual world, but they can also be treated quantitatively and be included in an objective, scientific description at the same level as form, movement, depth, texture, orientation and direction. When I experience colors, organize them, and explore their relation to physical stimuli and neural activity, this happens within logical structures and with reference to my culture’s understanding and knowledge. While I would maintain that my theoretical analysis of color phenomena and my qualitative experience of them are two complementary activities, and that they require different concepts and verbal expressions, I recognize that this is rooted in my preliminary understanding of the phenomenon color. Neuroscience makes it possible to investigate these and other relationships experimentally, and, as culture and science develops, we expect our understanding to change, as has been the case from ancient times to today.

Color and art in a scientific perspective

When focusing on the evolutionary advantages of trichromatic color vision, one usually explains its development from previous achromatic vision by the advantage it provided for better discrimination of fruits from foliage, to discover enemies, etc. Beside this biological advantage, is there no other value to color vision? What about perception and esthetics? The auditory sense, for instance, together with our intellectual and emotional capacities and the faculty to develop culture, has made it possible for us to appreciate a rich world of sound and music. Color vision obviously provides humans with yet another esthetic dimension. It has enabled us to produce new visual experiences and to admire and appreciate colors and texture, on clothes, in art and images, on artifacts, and in nature. The access to the dimension of color has opened our minds to a wealth of new esthetic adventures, much in the same way as our ability to perceive sound has allowed us to enjoy music.