The Folly of Fools: The Logic of Deceit and Self-Deception in Human Life - Robert Trivers (2011)

Chapter 2. Deception in Nature


Before we take a deeper look at self-deception, let us examine deception in other species. It is often easier to see patterns of importance if we cast our net of evidence widely—in this case, to include all species, not just our own. What can we learn about deception by viewing it in an evolutionary context? The evolutionary approach to deception is to study deception in all its forms while looking for general principles. So far, the forms of deception turn out to be very numerous and the principles very few. Deception hides from view, so its secrets often have to be pried out by meticulous study and analysis, of which, fortunately, there has been a lot, and several important principles have emerged that apply across species. First, there is a tremendous premium on novelty that in turn generates an enormous variety of deceptive ploys. Since novel tricks—almost by definition—lack defenses against the tricks, they usually spread quickly. This is the beginning of a so-called coevolutionary struggle between deceiver and deceived, acted out over evolutionary time. This struggle leads to complexity on both sides—to the evolution of bizarre, intricate, and beautiful examples of deception, as well as the ability to spot them. In general, but especially in birds and mammals, this evolutionary struggle also favors intelligence on both sides. Consider the simple matter of picking out an object against a background. If the object has not been selected to match the background, it should be easy to detect, differing in numerous random details. But if there has been selection to match, detection is an entirely different matter. Selection will have obliterated many of the random mismatches, leaving a much more complex cognitive problem for the observer to solve.


The most important general principle is that deceiver and deceived are locked into a coevolutionary struggle. Since the interests of the two are almost always contrary—what one gains by perpetrating a falsehood, the other loses by believing it—a struggle (over evolutionary time) takes place in which genetic improvements on one side favor improvements on the other. One key is that these effects are “frequency dependent”—deception fares well when rare and poorly when frequent. And detection of deception fares well when deception is frequent but not when it is rare. This means that deceiver and deceived are locked into a cyclic relationship, in the sense that neither can drive the other extinct. Over time the relative frequencies of deceiver and deceived oscillate, but they do so within bounds that prevent either from disappearing. Likewise, in a verbal species like our own, we will be warned about new tricks more often by others the more frequent the tricks become. Note that no role is exclusive to some and not others—all of us are both deceiver and deceived, depending on context.


You don’t have to look far to find evidence of frequency-dependent selection in systems of deception between prey and their predators. For example, in model/mimic systems, such as are found in butterflies (and snakes), a distasteful or poisonous species (model) evolves bright coloration to warn predators that it is distasteful. This selects for mimics, species that are perfectly tasty and harmless but gain protection by resembling the model. In West Africa, there is a genus of butterfly that is distasteful and as many as five species of the genus, all differing in coloration, may be found in the same forest. It turns out that there is a single species capable of mimicking all five model species. That is, females of the mimetic species can lay five kinds of eggs, each of which grows up to resemble one of the poisonous species.

This unusual system of mimicry provides striking evidence of frequency-dependent selection. Here, one species is delicious but mimics any one of five related poisonous species. These differ in color and pattern, and so do their respective mimics. When several poisonous species are found in the same forest with their mimics, the frequency of each mimic within this species matches the frequency of the model among its group of related, distasteful species. This could have been brought about only by frequency-dependent selection, where each mimetic form loses value when it becomes too common relative to its own model. If all the tasty butterflies looked the same, the predatory birds would rapidly specialize on that one form, decimating it.

One implication of frequency dependency is a perpetual premium on novelty. Indeed, in the above example, novel forms are more common in the mimetic species the more it outnumbers its model. That is, the more frequent the deceivers are, the more they begin to diversify, the better to avoid detection. Every new deception, by definition, starts rare and thereby gains an initial advantage. Only with success will one’s disguise become part of the backdrop against which another novelty can begin rare and flourish. We can also see how easily break-off forms in the mimic might happen to resemble a second poisonous species, leading to two forms mimicking two species.


A very rich illustration of coevolutionary principles is found in the relationships between brood parasites and their disadvantaged hosts, especially in birds but also in ants. A surprising percentage of all bird species, about 1 percent (usually cuckoos and cowbirds but also including a species of duck), is entirely dependent on other species to raise their young. Naturally this arrangement is rarely to the advantage of the “host” birds, who may end up raising unrelated young in addition to their own—or worse still, as is often the case, unrelated young instead of their own. This particular host/parasite relationship has been studied in unusual detail. Indeed, it is mentioned almost as early as human writing permits, some four thousand years ago in India, later described by Aristotle, and recently studied intensively by very clever field experiments designed to tease apart how the relationship works.

