THREE PRIMARY COLORS: The traditional paradigm of color suggests that light is a combination of three primary colors: red, green, and blue. Pigments such as those used for printing are combinations of three different primary colors: yellow, cyan, and magenta. While these paradigms of primary colors have worked well for human printing and light uses for over a century, it is likely that the three primary colors are not descriptive of the world, but rather an artifact of our eyes, the tools we use to perceive the world. In reality, there is an infinite number of colors just as there are an infinite quantity of real numbers on a number line. The cone shaped light detectors in human eyes are sensitive to three different color ranges, that we call red, green, and blue. If the human eye had cones that respond to additional color ranges, then additional primary colors would be necessary to produce all colors that humans see. A paradigm based on light spectra may match reality better.
HOW OUR EYES DETECT COLOR: Each cone is sensitive to a different range of colors. One is sensitive to primarily red, a second to green, and a third blue. The three curves in Figure 1 crudely represented by the sensitivities of the three types of cones. The three sensitivity ranges actually overlap so Figure 2 might be a better representation. (See below for precise curves). Each cone sends a nerve pulse to the brain at a rate proportional to the intensity of the light which that cone detects. That nerve pulse carries no information about the actual color of the light, only the brightness. By integrating the signal rates sent from three kinds of cones, the brain attempts to infer what color and brightness of the light must have been. For example, if only the R cone detects light, the brain interprets the color must be red. If R detects a little light, but G detects more light, the brain interprets the nerve signals to mean that the color must be yellow.
TRICKS ON OUR EYES: The brain can easily be fooled. For example a television image of a rainbow actually contains only three specific colors of light, red, green, and blue, one for each type of cone. Yet our brain common interprets different intensities of these three colors to mean that the infinite variety of colors of a real rainbow are present in the image. Only the three primary additive colors are necessary in televisions, computer monitors, and colored image projection systems to convince our brain of the detection of any of the infinite variety of visible colors. For example if a television images shows both red and green lights close together, you brain will falsely believe that yellow light was present instead. Each color of light need not be a precise wavelength to create such a false image, but only be within the range that a particular cone is primarily sensitive.
Figure 3 shows an approximation of the integrated (i.e., "added up") sensitivities of the three types of cones. If the full visible spectra comes from the same direction, the human brain interprets the color to be what we call "white." Isaac Newton first demonstrated that white is actually a combination of all the visible colors by using two prisms to separate then recombine the light. When Newton blocked some colors where the light was separated between the two prisms, some color other than white resulted where the light was recombined by the second prism.
HOW WE SEE PRINTED AND PAINTED COLORS:
The subtractive colors used in painting and printing can be understood by adsorption of light. A pigment that adsorbs light primarily from the red end of the spectra will reflect the remaining light that the eyes and brain will average and interpret as cyan. (See Figure 4.)
A pigment that adsorbs light the middle portion of the visible spectra will appear magenta, the average of the remaining reflected light.. (Figure 5)
And a pigment that adsorbs light from the blue end of the spectra will reflect colors that center on the yellow (Figure 6) and therefore appear yellow to human senses.
A mixture of all three pigments will adsorb nearly all visible colors, appearing nearly black.
A BETTER PARADIGM: A better understanding of the primary colors may come from a considering the (remote?) possibility that with the advent of artificial light sources that emit significant amounts of ultraviolet light, human eyes might evolve (perhaps over thousands of years?) to contain a fourth type of cone sensitive to the ultraviolet light. (See figure 7.) To reproduce this entire spectra would then require FOUR primary colors, one producing light for each of the kinds of cones!
With an additional kind of cone, the primary subtractive colors of pigments will not be the same. A pigment that absorbs only the ultraviolet light will reflect all the colors that we currently see, so such a pigment could continue to be called "white" although it no longer would contain all of the new visible colors.
If the color matching the name "white" is not changed, then a new name will be needed to describe light that is detected by all four cones, perhaps "allcolor."
The pigment that adsorbs the red portion of the spectra, previously called "cyan" (see figure 8), will reflect light now centered on blue. Thus the pigment perceived as "cyan" by three cone people would be perceived as "blue" by those people having four types of retina cones.
What is presently called "magenta", when containing additional ultraviolet light (as shown in Figure 9) might seem "cyan" to those with four cones.
What is currently called "yellow" pigment (Figure 10), might seem "green" to those with four cones.
In short, it is likely that the colors of we perceive from paints and inks depend on our eyes as well as the nature of the absorption of the pigments.
SUMMARY: Today's primary colors are not in reality fundamental properties of light, but are instead consequences of the number of cones in human eyes. The human brain can be easily tricked to believe it is detects colors not actually present. Our understanding of color might be better if we considered the spectra of light and which colors are absent or present in any particular situation.
Details: The above diagrams are intended to explain vision using the concept of a spectra containing an infinite variety of colors. The actual sensitivities of the cones is more complicated. More accurate representations of the sensitivities of human retina cones are presented in graphs plotting measured % sensitivities and (perhaps more accurately) log of sensitivities. The above explanation remains accurate for the more accurate graphs.
Cone sensitivity data was obtained from the UCSD Color and Vision database which is apparently no longer available online.