Part 2 (2/2)

ENTER THE KLUGE.

The answer lies in Arnold Toynbee's brain. It was one of the finest of its kind, which is saying something because any human brain is a truly marvelous thing. But no brain is perfect, not even the brain of a genius like Toynbee.

The brain was not designed by a team of engineers. It was not betatested, reworked, and released with a big ad campaign; it evolved. When the ancestors of today's humans parted ways with the ancestors of today's chimpanzees some five to seven million years ago, the protohuman brain was much smaller than the modern brain's fourteen hundred cubic centimeters. Around 2.5 million years ago, and again 500,000 years ago, our ancestors' brains went through growth spurts. The final ballooning occurred some 150,000 to 200,000 years ago. Throughout all this vast stretch of time-and in the much longer years when the brains of our ancestors' ancestors were forming-evolutionary pressures shaped the brain's development.

Genes normally replicate and make exact copies of themselves, but occasionally they misfire and produce mutations. If a mutation makes the person who has it significantly less likely to survive and reproduce, it will die off along with the unlucky person who got it. But mutations that a.s.sist survival and reproduction spread. They may even, eventually, become universal features of the species. This is true of mutations involving muscles, bones, and organs. And it's true of mutations involving the brain.

So positive changes proliferate. Mistakes are removed. And we get smarter and smarter. It's all so simple, neat, and efficient.

And yet the human brain is anything but ”simple, neat, and efficient.” Borrowing a term from engineering, psychologist Gary Marcus has dubbed the brain a ”kluge”-an inelegant but effective solution to a problem. When the carbon dioxide filters on board the Apollo 13 capsule failed, engineers at mission control dreamed up a replacement made out of ”a plastic bag, a cardboard box, some duct tape, and a sock.” That's a kluge.

”Natural selection, the key mechanism of evolution, is only as good as the random mutations that arise,” Marcus writes. ”If a given mutation is beneficial, it may propagate but the most beneficial mutations imaginable, alas, never appear.” As a result, an evolutionary solution to an environmental problem that is flawed or suboptimal but nonetheless does the job-a kluge, in other words-may spread and become standard operating equipment for the species. Once in place, the new equipment may be used to deal with other problems if, once again, it does the job adequately. And when new challenges arise, it may be the platform on which new less-than-perfect solutions will be built-thus multiplying the quirks and oddities. This is how we got spines that allow us to walk upright but are so flawed they routinely leave us bent over with back pain; vision marred by a built-in blind spot caused by the absurd design of our retina; and wisdom teeth that emerge to inflict pain for no particular reason. And then there is the brain. Imagine its ma.s.s, complexity, and general kluginess growing as our ancestors encountered one problem after another, across unfathomable spans of time, and it becomes obvious why the brain is anything but simple, neat, and efficient.

It's also critical to remember that natural selection operates in response to pressures in a particular environment. Change the environment and a solution may no longer be so helpful. Consider pale skin. It was a useful adaptation for human populations living at high lat.i.tudes, where sunlight is weaker, because it allowed the body to maximize production of vitamin D. But that advantage is strictly limited to the environment in which pale skin evolved. Not only is pale skin unnecessary at lower lat.i.tudes, where the sun's rays are stronger, it puts people at greater risk of skin cancer. The fact that evolutionary adaptations are specific to the environments in which they evolved didn't matter much throughout most of human history, for the simple reason that the environments in which we lived changed slowly. But as a result of the explosion of technology and productivity of the last several centuries, most people live in human-constructed environments that are dramatically different from the natural environments in which their ancestors lived-producing such novel sights as pasty-faced Englishmen clambering aboard airplanes for tropical destinations where the lucky vacationers will lie in the sun, sip fruity drinks, and boost their risk of skin cancer. From the perspective of one person, several centuries is a very long time, but in biological terms, it is a blink of a chimpanzee's eye. Human evolution doesn't move at anything like that speed and thus we are left with one of the defining facts of modern life: We live in the Information Age but our brains are Stone Age.

These two facts-the brain's kluginess and the radically changed environment in which we live-have a vast array of consequences. And almost all of us, almost always, are blissfully unaware of them.

