Seeing a color involves making comparisons. "All that a single cone can do is capture light and tell you something about its intensity," Nathans points out; "it tells you nothing about color."
To see any color, the brain must compare the input from different kinds of cone cellsand then make many other comparisons as well.
The lightning-fast work of judging a color begins in the retina, which has three layers of cells. Signals from the red and green cones in the first layer are compared by specialized red-green "opponent" cells in the second layer. These opponent cells compute the balance between red and green light coming from a particular part of the visual field. Other opponent cells then compare signals from blue cones with the combined signals from red and green cones.
On a broader scale, comparisons of neighboring portions of an image lead to our amazing ability to see colors as constants in an ever-changing world.
Nathans vividly remembers demonstrations of this "color constancy" by the late Edwin Land, the inventor of instant photography and founder of the Polaroid Corporation. Land and his colleagues had made a large collage of multi-colored geometric shapes, called a "Mondrian" after its resemblance to the works of the Dutch painter Piet Mondrian. They used three projectors that beamed light matching the wavelength-sensitivity of the three human cone types. With these projectors, the exact wavelength composition reflected from any given patch on the Mondrian could be exactly controlled.
"Land pointed out a patch on the Mondrian that looked orange in the context of the surrounding colors," Nathans recalls. "Then he gave me a tube, like the tube inside a paper towel roll, and had me look at this patch in isolation. And it wasn't orange anymore. It was a perfect red."
The patch was in fact painted orange, but Land had beamed a high-intensity long-wave light from the red end of the spectrum on it so that it reflected a high proportion of red light. Under normal viewing conditions, howeverwhen the patch was surrounded by other Mondrian colorsNathans still saw the orange figure by its true color.
Somehow, by comparing a patch of color to the surrounding colored region, the brain is able to discount the wavelength of the illuminating light and reconstruct the patch's real daylight color.
"Color constancy is the most important property of the color system," declares neurobiologist Semir Zeki of University College, London. Color would be a poor way of labeling objects if the perceived colors kept shifting under different conditions, he points out. But the eye is not a camera. Instead, the eye-brain pathway constitutes a kind of computervastly more complex and powerful than any that human engineers have builtdesigned to construct a stable visual representation of reality.
The key to color constancy is that we do not determine the color of an object in isolation; rather, the object's color derives from a comparison of the wavelengths reflected from the object and its surround. In the rosy light of dawn, for instance, a yellow lemon will reflect more long-wave light and therefore might appear orange; but its surrounding leaves also reflect more long-wave light. The brain compares the two and cancels out the increases.
Land's "Retinex" theory of color visiona mathematical model of this comparison processleft open the question of where in the pathway between retina and cortex color constancy was achieved. This issue could only be addressed by studying the brain itself.
Working with anesthetized monkeys in the 1960s, David Hubel and Torsten Wiesel of Harvard Medical School had shown that the primary visual cortex (V1), a credit-card-sized region at the back of the brain, possesses a highly organized system of neurons for analyzing the orientation of an object's outlines. But in their early studies they found few signs of color-sensitive cells. Then in 1973, Semir Zeki identified a separate area called V4, which was full of cells that crackled with activity when exposed to different colors.
A few years later, Edwin Land paid Zeki a visit in London. "He showed me his demonstration, and I was much taken by that," Zeki says. "I was converted, in fact. So I used his Mondrian display to study the single cells in area V4."
In this way, Zeki discovered that some of the cells in area V4 consistently respond to the actual surface color of a Mondrian patch, regardless of the lighting conditions. He believes these are the cells that perform color constancy.
More recently, with the aid of PET scans, Zeki found an area similar in location to the monkeys' V4 that is specifically activated in humans when they look at Mondrian color displays. The color displays also stimulate the primary visual area and an area that is adjacent to it, V2.
Much controversy exists about all aspects of the color pathway beyond the retina, however. Researchers disagree about the exact role of cells in human V1 and V2, about the importance of V4, about the similarities between monkey and human brains.
To resolve such issues, scientists await the results of further experiments on humans. The new, noninvasive brain-imaging techniques that can show the brain at work may supply key answers. Within a few years, researchers hope, these techniques will reveal the precise paths of the neural messages that make it possible for us to see the wealth of colors around us.
For more information on Jeremy Nathans and the study of vision, see 1997 Holiday Lectures: Senses and Sensitivity.
< Previous | Top of page