The first move is for the deceiver to lay one of its eggs in the victim’s nest. This selects in the victim for the ability to recognize a strange-looking egg and eject it. This, in turn, selects for egg mimicry in the brood parasite—the tendency to produce eggs that have the same spotting and coloration as the eggs of the species whose parental care is being borrowed. Some parasitic species lay in the nests of multiple species, with individual species specialized to lay eggs that match in coloration the eggs of the species in whose nest they are laying. It is now advantageous for the host to be able to count total number of eggs, and reject nests with one too many. This is especially valuable if the parasite’s young hatches before those of the host, ejecting all of its eggs so as to monopolize parental investment, leaving the host no offspring of its own to rear. Better for the host to start over. This selects for parasites that remove one egg for each one laid, leaving total number the same, and the egg is eaten or moved some distance from the nest, perhaps to hide the crime.

Once the egg has safely hatched, selection may favor brood-parasite mouth colors that resemble those of the host species, since parents feed more strongly mouth colors that resemble those of their own species. Within their own brood, evidence from other birds suggests that mouth color may be brighter for healthier chicks, so it is interesting that brood parasites make their mouth colors especially bright. By pushing out its foster siblings, the host young can monopolize parental investment, but since parents adjust their feeding to the total begging calls they hear, a single cuckoo chick may evolve to mimic the calls of an entire brood of the host. In an even more bizarre twist, a species of hawk cuckoo that parasitizes a hole-nesting species in Japan has evolved inner-wing patches that resemble the throat coloration of its host, so that when begging for food, a chick can flap its wings and simulate the begging of three offspring instead of one. The wing patches are even occasionally fed, a case of deception being too convincing for its own good.

A very important selective factor is errors in recognizing a host’s own offspring—so-called false positives—that are an inevitable feature of any system of discrimination (see spam versus anti-spam in Chapter 8). For weak systems of discrimination, a host rarely rejects itself, but it is fooled too often into accepting cowbird chicks. Stronger systems of discrimination cut down on the host’s loss due to the cowbirds but also impose a cost on the host, as it inevitably accidentally rejects its own offspring more of the time. In reed warblers, parents learn the appearance of their own eggs and then reject those differing by a certain amount. If their nests are parasitized about 30 percent of the time, it makes evolutionary sense for them to reject strange eggs, but if they are parasitized less often, the cost in destruction of their own eggs is too great. Sure enough, reed warblers are parasitized only 6 percent of the time in the UK and do not reject new eggs—unless a cuckoo is seen near the nest at about the right time (perhaps pushing probability above 30 percent). In one population, a drop in parasitism rate from 20 percent to 4 percent was matched by a one-third reduction in rejection rate, an effect too rapid to be genetic, so reed warblers probably often adjust their degree of discrimination to evidence of ongoing brood parasitism.

Note the important frequency-dependent effect. When almost all eggs are their own, discrimination will result in the warblers destroying some percent—say, 10 percent of their clutches—with only rare gain. But at 30 percent parasite frequency, they risk harming themselves only 7 percent of the time, while with perfect discrimination, they save themselves a substantial cost (in nurturing other species almost 30 percent of the time). At low frequency, deceivers are hardly worth detecting—only at high frequency are important defenses expected to kick in.

There is one striking peculiarity in the entire system. Birds repeatedly fail to evolve the ability to see that the cuckoo or cowbird chick bears no resemblance to their own chicks beyond mouth color and begging call. In size, a cuckoo chick is often six times or more larger than its host, so that a foster parent may perch on the shoulder of the chick it is about to feed. Since it would seem beneficial to note this absurd size discrepancy and act accordingly, why are birds, in species after species, unable to do so? The answer to the mystery is by no means certain, but there are some interesting possibilities. Failure to make the appropriate discrimination happens preferentially in species in which the brood parasite ejects its foster siblings before they hatch. Thus, if the parent learns the appearance of its own chicks by imprinting on the first ones produced, this will work fine if the first brood is its own, but it will prove fatal if at their first attempt they are parasitized. The host will imprint on the brood parasite and kill its own young whenever it sees them. This will wipe out the host’s entire lifetime reproductive success, since it will now see all of its own chicks as foreign.

More generally, some of the brood parasite’s characteristics are super-optimal from the foster parents’ standpoint. We expect parents often to favor the larger of their chicks as being healthier, stronger, and more likely to provide a good return on investment. This may make foster parents vulnerable to implausibly large chicks that nevertheless release the bias that bigger is better. More to the point, many brood parasites have evolved begging calls that are louder than the host’s and hence presumably harder to resist. Likewise, parasite mouth colors are especially brightly colored. These signals are less costly to magnify than is body size.