Consider an experiment in which psychologists dropped 240 wallets on various Edinburgh streets. In each wallet, there was a personal photo, some ID, an old raffle ticket, a members.h.i.+p card or two, and a few other minor personal items. There was no cash. The only variation in the wallets was the photograph, which could be seen through a clear plastic window. In some, it showed a smiling baby. In others, there was a puppy, a family, or an elderly couple. A few of the wallets had no photo at all. The researchers wanted to know how many of the wallets would be dropped in mailboxes, taken to the police, or otherwise returned. More specifically, the researchers wanted to know if the content of the photograph in each wallet would make a difference. It shouldn't, of course. A lost wallet is important to whoever loses it and returning it is a bother no matter what's in it. In strictly rational terms, the nature of the photograph is irrelevant.

And yet psychologist Richard Wiseman discovered that the photograph made an enormous difference. Only 15 percent of those without one were returned. A little more than one-quarter of the wallets with a picture of an elderly couple were returned, while 48 percent of the wallets with a picture of a family, and 53 percent with the photo of a puppy, were returned. But the baby walloped them all; an amazing 88 percent of wallets with pictures of infants were returned.

This doesn't make sense-until we consider the ”two-system” model of decision making. Researchers have demonstrated that we have not one mind making decisions such as ”Should I bother sending this wallet back to its owner?” We have two. One is the conscious mind, and since that mind is, by definition, aware only of itself, we think of it as being the single, unified, complete ent.i.ty that is ”me.” Wrong. Quite wrong, in fact. Most of what the brain does happens without our having any conscious awareness of it, which means this ”unconscious mind” is far more influential in our decision making than we realize.

The two minds work very differently. Whereas the conscious mind can slowly and carefully reason its way to a conclusion-”On the one hand, returning the wallet is the nice thing to do, but when I weigh that against the time and bother of returning it . . .”-the unconscious mind delivers instantaneous conclusions in the form of feeling, hunches, and intuitions. The difference in speed is critical to how the two systems work together. ”One of psychology's fundamental insights,” wrote psychologist Daniel Gilbert, ”is that judgments are generally the products of nonconscious systems that operate quickly, on the basis of scant evidence, and in a routine manner, and then pa.s.s their hurried approximations to consciousness, which slowly and deliberately adjusts them.” The unconscious mind is fast so it delivers first; the conscious mind then lumbers up and has a look at the unconscious mind's conclusion.

When someone spots a wallet on the streets of Edinburgh, picks it up, and decides to return it, she thinks that's all there is to the story. But far more happened. Even before her conscious thoughts got rolling, unconscious systems in her brain took a look at the situation and fired off a conclusion. That conclusion was the starting point for her conscious thoughts.

One unconscious mental system equates a photographed object with the real thing. That sounds mad-only a deranged person thinks a photo of a puppy is a puppy-until you recall that in the environment in which our brains evolved, there were no images of things that were not what they appeared to be. If something looked like a puppy, it was a puppy. Appearance equals reality. In the ancient environment in which our brains evolved, that was a good rule, which is why it became hardwired into the brain and remains there to this day. Of course, the reader will object that people do not routinely confuse photos of puppies with puppies. This is true, fortunately, but that's only because other brain systems intervene and correct this mistake. And a correction is not an erasure. Thus there remains a part of our brain that is convinced an image of something actually is that something. This quirk continues to have at least a little influence on our behavior, as jilted girlfriends reveal every time they tear up photos of the cad who hurt them or parents refuse to throw duplicate photos of their children in the trash. Still doubt this? Think of a lovely, tasty piece of fudge. But now imagine this fudge is shaped like a coil of dog poo. Still want to eat it? Right. That's exactly the reaction psychologists Paul Rozin and Carol Nemeroff got when they asked people to eat fudge shaped like dog poo, among other experiments involving a gap between image and reality. ”In these studies, subjects realized that their negative feelings were unfounded but they felt and acknowledged them anyway.”

We really like babies. The sight of a chubby little infant gurgling and grinning is enough to make even Scrooge smile. Babies are the best. And for good reason. In evolutionary terms, nothing is more important than reproducing. Among our ancestors, parents who didn't particularly care if their babies were well-fed, healthy, happy, and safe were much less likely to see those babies become adults with children of their own. So that att.i.tude was going nowhere. But those who felt a surge of pleasure, compa.s.sion, and concern at the very sight of their darling little ones would take better care of them and be more likely to bounce grandchildren on their knees. Thus the automatic emotional response every normal person feels at the sight of a baby became hardwired, not only among humans but in every species that raises its young to maturity: Never come between a bear and her cubs.

