Color vision

• 1. Introduction

  ›

  Color vision is not something every mind necessarily builds the same way, and the philosopher John Locke once imagined a case that makes the point vivid. He proposed that a person might experience green wherever someone else experiences red, and red wherever the other person experiences green, with no way for either one to ever notice. That person would still call a stop sign red and a leaf green, because the words would have been learned to match. Nobody has ever demonstrated this inverted spectrum in an actual experiment. But the thought experiment points at something real: color begins as light hitting a small set of cells in the eye, yet it ends as a rich, meaningful experience in the mind, and the gap between those two things is enormous. How does a handful of light sensitive cells give rise to the full range of hues a person can name? Why do other species, from bees to birds to shrimp, seem to inhabit an entirely different visual world? And why does an object appear to keep its color even as the light falling on it changes from moment to moment?

• 2. Splitting White Light

  ›

  In the seventeenth century, Isaac Newton discovered that a prism could split white light into its component colors, and that a second prism could recombine those colors back into white light. The visible spectrum that Newton was working with runs from about 380 to 740 nanometers. Within that range sit the spectral colors: red, orange, yellow, green, cyan, blue, and violet, each corresponding not to a single wavelength but to a band of them, with red spanning roughly 625 to 740 nanometers and violet spanning roughly 380 to 450. Light outside this window is called infrared or ultraviolet, and while people cannot generally see it, many other animals can. The human eye is remarkably sensitive to small shifts within this range. A change of about one nanometer is enough to register as a different hue in the blue-green and yellow parts of the spectrum, though it takes ten nanometers or more to matter in the longer red and shorter blue wavelengths. On top of the few hundred pure hues people can tell apart, mixing and diluting those hues with white light multiplies the number of distinguishable shades far higher. Not every visible color comes from a single band of wavelengths, either. Grayscale tones such as white, gray, and black arise not from hue at all but from the rod cells' pigment rhodopsin reacting to intensity. Pink comes from mixing red with white, brown from mixing orange with gray or black, and navy from mixing blue with black. There are even colors the eye cannot naturally produce: medium wavelength cones cannot fire completely on their own, and if they somehow did, the result would be a shade of green so intense scientists call it impossible. Perceiving white itself does not require the whole spectrum. Combining just two complementary colors, such as blue and yellow, is enough to trigger the sensation, and so is combining red, green, and blue together.

• 3. Cones, Rods, and Genes

  ›

  Rod cells, the retina's detectors for dim light, reach their peak sensitivity near 500 nanometers and play almost no role in color perception. In bright daylight, vision shifts to the cone cells, which peak in sensitivity near 555 nanometers and are entirely responsible for seeing color; this transition between rod-dominated and cone-dominated vision produces the Purkinje effect, a well-documented shift in which colors appear to change relative brightness as light dims or brightens. Humans carry three cone types, conventionally labeled short, medium, and long according to where their sensitivity peaks: the short type peaks between 420 and 440 nanometers, the medium between 534 and 555, and the long between 564 and 580. These labels do not map neatly onto blue, green, and red the way people assume. Microspectrophotometry shows that the long cones, often called red receptors, actually peak in the greenish-yellow part of the spectrum, and the short and medium cones do not correspond cleanly to blue and green either. Cone density is also uneven across the eye. Cones cluster densely at the fovea and thin out elsewhere, so most color information is gathered by looking almost straight at something, and people have surprisingly poor color perception at the edges of their vision, which the brain often fills in from memory and context. The genes behind two of these cone types, OPN1LW and OPN1MW, sit on the X chromosome, and a defect in either produces the most common forms of color blindness. OPN1LW turns out to be highly variable: one study found 85 different variants across a sample of 236 men. Because women carry two X chromosomes, a small percentage may carry two different versions of the long-wavelength gene, and because only one gene is active in each individual cone cell, some of those women end up with four functioning cone types instead of three, a rare condition called tetrachromacy. Even setting genetics aside, the peak sensitivity of a person's cone cells varies somewhat from one individual to the next, and in some other species that variation runs even wider.