There is yet another explanation for hosts’ not discriminating against obvious mimics—fear of the consequences. “Mafia-like” behavior has been described in a couple of bird species, in which a cuckoo or cowbird punishes those hosts who eject their eggs by destroying their entire nest. It becomes a matter of accepting a degree of parasitism or being really badly treated—like a demand payoff (tax) instead of an outright killing. Good evidence from one system shows that accepting the mafia tax leads to greater reproductive success than fighting it—and ending up with a destroyed nest.

Recently it has been shown that there is something resembling cultural transmission of knowledge regarding brood parasites. At least reed warblers can learn from the mobbing behavior of their neighbors toward models of cuckoos (whereas they do not bother to learn from induced mobbing toward innocuous species, such as parrots). Warblers are attracted to the sound of nearby mobbing and approach to observe. If it is a cuckoo being mobbed, they are more likely to approach quickly a model of a cuckoo in their own territory and to mob it. This social learning permits a much more rapid spread of defenses against brood parasites than can occur through genetic change alone. A brood parasite has also evolved to resemble a local hawk, and this resemblance reduces the degree to which it is be mobbed by potential hosts.

Birds are not the only group subject to brood parasites. Ants spend an enormous amount of energy raising large broods that are highly attractive as nurseries to other species. There are as many species of social parasites on ants as there are ant species (about ten thousand of each). Even though the nest may be fiercely defended, parasites have ways of gaining entry, usually by mimicking some part of the ant’s communication system. Caterpillars of one butterfly species manage to get into an ant’s nest by curling up in a ball and emitting the smell of ant larvae. They are then carried into the ant nest, where they imitate the sounds of a queen ant, the very sounds that lead actual queens to be preferentially fed and protected. When food is short, workers will feed young larvae to the pseudo-queens and, when the nest is disturbed, will rescue them over ant larvae. The caterpillars are even sometimes treated as rivals by the real ant queen. This is another example of a deception being too effective for its own good. These kinds of relations have been described for dozens of butterfly species that parasitize ant nests.

In sum, each move is met by a new countermove, resulting in principle in an evolutionary struggle that may last millions and millions of years. This is especially true of relationships between different species, where issues of relatedness no longer apply, but it may be true of many similar relationships within species as well. The two sexes, for example, are partly cooperative and partly in conflict, with move being matched by countermove, locking them into a tight frequency-dependent relationship that usually stabilizes at equal number of the two sexes (see Chapter 5).

Deception can be beautiful, complex, and very amusing. It can also be very, very painful. To be victimized by systematic deception in your own life can cause deep pain. Even watching another species victimized by deception can sear your heart. Every spring in Jamaica, I watch a few dove couples trying to reproduce by raising their young in my trees. These are birds I love to watch, and I wish them every success. Emerge from stage left the anis—large, black, ominous-looking birds that prey on the nestlings of other birds, eagerly gobbling up the chicks. Arriving in groups of about six to twelve, the anis are noisy and fast-moving, saturating their terrain and relying on one heartless trick. One ani gives a loud call that mimics the generic chick begging call of other species—a plaintive kind of squawk that the chick is most likely to give when it is hungry and the parent is nearby. Now the victim chick hears the ani’s begging call and promptly begs itself, the better to outcompete its imaginary sibling. The ani (or one of its group members) makes a beeline to the chick and gobbles it up, along with any other nestlings. Your heart goes out to the victim, fooled by its naive tendency to beg when it hears a begging call. Or worse, you suffer for its silent and completely innocent siblings, only fated to have a fool in their nest. I spent one long evening flinging stones at anis who were about to devour a nest they had detected through this deception. They stayed nearby overnight and consumed the nest contents first thing in the morning.


Deception spawns the mental ability to detect it. In the above case, this includes the ability to discriminate very similar objects, the ability to count, the ability to adjust discriminatory powers to contextual factors, and the ability to act as if making multiple inferences: eggshells on ground, egg destroyed, nest parasitized, investment best curtailed, and so on.

These improved intellectual abilities select for more subtle means of deception, which, in turn, select for greater abilities to detect the deception. In short, deception continually selects for mental ability in the deceived. Since the target of apprehension is a moving target—that is, evolves away from your ability to detect it—ever-new discriminations proliferate. The ability to see through a deception requires special talents unnecessary for discriminating a target that has no ability or interest in hiding. Thus, deception has probably been a major factor favoring intelligence, certainly in highly social species.

Intelligence also helps deceivers. In behavioral deception, intelligence presumably increases the range and quality of the deception displayed. In humans, at one extreme, the behaviorally retarded will largely be limited to nonverbal forms of deception—a lunge in one direction when the opposite is intended—but rarely sophisticated patterns of verbal deception. By contrast, the very bright can lie in multiple dimensions. Thus, deception selects for intelligence on both sides, though more reliably on the perceptual side. For example, a moth’s back comes to more and more exactly represent tree bark. This requires no new mental abilities on the part of the moth but implies growing powers of discrimination in its visual predators, such as birds and lizards. Not so for behavioral deception.

