Human color vision and the cone photoreceptors

Color is a fundamental visual feature that imparts a richness to the man-made or natural world. Color vision has long fascinated those with an interest in sensation and perception. Isaac Newton (1730/1979), in his famous experiments using prisms to produce and mix colored lights, deduced that ìcolorî is not a property of light wavelengths per se, but rather a property of the eye. This raises an interesting question: Do all eyes perceive colors similarly? The answer is no.

Many of the interesting research questions in this area have been directed at individual differences among humans and inter-species differences. The rationale for this is straightforward: An understanding of "how" and "why" these differences exist will tell us a great deal about how color vision is achieved by the eye and brain.

One of the most fascinating features of color vision is the fact that, more than any other aspect of perception, color vision shows large individual differences both across and within species. Not all animals, in fact not even all humans, share the same color experiences. For many species, including humans, there are even differences between males and females. Oddly enough, there are species of monkey where most of the females have normal color vision but all the males are color-deficient (Jacobs, 1998). We consider below some exciting recent advances in our understanding of human color vision.

A long-standing and fundamental property of human vision is trichromacy. The trichromatic theory explains our color perceptions and color discriminations. The anatomical basis of trichromacy begins with the complement of cone photoreceptors in the retina. For over one hundred years researchers thought that the color-normal eye contained three cone types whose photopigments were later psychophysically estimated to have peak spectral sensitivities near 440, 540, and 560 nanometers. A Figure showing the spectral sensitivities of these pigments, across the visible spectrum, is shown below:

Figure 9. The relative sensitivity to different light wavelengths is shown for each cone photopigment.
Note the overlap in sensitivity of the middle wavelength sensitive and long wavelength sensitive cone types. [Figure 9 description]

Over the years, however, psychologists questioned whether subtle variations may exist in normal color vision based on small individual differences in the spectral sensitivities of the photopigments (Alpern & Wake, 1977; Neitz & Jacobs, 1986). The findings of the early studies were viewed with some skepticism, however, because of the difficulty in ruling out measurement error and confounding factors. As the psychophysical evidence grew, researchers began to investigate this possibility from many angles.

Today, psychophysical (Neitz & Jacobs, 1990; Mollon, 1992), microspectrophotometric (Dartnall, Bowmaker, & Mollon, 1983), and molecular genetic studies (Nathans, Piantanida, Eddy, Shows, & Hogness, 1986; Winderickx et al., 1992) provide evidence of substantial variation in the number and spectral sensitivity of the cone types in the color-normal eye (also see Mollon, Cavonius, and Zrenner, 1998). The evidence now suggests the presence of three families of normally occurring cone photopigments. There is thought to be only one photopigment with a peak spectral sensitivity in the shortwavelengths (blue), but there is now evidence that there are multiple middlewavelength (green) photopigments and multiple longwavelength (red) photopigments. The difference in spectral sensitivity among he middlewavelength pigments or among the longwavelength pigments has been estimated to be approximately 5-7nm(Neitz, Neitz, & Jacobs, 1995). A representation of all these pigments is shown in the figure below:

Figure 10. A representation of all the identified cone photopigments.
It is easy to see how two avoided detection for so long--they are nearly identical to two others. [Figure 10 description]

Molecular genetic analyses show that individuals may inherit a surprisingly large number of different X-linked, recessive genes that encode the production of these photopigments (Neitz, Neitz, & Grishok, 1995). A representation of different gene arrangements is shown below. An obvious question is why do we have so many color vision genes? The genes that encode the middle and longwavelength sensitive pigments reside near the end of one of the arms of the X chromosome and they have very similar DNA sequences. In fact, the substitution of one amino acid in the DNA of a photopigment gene is sufficient to cause a change in the spectral sensitivity of that photopigment and in our color perceptions. The location and similarity of these genes makes them susceptible to the kinds of genetic errors that produce multiple gene copies, as well as hybrid genes that are genetic composites of the original ones (Nathans, et al., 1986).

