ALTHOUGH OUR PERCEPTION of the world seems effortless and instantaneous, it actually involves considerable image processing, as we have noted in many of our previous columns. Curiously enough, much of the current scientific understanding of that process is based on the study of visual illusions.

Analysis and resolution of an image into distinct features begin at the earliest stages of visual processing. This was discovered in cats and monkeys by a number of techniques, the most straightforward of which was to use tiny needles—microelectrodes—to pick up electrical signals from cells in the retina and the areas of the brain associated with vision (of which there are nearly 30). By presenting various visual targets to monitored animals, investigators learned that cells in early-processing brain areas are each sensitive mainly to changes in just one visual parameter, not to others. For instance, in the primary visual cortex (V1, also called area 17), the main feature extracted is the orientation of edges. In the area known as V4 in the temporal lobes, cells react to color (or, strictly speaking, to wavelengths of light, with different cells responding to different wavelengths). Cells in the area called MT are mainly interested in direction of movement.

One characteristic of these cells that may seem surprising is that their activity when stimulated is not constant. A neuron that responds to red, for instance, will initially fire vigorously but taper off over time as it adapts, or “fatigues,” from steady exposure. Although part of this adaptation may result from depletion of neurotransmitters, it also likely reflects the evolutionary logic that the goal of the cell is to signal change rather than a steady state (that is, if nothing changes, there is literally nothing for the cell to get excited about).

How do we know that such cells also exist in humans? Simply put, we descended from apelike ancestors, and there is no reason why we would have lost those cells during evolution. But we can also infer the existence (and properties) of feature-detecting cells in humans from  the results of psychological experiments in which the short-term viewing of one pattern very specifically alters the perception of a subsequently viewed pattern.

For example, if you watch a waterfall for a minute and then transfer your gaze to the grass on the ground below, the grass will seem to move uphill. This illusion occurs because the brain normally interprets motion in a scene from the ratio of activity among cells responding to different directions of movement. (Similarly, the wide range of hues you see on the screen of your television set is based on the relative activity of tiny dots reflecting just three colors—red, green and blue.) By gazing at the waterfall, you fatigue the cells for downward movement; when you then look at a stationary image, the higher baseline of activity in the upward-motion cells results in a ratio that is interpreted as the grass going up. The illusion implies that the human brain must have such feature-detecting cells because of the general dictum that “if you can fatigue it, it must be there.” (This is only a rule of thumb. One of us “adapted” to the dreadful climate and food in England, but there are no “weather cells” or “food-quality cells” in his brain.)

The waterfall effect (or motion aftereffect, as it is also known) was first noted by Aristotle. Unfortunately, as pointed out by 20th-century philosopher Bertrand Russell, Aristotle was a good observer but a poor experimenter, allowing his preconceived notions to influence his observations. He believed, erroneously, that the motion aftereffect was a form of visual inertia, a tendency to continue seeing things move in the same direction because of the inertia of some physical movement stimulated in the brain. He assumed, therefore, that the grass would seem to move downward as well—as if to continue to mimic the movement of the waterfall! If only he had spent a few minutes observing and comparing the apparent movements of the waterfall and the grass, he would not have made the mistake—but experiments were not his forte. (He also proclaimed that women have fewer teeth than men, never having bothered to count Mrs. Aristotle’s teeth.)

The principle of motion adaptation isn’t all that different from the one illustrated by the color aftereffect. Stare at the fixation spot in a between the two vertically aligned squares—the top one red, the bottom one green. After a minute, look at the blank gray screen in b. You should see a ghostly bluish-green square where the red used to fall in your visual field and a reddish square where the green used to be. The effect is especially powerful if you blink your eyes.

This color-adaptation aftereffect occurs mainly in the retina. The eye has three receptor pigments–for red, green and blue—each of which is optimally (but not exclusively) excited by one wavelength. Light that contains all wavelengths and thereby stimulates all three receptors equally yields a ratio that the brain interprets as white. If your red color receptors become fatigued from staring at a red square, then when you look at a field of white or light gray, the ratio of activation shifts in favor of greenish blue, which is what you see.