The best evidence for a robust role of intelligence in deception comes from a study of monkey and ape brains. The size of the neocortex (so-called social brain)—or better still, its relative portion of total brain—is positively associated with the use in nature of tactical deception, which includes any kind of deception that can be seen to give an advantage. The relative size of the neocortex is, in turn, a good measure of relative intelligence, especially social intelligence. Scientists used published studies of monkey and ape behavior in nature to assemble a large set of examples of deception, then solicited a still larger sample of unpublished studies. They next made sure the evidence was not biased by group size, or degree to which a species had been studied, or applied only to some monkeys and apes but not others. The strong conclusion was that among monkeys and apes, the smarter the species, the more often deception occurs. So perhaps does self-deception. We shall see later that the brighter children are, for a given age, the more often they lie. The importance of this can’t be overemphasized. We often think that greater intelligence will be associated with less self-deception—or at least intellectuals imagine this to be true. What if the reverse is true, as I believe it is—smarter people on average lie and self-deceive more often than do the less gifted?


You would think that telling the sexes apart would evolve easily and reliably, but in an extraordinary number of cases, one sex imitates the other (or the same sex of another species). In each case, females are being mimicked, as in the following three examples. In many groups of fireflies, particular species have evolved to prey on others by sexual mimicry. A predatory female of one species responds to the courtship flash of a male of another species by giving not her own flash of interest but that of a female of his species. He turns toward her, expecting to enjoy sex, and is seized and eaten instead. Sex is a very powerful force and especially in males often selects for “indiscriminate eagerness,” which provides fertile ground for deception to parasitize.

In another example, that of orchids, fully one-third of all species are pollinated through deception—that is, the plant offers no actual reward to its pollinators, only the illusion of one. Most species mimic the smell of their pollinators’ food without supplying any. A smaller number (about four hundred species) mimics an adult female of the pollinator species in both appearance and smell, so as to induce pseudo-copulation by the aroused male. The plant takes care not to give the male a full copulation with ejaculation, presumably to keep him in a perpetually aroused state, driven to seek out new “female” after new “female,” pollinating the flowers all the way. Males who find pseudo-females do not linger and test nearby flowers as do males in plant species that have just given a nectar reward. Instead they fly immediately to a new patch of flowers, presumably in search of actual rewards. Thus, sexual mimics tend to be more outbred than closely related species that offer a real reward—a side effect of being deceived that may actually benefit the species itself.

Selection has also repeatedly favored males who mimic females within their species to fool territorial males into thinking they are females so they can get close enough to steal paternity of some or all of the eggs about to be laid by real females. These eggs will be cared for by the territorial male as his own. Sometimes selection for deception has been strong enough to mold morphs that are permanently committed to deception, that is, morphological forms whose strategy depends entirely on a life spent deceiving others. A classic example occurs in the bluegill sunfish, where a specialized male form has evolved that mimics a female in appearance and behavior, being one-sixth the size of a territorial male and roughly the size of an actual female. This female-mimic seeks out a territorial male, permits himself to be courted, and responds enough to keep the other male interested, so that when a true female spawns, the pseudo-female is ready nearby to help fertilize the eggs. It is as if the territorial male imagines he is in bed with two females when in fact he is in bed with one female and one male. The female almost certainly knows the truth.

The two kinds of males appear to be distinct forms that never turn into each other. To have persisted for so long, their long-term reproductive success must be identical—that is, over evolutionary time, the deceiver is doing exactly as well as the deceived—and this equality must, in turn, be enforced by frequency-dependent selection. When the female-mimic is relatively rare, he will do relatively well; when common, less so. Whether the female expresses any kind of preference for either male is unknown, but in general, females prefer rare males, that is, the less frequent of two choices. Perhaps one of the most spectacular cases of sexual mimicry is performed by a tiny blister beetle, itself a parasite on a solitary bee. To achieve dispersal, one hundred to two thousand individuals aggregate in groups that mimic in size, color, and perching location a single female of the host bee species, even moving as a unit up and down a tree. So here a kaleidoscopic falsehood is produced, its individual parts one-hundredth or less the size of the picture they are creating. In turn, a male bee copulating with the picture will serve to disperse the beetles to future bee nests since the beetles attach to him.


Alarm calls occur in a variety of species, especially birds, and serve to warn other individuals (often relatives) that a predator is nearby. An alarm call is obviously a key moment—with little room for error on the receiving end. Thus, it is not surprising that true alarm calls have served as a template for the repeated evolution of false alarm calls. In mixed-species flocks of birds found in the tropics, an individual will give a false warning call when another bird has caught and is about to eat a large, tasty insect. Half the time, this causes the bird to drop the insect and dive for cover. In the other half of the cases, the bird is not fooled—while it always responds to a true alarm call with immediate flight. Thus, the birds have evolved to tell false from real alarm calls half the time.