• 4. Two Rival Theories

  ›

  Thomas Young and Hermann von Helmholtz argued in the nineteenth century that three cone types, tuned respectively toward blue, green, and red, could account for the whole of human color vision. This became known as the trichromatic, or Young-Helmholtz, theory. In 1872 the physiologist Ewald Hering proposed a different explanation, the opponent process theory, which holds that the visual system reads color in pairs that oppose each other: red against green, blue against yellow, and black against white. Under this model, a single eye can never perceive a reddish-green or a bluish-yellow, in the same way a number cannot be slightly negative and slightly positive at once. One striking piece of evidence for opponency comes from afterimages. Staring at a vivid color long enough to tire the relevant cells, then looking at a plain white surface, produces an afterimage in the complementary color. That effect reveals that cyan, not green, is the true complement of red, and magenta, not red, is the true complement of green, which means the reddish-green that opponent theory calls impossible is actually just the color yellow appearing where it should not be able to. In 1963 the scientist and inventor Edwin H. Land proposed a third framework, the Retinex theory, built around demonstrations of color constancy: an object's perceived color on a complex natural scene stays largely the same regardless of the exact wavelengths bouncing off it, and even the afterimage produced by staring at that object stays the same too, which seems to rule out an explanation based purely on the cones themselves adapting. Land argued instead that the brain assigns color by comparing how brightly each of the three cone systems reads every surface across an entire scene, then judging their relative lightness together. All three theories remain in active use because they describe different stages of the same process, and the signals that give rise to them do not stay inside the eye. They travel onward through a relay of specialized cells toward the brain.

• 5. The Path to the Brain

  ›

  Retinal ganglion cells carry color signals out of the eye, and some of them are built specifically to pit one cone type against another. Midget ganglion cells oppose the long and medium cones, producing a signal that loosely resembles red-green opponency but actually runs along an axis from blue-green to magenta, while small bistratified ganglion cells oppose the short cones against the long and medium ones, running along an axis from yellow-green to violet rather than a strict blue-yellow line. From there, signals travel down the optic nerve to the optic chiasma, the point where the two nerves meet and information from each eye's outer visual field crosses to the opposite side of the brain. Past that crossing point, the visual tracts enter the thalamus and synapse at the lateral geniculate nucleus, a structure divided into zones called laminae: M-laminae built mostly from M-cells, P-laminae built mostly from P-cells, including the midget cells, and koniocellular laminae, which receive input from the small bistratified cells. The signal then reaches the primary visual cortex, known as V1, at the back of the occipital lobe, where color processing grows far more complicated. Many V1 cells respond differently to color depending on how the visual system has adapted to recent light levels, so researchers believe a distinct, smaller population handles color specifically. These color cells, first identified in the goldfish retina by the researcher Nigel Daw, were later suggested to exist in primates by David Hubel and Torsten Wiesel, first demonstrated there by C. R. Michael, and subsequently confirmed by Bevil Conway. Margaret Livingstone and Hubel showed that these double-opponent cells cluster in localized patches within V1 called blobs, split into red-green and blue-yellow varieties that compare local color contrast between adjacent parts of a scene. From the blobs, signals move to a second visual area, V2, where the most strongly color-tuned cells cluster in bands called thin stripes, distinct from the interstripes and thick stripes that handle motion and fine form. V2 then feeds an extended region around V4, first proposed by Semir Zeki as devoted entirely to color, later shown to contain millimeter-sized clusters called globs where the full range of hues gets built. From there, signals reach the inferior temporal lobe, believed to fuse color with shape, completing a chain researchers call the ventral stream, or the what pathway, distinct from the dorsal stream that tracks motion and location.

