IF SOMEONE SHOWED you a caricature of Richard Nixon—a man’s face with oversize shaggy eyebrows, a bulbous nose and pronounced jowls—you would probably recognize the former president immediately, even though the drawing is not true to life. A cartoonist creates such a sketch by taking the average of many male faces and subtracting it from Nixon’s face, then amplifying those distinctive differences. To an observer, the result looks more like Nixon than Nixon himself. Why is it that our brains respond so intensely to extremes?

When the cartoon’s “Nixon-ness” jumps out at you, you are experiencing what scientists call “peak shift.” To understand the concept, imagine, for argument’s sake, that you want to teach a rat to distinguish a rectangle from a square. It’s quite easy to do. Simply give the animal cheese every time it picks the rectangle, and it will soon learn to select the rectangle every time. Once the rat has developed this preference, let’s say you show it a longer, skinnier rectangle. Inevitably, you will find that the rat prefers the exaggerated one to the original. What the rat has learned to recognize is not a particular rectangle but rather rectangularity itself: the more rectangular the better. The savvy rodent looks at the longer, skinnier quadrilateral and goes, “Wow, what a rectangle!” In scientific parlance, the rat’s “peak response”—its strongest reaction—has shifted away from the original—hence the term “peak shift.”

The sway that exaggerated characteristics hold over us is a special kind of illusion—and a powerful one, we believe. In the five years that we have been writing about perception for this magazine, we have described a range of illusions, from geometric patterns that seem to move because they activate our motion perception systems to optical tricks that arise because each of our eyes sees the world from a slightly different position.

Now we would like to make a daring suggestion: that illusions are not merely fascinating windows into our minds and the way we perceive the world. They help to drive the most powerful force that shapes life on earth: evolution.

The standard theory of evolution is that animals that randomly inherited genes that produced beneficial traits—in the case of the giraffe, a longer neck, which made it easier to reach tall acacia trees—ate better, reproduced more often and passed these gene variants to their offspring. Hence the progressive lengthening of the giraffe’s neck across successive generations.

What we are proposing is yet another mechanism of evolution. Our hypothesis involves the unintended consequences of aesthetic and perceptual laws that evolved to help creatures quickly identify what in their surroundings is useful (food and potential mates) and what constitutes a threat (environmental dangers and predators). We believe that these laws indirectly drive many aspects of the evolution of animals’ shape, size and coloration.

Let’s return to the giraffe. Giraffes need to recognize and mate with others of their own kind—and not, say, with antelopes or okapi. Wired into the animals’ visual centers is a recognition system that automatically prefers mates that have more “giraffelike” characteristics. In this formulation, the longer necks were selected not because of any functional reason but simply because in scanning for desired traits, the visual system lights upon exaggerated ones first. They stand out, like Nixon’s prominent brows. Across successive generations, the long neck would have become an ever more reliable species marker for giraffeness, thereby enabling a partner to be spotted even from a great distance.

Our theory is not intended to replace Charles Darwin’s but to point out that other powerful forces besides the natural selection of fitness-conferring genes may be involved. Darwin, of course, acknowledged as much when he observed that mating behavior—so-called sexual selection—can exert its own, often maladaptive, impact on evolution. Because female peacocks prefer males with large tails, big-tail genes multiply in the population, eventually culminating in modern peacocks’ magnificent but absurdly impractical tails.

The aesthetic theory of evolution that we are proposing also revolves around mating behavior, but it differs from sexual selection. For one thing, sexual selection only explains why secondary sexual characteristics in males (the peacock’s tail, the rooster’s wattle, the unwieldy antlers of the Irish elk) become exaggerated. The peak-shift effect, in contrast, helps to explain extreme traits and behaviors that pertain to all members of a species (both male and female giraffes must identify potential mates, which helps to explain why both genders have long necks).

Because humans (taxonomists included) are diurnal and hence visual creatures, we tend to place a strong emphasis on appearances. But the principle of peak shift can apply as easily to nonvisual signals. For nocturnal critters such as rodents that use smells to find mates and interpret their world, attractions to strong scents could drive evolutionary change. These changes would be hard to see but just as real. If dogs were taxonomists, the evolutionary trees in their textbooks would look very different from ours.

The Gull Chick Principle
Other rules of aesthetics besides peak shift can also be invoked to explain the astonishing diversity of species. One is what we call the “gull chick principle.”