In skuas, false warning calls by parents are used to frighten warring offspring into separating and fleeing for cover, at which point the parents intervene to prevent further strife. In swallows, males apparently use false alarm calls to guard their paternity. They will give an alarm call when they spot their mate near another male, often causing both birds to dive for cover. Males breeding in colonies almost always give such calls when returning to an empty nest during egg laying (when female copulations outside the pair are frequent and threaten his paternity of the offspring) but not at other times (even swallows do not wish to cry “wolf”). Antelopes have been discovered playing the same trick. After a male has spent a day or two in sexual consort with an adult female, he will give a warning bark if the female seeks to move on, as if signaling that a predator lurks nearby and she should remain with him.


Camouflage is so common in nature as almost to escape notice. Most creatures are selected at the very least to blend in to their backgrounds, with stick and leaf insects merely extreme examples. But at the behavioral level, octopuses and squid are so advanced as to be worth special note.

Octopuses and squid are fat, tasty creatures without a protective shell, so they are naturally sought after by a wide range of predators, mostly fish but also mammals and diving birds. Their only defense (beyond ink clouds and biting) is camouflage, and here they have evolved a remarkable system in which each skin-color cell is innervated by a single neuron, thus cutting out all synaptic delays and permitting a near-perfect adjustment to the background in about two seconds. While feeding, the animal can move very slowly across a great range of backgrounds, continuously remaining nearly invisible to others by adjusting its color to each new surface—sand, mud flats, coral reefs, rocks, sea-grass beds, and so on. Octopuses look as if they are slowly rolling while continuously adjusting to what is below. When they want to swim fast, they mimic flounders, in shape, color, swimming movements, and speed, darting swiftly along the sea bottom.

At intermediate speeds (when foraging), they adopt a most unusual strategy of randomly displaying variant phenotypes at about the rate of three per minute, for hours at a time, as if they are shuffling through a deck of cards featuring different camouflaged versions of themselves. This helps prevent the predator from forming a specific search image for any particular version. Just as the predator recognizes potential prey, the prey has morphed into a novel camouflaged form. One species of squid has also evolved a female mimic, one so good that he sometimes fools even fellow female mimics, who approach in search of copulation. This is yet another case of deception being too convincing for its own good.


It has long been known in predator/prey relations that deception can work anywhere from first detection until final consumption. Consider two examples near the time of death itself. The feigning of death typically occurs after the prey is caught, and is thought to inhibit the final death-dealing strike. The bird acts dead, lifeless, but remains conscious and alert so that often the only sign of life is its open eyes. Chickens run at the first opportunity, typically when the predator lets go, but a duck threatened by a fox often remains immobile for some time after release, especially if other foxes appear to be present. The fox’s counteradaptations are to kill some prey immediately upon capture and to disable the remaining ones by severing a wing on each.

In the broken-wing display, a bird near its nest tries to distract a potential predator by acting like an injured bird, with one broken and extended wing. The bird moves awkwardly near the predator with wing extended but flies away quickly when attacked. This display is much more dramatic the closer the predator is to the nest. Birds have a variety of other acts they conduct when their nest is threatened. Crakes, ground-nesting birds, will mimic rats scurrying away from their nest, their backs slightly hunched, with both wings partly open and drooping to mimic a fat rat scurrying away in the wide open—an easy prey that looks tasty to various mammals and birds, but one that can suddenly take to the air when attacked. At other times, among reeds, the crake will drop like a stone into the water, creating a big splash, and then move loudly through the reeds, much like a frog staying at the surface. What is noteworthy is that the crake calls attention to itself while acting as if it is not. It must not be such a good rat or frog that it remains undetected, yet it must act like a target trying to avoid detection. Thus, movements are outwardly furtive but louder than usual.


We use patterns to detect deception, and randomness is the absence of pattern. It is often not appreciated how valuable randomness is as part of a deceptive strategy designed to avoid detection. Consider a couple of examples. Fake butterfly eggs are actually plant structures evolved to prevent butterflies from laying their eggs—since butterflies avoid laying eggs where they see some have already been laid. The fake eggs appear at random on the surface of the plant’s leaves. Yet in closely related species, where the plant structures serve their original function, they are symmetrically located on each side of the leaf. Thus, natural selection created the randomness, presumably since butterflies had evolved to treat symmetrical patterns of eggs as if they were not really eggs (as indeed they are not). An ongoing struggle for randomness occurs in a pronghorn antelope. The pronghorn mother leaving her offspring hidden between nursings while she eats initially orients herself away from her offspring, then for much of the time she faces in random directions. Finally, only just before returning to nurse does the mother reveal the offspring’s position by facing it.

