We ourselves could not confirm the accuracy of this, because our color vision interfered with our ability to visualize the gray scale, as, earlier, normally sighted viewers had been unable to perceive the tonal sense of his confusingly polychromatic flower paintings. But a black-and-white photograph and a black-and-white video camera confirmed that Mr. I. had indeed accurately divided the colored yarns in a pure gray-scale manner. There was, perhaps, a certain crudeness in his categories, but this went with the sense of sharp contrast, the paucity of tonal gradations, that he had complained of. Indeed, when shown an artist’s gray scale of perhaps a dozen gradations from black to white, Mr. I. could distinguish only three or four categories of tone.
One anomaly again showed itself in the wool-sorting test: he ranked bright saturated blues as “pale” (as he had complained that the blue sky seemed almost white). But was this an anomaly? Could we be sure that the blue wool was not, under its blueness, rather washed-out or pale? We had to have hues that were otherwise identical—identical in brightness, saturation, reflectivity. This is the case with a set of carefully produced color buttons known as the Farnsworth-Munsell test, which we now gave to Mr. I. He was unable to put the buttons in any order, but he did separate out the blue ones as “paler” than the rest.
When we asked Mr. I. to examine and paint a copy of a colored spectrum (we used the printed one in Helmholtz’s Physiological Optics), he could see only black and white and varying shades of gray, and painted it as such. Intriguingly, his perception of the spectrum bore no resemblance to that of the retinally colorblind (which has a single peak of luminosity in the green around 500 nanometers) but did resemble that of people with normal (“photopic”) vision, whose perception of luminosity reaches a peak in the yellow-green (around 560 nanometers). This showed that his cone mechanisms and discrimination of wavelengths were intact, and only color “perception” (or “construction”) was deficient. There was, however, a strange, additional anomaly: an additional luminosity in the blue part of the spectrum, similar to the transformation of blue already observed in the Farnsworth-Munsell test.
We now came to the classic color-dot test plates always given as a test for colorblindness—the Ishihara plates, in which configurations or numerals of subtly differentiated colors may stand out clearly for the normally sighted, but not for those with various types of colorblindness. Mr. I. was unable to see any of these figures 9although he had no difficulty with certain “trick” plates, which are designed to catch pretended or hysterical colorblindness).
Though problems arose occasionally when he was shown reproductions of colored paintings, Mr. I. had no difficulty describing black-and-white photographs or reproductions accurately; he had no difficulty recognizing forms. His imagery and memory of objects and pictures shown to him were indeed exceptionally vivid and accurate, almost eidetic, though always colorless. Thus, after being shown a classic test picture of a colored boat, he looked intensely, looked away, and then rapidly reproduced it in black-and-white paint (see illustration on page 32). Vivid positive and negative afterimages occurred after he was shown bright colors, but these were also devoid of color. When asked the colors of familiar objects, he showed no difficulties in color association or color naming. Nor did he (now) have any difficulties reading.
Testing up to this point—other forms of visual testing, and a general neurological examination, were entirely negative—had shown an isolated but total achromatopsia or colorblindness, but one with some atypical features. Clearly his case did not resemble “ordinary” colorblindness, in which the color receptors of the eyes are defective or missing. Mr. I. made distinctions where the retinally colorblind could not—e.g., the blues (although these were seen not as “blue” but as “pale”) on the Farnsworth-Munsell and other tests. In some sense, it seemed, he was “seeing” the blue, at least seeing something about it, although (to use the current word) he could not, apparently, “process” this internally to create the cerebral or mental construct of “color.” Thus we needed more sophisticated tests, designed to explore the brain’s mechanisms for generating and perceiving color.
Such cerebral mechanisms may be examined by the active responses of a subject (human or animal), responses that indicate what the subject is perceiving. They may also be examined in a purely anatomical or physiological way, by visualizing or measuring the electrical activity of the brain. The first (or neuropsychological) approach is of particular use for examining color perception, since the areas of the brain involved in this are so minute that they may elude direct visualization. Efforts had indeed been made to delineate the brain damage in Mr. I.’s case (by the use of special scan techniques: CAT scan, NMR scan), and to measure the physiological reactions of the visual cortex (with evoked potential tests), but these tests were all negative. With more sophisticated brain imaging we might well be able to identify the minute brain areas affected; but Mr. I. was getting tired of “all those tests,” and for the present it seemed best to return to perceptual testing, but in a more elaborate form.
“Higher” forms of color perception have engaged the interest of Edwin Land in this country and S. Zeki in England, who have both devised a number of experimental and clinical tests. These use complex, subtly juxtaposed blocks of different colors, with a vague resemblance to some paintings of Mondrian (and hence sometimes called “Mondrians”). The colored shapes are projected on a screen through filters that can quickly be changed. In January 1987, with the patient, we met with Professor Zeki, and performed more elaborate testing. A “Mondrian” of great complexity was used as a test object, and this was projected with white light and with extremely narrow-range gel filters allowing the passage of only red, green, and blue light. Strictly speaking, of course, one should refer, as we did during the testing, not to color but to the wavelengths that are associated with each color—to long, medium, and short wavelengths respectively.