Niko Tinbergen, a pioneering investigator of animal behavior, experimented with herring gulls 50 years ago, but the relevance of his work to evolutionary theory has not been widely appreciated. The adult herring gull has a long, yellow beak with a red spot near the tip. As soon as a chick hatches, it starts pecking at this spot, which triggers the parent to regurgitate food into the chick’s mouth. How does the chick recognize its mother? Tinbergen found that it doesn’t: chicks will peck as intently at a disembodied beak.

Why is a beak sufficient? The purpose of vision is to identify and interpret objects and events while expending the least amount of mental processing power. Through millions of years of evolution, the chick’s brain has acquired the wisdom that this long thing with a red spot always has a mother, not an inquisitive ethologist, attached to it, and it makes an interpretive shortcut.
Tinbergen next found that a beak is not even required. He held out a long, yellow stick with three red stripes on it, and the chicks pecked it—more, in fact, than they would have pecked at a real beak. Tinbergen had stumbled on a superbeak!

Why does this happen? Clearly, there are neural circuits in the visual pathways of the chick’s brain that are specialized to detect the red spot on a beak as soon as the chick hatches. Perhaps the neurons’ receptive field embodies a rule such as “the more red contours the better.” So even though the stick does not look like a beak—maybe not even to the chick—this strange object is more effective than a real beak at activating the bird’s beak-detection system. Hence, we  predict that a species of gull will emerge that has two or even three red beak stripes instead of just a bigger red splotch. Another, even more striking example of the gull chick principle is the idiosyncratic preference (demonstrated in the lab) that guppies show for potential mates that have been painted blue—even though in nature guppies are not blue. Again, we anticipate the emergence of a new species: the blue guppy. It’s not often in evolutionary theory that one can make such specific predictions.

The gull chick principle may apply widely, because the visual system of every animal is wired to use specific characteristics to identify others of its species. If a potential mate diverges from the standard in a way that more optimally excites “species-identifying” brain circuits, the genes that promote such supertraits will flood the population. Unlike the peak-shift principle, no obvious parameter is being exaggerated (such as a long neck); the changes in appearance are selected because of idiosyncratic aspects of neural wiring. Even the florid, almost comical, exaggeration of dance rituals in some bird species may be influenced by this principle.

Marian Stamp Dawkins, an animal behavior expert at the University of Oxford, has championed the idea that aspects of sensory processing can influence the evolution of communication signals; for example, a nocturnal species whose predators are color-blind would not evolve colored warning splotches. Our idea complements hers but takes it further, by arguing that higher-order principles of perception may also play a role.

Another principle that may affect evolution is known as grouping. The visual system has an obsessive desire to make whole objects from fragmentary evidence—such as a lion largely obscured by leaves and shadows. Like-colored fragments are interpreted as bits of a single object that is partially hidden by another, closer object. As naturalists have long recognized, this tendency is cunningly exploited by reef fish, which evolved bold colored splotches that “break” their outlines and confuse predators seeking continuous contours.

Proofs of Concept
If perceptual laws influence the development of species, what would evolutionary biologists expect to see? For one, the progressive “caricaturization” of easily recognizable physical traits over time. And indeed, such trends are commonly seen in the evolution of mammoths, ankylosaurs, titanotheres and other animals.

Another prediction from the theory is that unseen parts—internal organs—would not be subject to perceptual selection pressures and hence should diverge considerably less. Overall, this appears to be true. A rhesus monkey’s liver doesn’t look much different from a human one.

Finally, because plants do not have sophisticated sensory systems, they should vary less in appearance than animals do, except when selection has been done for them by animals. This would explain why leaves and trunks look much alike, whereas flowers, which “compete” to be visited by insects and hummingbirds, are stunningly conspicuous and variable. There is even one species, the bee orchid, whose flower perfectly resembles a caricature of
a female bee—a superbee—to attract pseudocopulation and cross-pollination by male bees.

Ultimately, our hypothesis is not a mechanism outside Darwin’s theory but an unexpected interaction within it. His principle of natural selection leads to the emergence of brain mechanisms that enable an animal to quickly detect healthy sexual partners of the same species. But inevitably these cognitive processes have side effects. They evolved to increase a species’ fitness but may lead to perceptual quirks that do not promote fitness—and may even work against it. Thus, the study of visual illusions—and the laws they exploit—offers clues to certain otherwise mysterious trends in evolution.