Now consider a human example. In the old days, when customs officers routinely searched most bags in the owner’s presence, a tried-and-true method to detect smuggling was to poke around randomly while watching the owner out of the corner of their eye. Whenever the owner became agitated or showed undue attention, the customs officer eliminated the rest of the bag and concentrated on the suspicious section. Again, by poking around (and paying close attention), the officer allowed the owner to guide him or her to the problem, presumably something illegal. Note that lack of preparation for this eventuality—being caught—only heightens one’s anxiety and inadvertent information leakage.

For years, I have been well aware of the importance of information limitation. I have not used it with customs officials, but if a police officer is searching the trunk of my car, I simply turn my back. The officer may think I have something to hide, but he or she will learn nothing from me about where it is, if indeed there is something. Of course, when being watched for other purposes, we may also busy ourselves with semi-random behavior to hide the truth.

Once, when trying to get readmitted to Harvard after a medical leave, I had to take the famous “What do you see in this inkblot?” (Rorschach) test. I had learned that results were graded based on whether you saw a picture or told a story, whether it was in color, whether the story was coherent, and so on, but I had forgotten what the “appropriate” answers were supposed to look like to signify “normal,” so I simply randomized my responses, figuring absence of a pattern was my best hope. Sometimes they got a story, sometimes a snapshot, sometimes in color, and so on. At least I did not appear to be rigid or compulsive. I was readmitted.

It may, indeed, be that a certain degree of randomness is built into the very core of our behavior. Not only will others not detect a pattern, but neither shall we—thus preventing us from inadvertently revealing ourselves.


How do animals react when they detect deception directed at them? Studies from a range of species—wasps, birds, and monkeys—suggest they often get angry and seek immediate retribution. At least this seems to be true of several species in which individuals have what appear to be arbitrary symbols that confer status—so-called badges—such as greater melanin (darker color) on the chest feathers of sparrows or the mouth-parts (clypeus) of wasps. In each case, the signals are on the part of the body most visible in a face-to-face encounter, and each is positively associated with body size and dominance. How is the association maintained between the arbitrary badge of status and the status itself? In wasps, for example, less than 1 percent of the body’s melanin is found on the clypeus. Why do cheaters not invade the system and produce higher-status badges than their size warrants? Precisely because they are immediately attacked and are usually unable to defend themselves. Those whose clypeuses are painted to look more dominant do not become more dominant but are attacked six times as often by truly dominant individuals, while wasps painted to look less dominant are attacked twice as often as nonpainted controls. And it is interesting that subordinate wasps attack those painted to look dominant more often than they attack those who look dominant to begin with. A key perceptual factor is incongruity between appearance and behavior—when individuals are painted darker and made more aggressive via hormone treatment, they gain in dominance, but when made more aggressive without the change in appearance, the wasps fail to establish stable dominance relations, presumably because others are continually tempted to challenge them.

When a sparrow’s chest is painted blacker to enhance the apparent badge size and, thus, status of the sparrow, the effect on status is usually the opposite. The altered bird is attacked more frequently than before, especially by those with the same apparent badge size or larger. The result is a drop in status—or ostracism from the group—for the individual with the deceptive badge. By contrast, those who, in effect, deceive downward—that is, who are bleached to appear less dark than they really are—often become hyperaggressive, whirling around and attacking their near neighbors who now act disrespectfully by standing too close to them based on their new (diminished) badge.

That deception might induce anger and attack was suggested to me very forcefully in my own life some thirty years ago. I was taking a walk, carrying my one-year-old son in my arms, when I spotted a squirrel in a tree. The problem was that my son did not see the squirrel, so I whistled as melodically as I could to draw the squirrel closer to us and, sure enough, the squirrel crept forward, but my son still could not see it. So I decided to reverse my relationship with the squirrel and mimic an attack. I suddenly lunged at it. I expected it to scamper away from me. I would have ruined a budding friendship but allowed my son to see the squirrel as it rushed away from us. Instead, the squirrel ran straight at us, chitter-ing in apparent rage, teeth fully exposed, jumping to the branch closest to me and my son. Now my son saw the squirrel, and I had the fright of my life, quickly running several steps away.

For my folly, the squirrel could have killed my son with a leap to my shoulders and two expert bites to his neck. Had I begun the relationship hostile, I believe the squirrel never would have become so angry. It was the betrayal implied by beginning friendly, only then to attack (deception), that triggered the enormous anger. There is nothing quite like the humility you feel as you sneak your son back into your home, not telling your wife, of course, that in a little pseudo-scientific work on the side, you had managed to enrage a squirrel to the point of putting her child at risk. I had no plans to try that stunt again anytime soon.