• 6. Color Across Species

  ›

  The earliest common ancestor of today's vertebrates is thought to have carried four distinct cone pigments plus rhodopsin, making it tetrachromatic by default, and many lineages since then have lost one or more of those pigment genes rather than gained them. The result is a wide spread across the animal kingdom: most skates have no color perception at all, seals, whales, and armadillos see in a single dimension, most mammals and about a quarter of colorblind humans see in two dimensions, most Old World monkeys, apes, and humans see in three, most birds, reptiles, and fish see in four, and a rare few vertebrates manage five dimensions or more, pigeons among them. Among invertebrates the range runs even wider. Honeybees and their relative Osmia rufa see in three dimensions but are largely blind to red while sensitive to ultraviolet, which they use to locate pollen. Papilio butterflies carry six types of photoreceptor and may see in five dimensions. The most elaborate color system yet documented belongs to stomatopods such as the mantis shrimp, which carry between twelve and sixteen distinct spectral receptor types, thought to function less like true color mixing and more like a bank of separate two-color detectors working in parallel. Birds achieve four-dimensional vision through as many as four cone types, each fitted with a colored oil droplet that narrows or shifts its sensitivity, and pigeons may push into a fifth dimension. Reptiles and amphibians typically carry four cone types, occasionally five, and some nocturnal geckos and frogs can perceive color even in dim light. Most mammals besides primates, including dogs and common farm animals, see only in two dimensions, distinguishing blue, green, and yellow but not orange from red, though a handful of species such as cats appear to have partly regained longer-wavelength discrimination through single amino-acid mutations in their opsin genes. Among primates, Old World monkeys and apes see much as humans do, but New World monkeys vary: most males in these species are dichromats while roughly 60 percent of females are trichromats, owl monkeys see in only one dimension regardless of sex, and howler monkeys of both sexes are trichromats. At least one marsupial, the fat-tailed dunnart, has trichromatic color vision of its own. Marine mammals, by contrast, carry only a single cone type, making them monochromats adapted for life in low light.

• 7. Why Color Vision Evolved

  ›

  In primates, researchers believe color vision developed under pressure to solve several visual problems at once: finding nutritious young leaves, ripe fruit, and flowers, spotting predators hidden by camouflage, and reading emotional states in other primates. That pressure appears linked to a broader shift, as the ancestors of monkeys, apes, and humans moved from nocturnal to daytime activity and began eating more fruit and leaves from flowering plants, favoring trichromatic vision able to pick reddish, ripe food out from green foliage. A similar dynamic shows up elsewhere in nature: hummingbirds often identify particular flower species by color alone. Nocturnal mammals, needing far less color discrimination in the dark, tend to have correspondingly weaker color vision, since cones require adequate light to function at all. Ultraviolet sensitivity plays an outsized role across many branches of the animal kingdom, especially insects: bees and other pollinators use it to find nectar, and the plants that depend on them may owe their reproductive success more to ultraviolet patterns invisible to humans than to how colorful their flowers look to the human eye. Birds, too, see into the ultraviolet range between roughly 300 and 400 nanometers, and some species carry markings on their plumage visible only at those wavelengths, used to signal sex to other birds. That sensitivity comes at a cost elsewhere: bees' visible range ends around 590 nanometers, just short of orange, and while birds can see further into red than bees can, they still fall short of the human range. It is a persistent myth that the common goldfish is the only animal able to see both infrared and ultraviolet light; in reality its vision extends into the ultraviolet but not the infrared at all. For birds specifically, ultraviolet vision serves several practical purposes at once: spotting small prey from a distance, navigating, avoiding predators, foraging accurately while flying at speed, recognizing other individual birds, and signaling during mate selection. Beyond vertebrates, only arthropods among land animals are known to share this same capacity for ultraviolet-inclusive color vision.

• 8. A Subjective Sense

  ›

  Color perception can be triggered by things that are not even light. People with synesthesia, sometimes called ideasthesia, report experiencing color set off by sounds or shapes, evidence that color exists as much as a mental event as a physical one. The Himba people offer a documented example of how differently color can be organized across cultures: researchers have found they distinguish close shades of green that most Westerners struggle to tell apart at all, while dividing the spectrum overall into categories unfamiliar to English speakers, including zuzu for dark shades, vapa for very light ones, and buru for vivid blues and greens together, a scheme the Himba have linked to their specific way of life. Context shapes color perception heavily too: the same object can appear to shift hue depending on what surrounds it, even when the light reaching the eye from that object never changes. Timing plays a role as well. Psychophysical experiments have found that people register color roughly 40 milliseconds before they register the orientation of a line, and about 80 milliseconds before they register directional motion, a gap in processing speed detectable only with very brief presentation times. Chromatic adaptation, or color constancy, explains why a white page still looks white under blue, pink, or purple light, even though it is reflecting mostly blue, pink, or purple light back to the eye; the brain compensates using cues from everything else in the scene. Digital tools now have to solve a version of the same problem deliberately. Photo-editing software applies what is called a chromatic adaptation transform to keep a neutral object looking neutral when an image moves between color profiles with different white points, and Adobe Photoshop, for instance, relies on a specific version of that transform known as the Bradford CAT.