But there's a problem here. Evolution is ruthless. It puts a priority on your reproduction, which means it cares about your offspring, not somebody else's. And yet that's not how we respond to babies. The sight of any gurgling and grinning infant makes us feel all warm and compa.s.sionate. Why is that? Our compa.s.sionate response to babies is a kluge. In prehistoric environments, we would seldom encounter a baby that was not our own, or that of our kin or our neighbor. Thus an automatic surge of compa.s.sion in response to the sight of any baby was not a perfect response-in ruthless reproduce-above-all terms-but it didn't cause us to do anything too foolish. So it did the job; it was good enough.

Now let's go back to those wallets on the streets of Edinburgh. Someone comes along, spots one, picks it up-and finds there is no picture. What does she do? Well, for her, there's not much more than rational calculation to go on. She knows the owner would probably like the wallet back, but there's no money in it so the loss wouldn't be too bad. And besides, returning it would be a ha.s.sle. Hence, only 15 percent of these wallets are returned. But another person picks up a different wallet, looks in it, and sees a photo of an elderly couple. This humanizes the problem-literally so, for the brain system that mistakes a photo of an elderly couple for the elderly couple themselves. An intuitive impulse is elicited, a feeling, a sense of compa.s.sion. Then the conscious brain steps in and thinks about the situation. Result: 25 percent of these wallets are returned, a significant increase.

The photo of a family generates a stronger unconscious response and a 48 percent return rate. And the baby, of course, produces an amazing 85 percent return rate. But what about the photo of the puppy? At 53 percent, its return rate is roughly equal to the photo of a family and double that of the elderly couple. And it's not even human! One might think that makes no sense from an evolutionary perspective, and yet, it does. The puppy and the baby may be different species but both have big eyes, a little mouth and chin, and soft features. Our automatic response to babies is triggered by these features, so anything or anyone that presents them can elicit a similar response. It's not a coincidence that all the animals and cartoon characters we find cute and adorable-from baby seals to Mickey Mouse-have the same features. Nor is it a coincidence that, as psychologists have shown, people often stereotype ”baby-faced” adults as innocent, helpless, and needy. It's our kluge at work.

SEEING THINGS.

It would be nice if feeling compa.s.sion for puppies and baby seals were the worst thing that happens when cognitive wires get crossed. Unfortunately, it's not. Thanks to the brain's evolutionary character, we often make mistakes about far more consequential matters.

In the last years of the Second World War, Germany pounded London with V-1 and V-2 rockets. These ”flying bombs” were a horrible new weapon, unlike anything seen before. At first, Londoners didn't know what to make of the threat. All they knew was that, at any moment, with little or no warning, a ma.s.sive explosion would erupt somewhere in the city. But gradually people began to realize that the rocket strikes were cl.u.s.tered in certain parts of the city, while others were spared. Rumors spread. n.a.z.i spies were directing the missiles, some said. The spies must live in the parts of the city that weren't being hit. But what was the German strategy? The East End was being particularly hard hit and the East End is working cla.s.s. Aha! The Germans must be trying to inflame cla.s.s resentment in order to weaken the war effort.

It was a compelling explanation and yet it was wrong. As terrifying as the rockets were, they lacked precision guidance equipment and the best the Germans could do was point them at London and let them explode where they might. In 1946, statistician R. D. Clarke made a simple one-page calculation that compared the extent of cl.u.s.tering in the flying-bomb attacks to the cl.u.s.tering that could be expected if the bombs had been randomly distributed. There was a near-perfect match.

We have a hard time with randomness. If we try, we can understand it intellectually, but as countless experiments have shown, we don't get it intuitively. This is why someone who plunks one coin after another into a slot machine without winning will have a strong and growing sense-the ”gambler's fallacy”-that a jackpot is ”due,” even though every result is random and therefore unconnected to what came before. Similarly, someone asked to put dots on a piece of paper in a way that mimics randomness will distribute them fairly evenly across the page, so there won't be any cl.u.s.ters of dots or large empty patches-an outcome that is actually very unlikely to happen in a true random distribution. And people believe that a sequence of random coin tosses that goes THTHHT is far more likely than the sequence THTHTH, even though they are equally likely.