The importance of aggression following knowledge of deception is that it may greatly increase the costs of deceptive behavior and the benefits of remaining undetected. Fear of aggression can itself become a secondary signal suggesting deception, and its suppression an advantage for self-deception. Of course, aggression is not the only social cost of detected deception. A woman may terminate a relationship upon learning of a lie, usually a crueler punishment than her giving you a good beating, assuming she is capable. Detected deception may lead to social shame—bad reputation, loss of credibility and status, so that there will always be pressure on the deceiver to hide the deception, not only to make it successful but to avoid the larger consequences of detection.


Naturally one must be careful in imputing particular kinds of consciousness to other species, but some situations strongly suggest that animals are conscious of ongoing deception in some detail. Ravens, for example, have evolved a set of elaborate behaviors surrounding their tendency to cache (that is, bury and hide) food for future consumption, which can be enjoyed by another bird who happens to view the caching. Accordingly, ravens who are about to hide food seem very sensitive to just this possibility. They distance themselves from others and often cache behind a structure that obstructs others’view. They regularly interrupt caching to look around. At any evidence they are being observed, they will usually retrieve the cached food and wait to rebury somewhere else, preferably while not under observation. If they do cache food, they will often return within a minute or two. The watchers, in turn, stay at a safe distance, often hiding behind a tree or other object. They stop looking if the other stops caching and wait a minute or more after the bird has left before going for the cache. Hand-reared ravens, in turn, can follow the human gaze by repositioning themselves to see around an obstacle. This suggests the possibility that ravens can project the sight of another individual into the distance. Likewise, when jays are caching in others’ presence, they maximize their distance from others and cache in the shade and in a confusing pattern, moving caches frequently. Experimental work shows that they remember who has watched them cache in the past and when being observed by such individuals are more likely to re-cache than when they are being watched by a newcomer—another example of intelligence evolving in the context of deception.

In the presence of other squirrels, gray squirrels cache farther apart, build false caches, and build with their backs turned to the other squirrels; no such responses are shown to crows who may be watching. Turning one’s back often shows up in other mammals, as well. A chimpanzee male displaying an erection to a female may turn his back when a more dominant male arrives, until his erection has subsided. Children as young as sixteen months will turn their backs to conceal the object in hand or what they are doing. I personally find it very hard in the presence of a woman with whom I am close to receive a phone call from another woman with whom I may have, or only wish to have, a relationship, without turning my back to pursue the conversation. This occurs even though there is nothing visual to hide and the act of turning gives me away. Perhaps this is a case of reducing cognitive dissonance—and cognitive load—by not having to watch one woman watch you while you pretend not to talk to another woman.

In ravens, the pilferers avoid searching for known caches when in the presence of those who cache but will go immediately to the caches in the presence of a noncaching bird (that is unlikely to defend). In addition, they actively search away from the cache in the presence of the cacher, as if hiding their intentions. In one experiment, when ravens were introduced into an area where food was hidden, a subordinate male quickly developed the ability to find food, which the most dominant quickly learned to parasitize. This in turn led the subordinate to first search in areas where no food was present, to lure the dominant away, at which point the subordinate moved quickly to the food itself.

Mantis shrimps are hard-shelled and their claws dangerous for seven weeks out of eight. On the eighth week, they are molting, and their body and claws are soft; they are unable to attack others and are vulnerable to attack by them. When encountered at this time, they greatly increase their rate of claw threats, sometimes combined with insincere lunges at the opponent. About half the time, this scares off their opponent. The other half, the soft-shelled shrimp runs for its life. The week before a mantis shrimp becomes soft-shelled, it increases its rate of claw threats but also increases the rate at which these threats are followed by actual attack, as if signaling that threats will be backed up by aggressive action just before the time when they will not.

In fiddler crabs, the male typically has a large claw used to fight and threaten other males and to court females. Should he lose this claw, he regenerates one very similar in appearance but less effective than the original. The size of the first claw does indeed correlate (independent of body size) with claw strength as well as ability to resist being pulled from one’s burrow, but the size of the replacement claw does not, and males can’t distinguish between the two kinds of claws in an opponent.

In primates, hiding information from others may take very active forms. For example, in both chimpanzees and gorillas, individuals will cover their faces in an apparent attempt to hide a facial expression. Gorillas in zoos have been seen to cover “play faces” (facial expressions meant to invite play) with one or both hands, and these covered faces are less likely to elicit play than uncovered play faces. Of course, a play face hidden in this fashion is hardly undetectable and may easily become a secondary signal. Chimpanzees will hide objects behind their backs that they are about to throw. They will also throw an object to one side of a tree to frighten another chimp into moving to the opposite side, where his opponent awaits him.