Mr. I., it was evident, could distinguish most of the geometric shapes, though only as consisting of differing shades of gray, and he instantly ranked them on a one-to-four gray scale, although he could not distinguish some color boundaries (for example, between red and green, which both appeared to him, in white light, as “black”). With rapid, random switching of the filters, the gray-scale value of all the shapes dramatically changed, some shades previously indistinguishable now becoming very different, and all shades (except actual black) changed, either grossly or subtly, with the wavelength of the illuminating beam. (Thus a green area would be seen by him as “white” in green [medium-wavelength] light, but as “black” in white or red [long-wavelength] light.)
All Mr. I.’s responses were consistent and immediate. (It would have been very difficult, if not impossible, for a normally sighted person to make these instant and invariably “correct” estimations, even with a perfect memory and a profound knowledge of the latest color theory.) Such a response was utterly unlike that which would be made by someone with retinal colorblindness—i.e., an absence of receptors sensitive to wavelengths in the eye. Mr. I., it was clear, could discriminate wavelengths—as no retinally colorblind person could—but he could not go on from this to “translate” the discriminated wavelengths into color, could not generate the cerebral or mental construct of color.5
This finding not only pinpointed the nature of the problem—the inability to “create” color, to “arrive at” colors on the basis of information about wavelengths, edge-matching, etc.—but also served to pinpoint the location of the trouble. For it has been established, directly, in animal experiments (conducted by Zeki), and the human cases of achromatopsia reported would support this, that the visual cortex deals with “color” (and other percepts) twice. First, it discriminates and categorizes the physical aspects of the stimulus (e.g., wavelength, displacement in time, parallactic displacement, etc., as these have been coded by the retina); this is done in the primary visual cortex. Second, it constructs from these the perceptual qualities required for an image (color itself, movement, depth, etc.); this is done in another region of the brain, the secondary, or associational, visual cortex.
What had been suggested by Mr. I.’s history, and by the other tests, was definitively corroborated by the “Mondrian” test: it was the visual association cortex, and this only, that had been damaged in Mr. I. Hence his inability, despite the intactness of the retinal output and processing in the primary visual cortex, to construct color (and, for a short time, letters) as an element of the visual world.
There is a simple or “naturalistic” way of regarding color, and indeed the whole perceptible world, that has its philosophical exemplar in Locke and its scientific exemplar in Newton. Here sensations are given an “absolute” status corresponding to the “absolute” status of physical stimuli: nothing is added, nothing is removed, in passing from the outer world to the inner world of each person or sentient being. The mental world, according to this philosophy, is a physical world—a little replica of it, perhaps, within the brain. Newton, in his famous prism experiment in 1666, had shown that “white” light was composite—could be decomposed into, and recomposed by, all the colors of the spectrum. The rays that were bent most (“the most refrangible”) were seen as violet, the least refrangible as red, with the rest of the spectrum in between. The colors of objects, Newton thought, were determined by the “copiousness” with which they reflected particular rays to the eye.
But it was not necessary to have all the spectral colors; artists had long known that one could obtain most colors by the admixture of as few as three brightly colored pigments. This, and perhaps also John Dalton’s description of his own colorblindness a few years before, moved Thomas Young, in 1801, to his “trichromatic” hypothesis, the hypothesis that the eye had just three color receptors, which were “tuned” to resonate to red, green, and blue. Young’s hypothesis was confirmed by Helmholtz a half-century later, so that we now speak of the Young–Helmholtz hypothesis.
But for Helmholtz there was something mysterious, nonmechanical, at work too. Objects retain their “color” even in very different illumination: for example, in the evening when they are bathed in long wavelengths. This obvious yet central phenomenon—of color constancy—was seized on by Helmholtz as implying that something active went on, not simply a mechanical translation of wavelength into color. He spoke of color vision as “an act of judgment.”
This was, for Helmholtz, a special example of the general act of “perceptual judgment” required to make a stable world from a chaotic sensory flux, a world that would not be possible if our brains merely reflected passively the ever-changing input that bathed our receptors. There is thus, in Helmholtz, even though he is seen as the great successor of the Newton and Young tradition, something that departs radically from the naturalistic tradition, in that it assigns an active role to the organism and to the brain.
When looking at the "Mondrian," Mr. I. consistently saw blues as a sort of brilliant grayish white (as had been the case on all the other occasions on which his color vision had been tested). Professor Zeki was puzzled by this, as we had been, and said that he could offer no explanation.↩
When looking at the “Mondrian,” Mr. I. consistently saw blues as a sort of brilliant grayish white (as had been the case on all the other occasions on which his color vision had been tested). Professor Zeki was puzzled by this, as we had been, and said that he could offer no explanation.↩