Orientation adaptation, discovered by Colin Blakemore, then at the University of Cambridge, is another striking example of this phenomenon, except that (like the waterfall effect) it occurs in the brain, not the eye. Stare at the anticlockwise-tilted lines in c for a minute (while moving fixation within the central disk) and then transfer your gaze to the vertical lines in d. You will be startled to find the vertical lines tilted in the opposite direction, clockwise. This perception allows the inference that orientation-specific cells do exist in the human brain: the adaptation to tilt “tilts” the balance of activity among the orientation-specific neurons, favoring those that are attuned to the opposite, clockwise direction.

Even more exciting was Celeste McCollough’s discovery during the early 1960s, while on sabbatical from Oberlin College, of “double duty” cells in humans. Her experiments showed that in addition to cells that respond specifically to a color or an orientation, there are cells that respond only to a line that is both tilted and colored appropriately (that is, a cell for “red line tilted 45 degrees clockwise” or for “green line tilted 10 degrees anticlockwise,” and so on).

Look at e (horizontal black and red bars) for 10 seconds, moving your eyes around the central fixation (don’t keep staring just at the fixation) and then at f (vertical green and black bars) for 10 seconds. Alternate between them about 10 times each. By doing so, you tire all the color receptors in your retina about equally. If you then look at white paper, you see white—no colors. But an astonishing thing happens if you look at g and h, which consist of black and white horizontal or vertical bars. (Move your eyes back and forth betweeen them.) The white horizontal lines now look tinged green and the vertical ones red! The effect is even more striking if you look at the patchwork quilt (i).

Why does this happen? The McCollough effect suggests that subsequent to the retinal processing, some cells in the brain’s visual pathway extract two features along independent dimensions simultaneously. For simplicity, assume there are just four types of these cells: red-vertical, green-vertical, red-horizontal and green-horizontal. Because e fatigues only the red-horizontal cells, you are left with nonfatigued green-horizontal cells, which are then relatively active when you look at white horizontal stripes. Consequently, the white horizontal stripes look greenish; f has the reverse effect on the cells: because green-vertical cells have been selectively adapted, white vertical stripes now appear reddish. But none of these aftereffects occurs when you look at blank white paper because your eye movements ensure that all color receptors are equally stimulated on the retina, whereas cortical cells that have an orientation specificity are not stimulated.

Therefore, with a 10-minute experiment, we have shown the existence of neurons in the brain that require the joint presence of a specific color and orientation to fire. The adaptation effects that result from fatiguing them are called contingent aftereffects. The McCollough effect is an orientation-contingent color aftereffect.

A peculiar aspect of the McCollough effect is that once it has been generated in your brain, it can survive for a long period. Look again next week, and the stripes may very well continue to look red- or green-tinged. (The strength of the aftereffect normally ebbs gradually over time, unless you are submerged in darkness, in which case it endures undiminished!) It has therefore been suggested that contingent aftereffects have more in common with memory and learning than with purely visual adaptation. It is as though during the initial adaptation (or exposure) phase, the brain were saying, “Every time I see horizontal stripes, there’s too much red in the world, so let’s pay less attention to red. Whereas every time I see vertical stripes, I see too much green. So let me damp down the green when I am shown vertical white stripes and damp down red when I see horizontal white.” (In the same way, your brain says, “Any time I set foot into the hot tub, it’s hot, so let me recalibrate my temperature judgment accordingly. I’ll expect it to be hot and won’t withdraw my foot in surprise.”)

It has been shown that certain drugs (including caffeine) can enhance the persistence of the McCollough effect. The phenomenon deserves further study as a way of approaching the neurochemistry of perceptual mechanisms. Visual aftereffects may thus give us insights not only into the neural channels that mediate perception but also into the neural—and possibly pharmacological—basis of memory and learning.