An important part of understanding deception is to understand it mathematically as an evolutionary game, with multiple players pursuing multiple strategies with various degrees of conscious and unconscious deception (in a fine-grained mixture). Contrast this with the problem of cooperation. Cooperation has been well modeled as a simple prisoner’s dilemma. Cooperation by both parties benefits each, while defections hurt both, but each is better off if he defects while the other cooperates. Cheating is favored in single encounters, but cooperation may emerge much of the time, if players are permitted to respond to their partner’s previous moves. This theoretical space is well explored.

The simplest application of game theory to deception would be to treat it as a classical prisoner’s dilemma. Two individuals can tell each other the truth (both cooperate), lie (both defect), or one of each. But this cannot work. One problem is that a critical new variable becomes important—who believes whom? If you lie and I believe you, I suffer. If you lie and I disbelieve you, you are likely to suffer. By contrast, in the prisoner’s dilemma, each individual knows after each reciprocal play how the other played (cooperate or defect), and a simple reciprocal rule can operate under the humblest of conditions—cooperate initially, then do what your partner did on the previous move (tit for tat). But with deception, there is no obvious reciprocal logic. If you lie to me, this does not mean my best strategy is to lie back to you—it usually means that my best strategy is to distance myself from you or punish you.

The most creative suggestion I have heard to mathematically model deception is to adapt the ultimatum game (UG) to this problem. In the UG, a person proposes a split of, say, $100 (provided by the experimenter) —$80 to self, $20 to the responder. The responder, in turn, can accept the split, in which case the money is split accordingly, or the responder can reject the offer, in which case neither party gets any money. Often the game is played as a one-shot anonymous encounter. That is, individuals play only once with people they do not know and with whom they will not interact in the future. In this situation, the game measures an individual’s sense of injustice—at what level of offer are you sufficiently offended to turn it down even though you thereby lose money? In many cultures, the 80/20 split is the break-even point at which one-half of the population turns down the offer as too unfair.

Now imagine a modified UG in which there are two possible pots (say, $100 and $400) and both players know this. One pot is then randomly assigned to the proposer. Imagine the proposer offers you $40, which could represent 40 percent of a $100 pot (in which case you should accept) or 10 percent of a $400 pot (most people would reject). The proposer is permitted to lie and tell you that the pot is the smaller of the two when in fact it is the larger. You can trust the proposer or not, but the key is that you are permitted to pay to find out the truth from a (disinterested) third party. This measures the value you place in reducing your uncertainty regarding the proposer’s honesty.

If you then discover that the proposer lied, you should have a moral (or, at least, moralistic) motive to reject the offer, and the other way around for the truth—all compared to uncertainty, or not paying to find out. Note that from a purely economic point of view, there is no benefit in finding out the truth, since it costs money and may lead to an (otherwise) unnecessary loss of whatever is offered. The question can then be posed: How much would a responder be prepared to pay to reduce the uncertainty and go for a possibly inconvenient truth? Note that the game can be played in real life with varying degrees of anonymity and also multiple times, as in the iterated prisoner’s dilemma. As ability to discriminate develops, the other person will benefit more from your honesty (quickly seen as such) and suffer less from deception (spotted and discarded).

When we add self-deception, the game quickly becomes very complicated. One can imagine actors who are:

• Stone-cold honest (cost: information given away, naive regarding deception by others).

• Consciously dishonest to a high degree but with low self-deception (cost: higher cognitive cost and higher cost when detected).

• Dishonest with high self-deception (more superficially convincing at lower immediate cognitive cost but suffering later defects and acting more often in the service of others).

And so on.


Those talented at the mathematics of simple games or studying them via computer simulation might find it rewarding to define a set of people along the lines just mentioned, and then assign variable quantitative effects to explore their combined evolutionary trajectory. Perhaps results will be trivial and trajectories will depend completely on the relative quantitative effects assigned to each strategy, but it is much more likely that deeper connections will emerge, seen only when the coevolutionary struggle is formulated explicitly. The general point is, of course, that there are multiple actors in this game, kept in some kind of frequency-dependent equilibrium that itself may change over time. We choose to play different roles in different situations, presumably according to the expected payoffs. Of course it is better to begin with very simple games and only add complexity as we learn more about the underlying dynamics.

It stands to reason that if our theory of self-deception rests on a theory of deception, advances in the latter will be especially valuable. I have known this for thirty years but have not been able to think of anything myself that is original regarding the deeper logic of deception, nor have I seen much progress elsewhere. Yes, signals in male/female courtship interactions may evolve toward costlier ones that are more difficult to fake (for example, antler size, physical strength, and bodily symmetry), but there is always room for deception, and many systems do not obey this simple rule regarding cost.