Figure 11. Each rectangle represents a single color vision gene; each row of genes represents an observed gene arrangement. The half red / half green genes represent hybrid genes; linked to "unusual" spectral sensitivity. [Figure 11 description]

At present, it appears that normal color vision results from inheriting at least one cone type from each cone ìclassî (short, middle, and long). It is unclear, however, which complement of genes and cone types result in specific types of color vision deficiency. There is a great deal of genetic variation among individuals with the same type of color defect, making this work difficult. However, it appears that both the type and severity of a color vision defect can be linked to the complement of different cone types in the retina. Hybrid genes, which have been associated with small differences in the spectral sensitivity of the photopigments, are thought to be involved.

These findings lead to an interesting question: if humans possess more than three cone types in their retina, do they still have trichromatic vision? The answer appears to be yes, presumably because the outputs of the different middle or longwavelength cone photoreceptors are summed together before leaving the retina. The resulting signals differ to a small but significant degree across individuals, though, because they affect color perception in some situations. Individuals with different complements of cone pigments will not accept each othersí color matches in the longwavelength end of the spectrum and they will disagree on color names for certain wavelengths of light (Neitz, Neitz, & Jacobs, 1993). For example, a particular mixture of red and green light might appear a perfect yellow to your eye, but appear a greenish-yellow or slightly orange to someone else. This type of color vision assessment, called the Rayleigh Match, is the most accurate method for measuring color discrimination and diagnosing the congenital color vision defects. Some Rayleigh match data is shown below.

Figure 12. The Rayleigh matches of 94 men. The length of the "error bars" represent all the different ratios of red to green light that can be mixed to make an acceptable match to a comparison yellow. Longer bars, therefore, represent poorer color discrimination. The individuals on each end of the figure have color vision deficiencies. [Figure 12 description]

Whereas most of the research to date has examined the color vision of men, the situation for women, especially women who have a color vision deficiency or who have a family history of color deficiency, is even more interesting. Women, by virtue of having two X-chromosomes, inherit two sets of color vision genes. The traditional view proposed that inheriting the normal complement on one or both X-chromosomes led to normal color vision while failing to inherit a normal complement of genes on at least one X-chromosome led to defective color vision (see Mollon, 1992).

However, we now know that early in development women may alternatively express genes from one or the other X-chromosome and that later in development one of the X-chromosomes is inactivated. Consequently, a womanís retina may have regions where the cones reflect the expression of genes on one X-chromosome and other areas where the cones reflect the expression of genes on the other chromosome. A representation of a normal distribution of cone type is shown below:

Figure 13. A representation of the distribution of photoreceptors in the retina. The ratio of R / G / B cone types varies, but the long wavelength cones are the most prevalent; short wavelength cones the least prevalent in the retina. [Figure 13 description]

Women who are heterozygous for the normal complement of color vision genes, therefore, may have a ìmosaicî retina: a patchwork of color-normal and color-deficient regions (Cohn, Emmerich, & Carlson, 1989). The nature of this mosaic depends on the inherited complement of color vision genes and on the point in development that X-chromosome inactivation occurred. That is, some women heterozygous for these genes may develop a color vision deficiency while others may develop normal color vision (Miyahara, Pokorny, Smith, Baron, & Baron, 1998). And, in fact, there are reports in the literature of identical (monozygotic) twins where one twin has normal color vision and the second is color-deficient (Jorgenson, et al., 1992).

In light of these current findings, sensory psychologists and other perception researchers are probably designing psychophysical tasks to try to tease apart the nature of color processing in the eyes of individuals with different complements of cone photoreceptors. The challenge will then fall to neuroscientists, molecular biologists, and others to support or refute our findings at the cellular level.

Future work for sensory psychologists will also involve investigating the extent to which these individual differences in color vision affect our interactions with our world. This knowledge is important because our society uses color to code information in a variety of settings, including education and transportation. In many occupations color discrimination is critical, for example, in discriminating electrical wiring and colored signal lights. While these individual differences are small, they may prove to be problematic in some settings.

In contrast to the exciting research directed at the earliest stages of sensory processing, today there is also substantial research interest at the other end of the S&P continuum: Research directed at higher level perceptual processes and phenomena in the gray area where perception and cognition meld. One area of intense research effort today is the study of visual search: The ability to scan our visual world, quickly distinguish among a variety of objects or forms, and locate and identify a specific target.