Many people experienced this intuitive failure listening to an iPod. When it's set on ”shuffle,” it's supposed to choose and play songs randomly. But it often doesn't seem random. You may have heard six in a row from one artist and wondered if the program is biased in favor of that guy. Or maybe it gave top billing to your favorites and you suspected it's actually mimicking your nonrandom choices. Or perhaps-as some conspiracy-minded bloggers insisted-it seemed to favor songs from record companies that have a close relations.h.i.+p with Apple, the maker of the iPod. Peppered with complaints and accusations, Apple subsequently reprogrammed the shuffle feature: The idea was to make it ”less random to make it feel more random,” Steve Jobs said.

People are particularly disinclined to see randomness as the explanation for an outcome when their own actions are involved. Gamblers rolling dice tend to concentrate and throw harder for higher numbers, softer for lower. Psychologists call this the ”illusion of control.” We may know intellectually that the outcome is random and there's nothing we can do to control it, but that's seldom how we feel and behave. Psychologist Ellen Langer revealed the pervasive effect of the illusion of control in a stunning series of experiments. In one, people were asked to cut cards with another person to see who would draw the higher card and to make bets on the outcome. The outcome is obviously random. But the compet.i.tor people faced was in on the experiment, and his demeanor was carefully manipulated. Sometimes he was confident and calm; sometimes he was nervous. Those who faced a nervous compet.i.tor placed bigger bets than those who squared off against a confident opponent. Langer got the same result in five other experiments testing the ”illusion of control”-including an experiment in which people put a higher value on a lottery ticket they chose at random than a ticket they were given, and another in which people rated their chances of winning a random-outcome game to be higher if they were given a chance to practice than if they had not played the game before.

Disturbing as these findings were, it was another of Langer's experiments that fully revealed how deluded the brain can be. Yale students were asked to watch someone flip a coin thirty times. Before each flip, the students were asked to predict whether the flip would come up heads or tails; after each, they were told whether they had ”won” or ”lost.” In reality, the results were rigged so there would always be fifteen wins and fifteen losses, but some of the students would get a string of wins near the beginning while others first encountered a string of losses. At the end of the thirty flips, students were asked how many wins they thought they got in total, how good they thought they were at predicting coin tosses, and how many wins they thought they would get if they did the test again with a hundred flips in total. Langer discovered a clear tendency: Students who got a string of wins at the beginning thought they did better than those who didn't; they said their ability to predict was higher; and they said they would score significantly more wins in a future round of coin flipping. So the string of early wins had triggered the illusion of control. Students then focused on subsequent wins and paid little attention to losses, which led them to the false conclusion that they had notched more wins than losses and that they could do it again.

Langer's results are particularly startling when we consider the full context of the experiment. These are top-tier students at one of the world's best universities. They're in a clinical environment in which they believe their intelligence is being tested in some way. Under the circ.u.mstances, Langer noted, they ”are likely to be 'superrational.'” And this is hardly a tricky task. A flipped coin is the very symbol of randomness and any educated person knows it is absurd to think skill has anything to do with calling ”heads” or ”tails.” And yet Langer's test subjects still managed to fool themselves.

Langer's research inspired dozens more studies like it. Psychologists Paul Presson and Victor Bena.s.si of the University of New Hamps.h.i.+re brought it all together and noticed that although psychologists use the term illusion of control, much of the research wasn't about ”controlling” an outcome. As in Langer's coin-flipping experiment, it was really about predicting outcomes. They also found the illusion is stronger when it involves prediction. In a sense, the ”illusion of control” should be renamed the ”illusion of prediction.”

This illusion is a key reason experts routinely make the mistake of seeing random hits as proof of predictive ability. Money manager and Forbes columnist Kenneth Fisher recalled attending a conference at which audience members were invited to predict how the Dow would do the next day. At the time, the index hovered around 800, so Fisher guessed it would drop 5.39 points. ”Then I noticed the gent next to me jotting down a 35-point plunge,” Fisher recalled in 1987. ”He said he hadn't the foggiest idea what might happen,” but he certainly had a strategy. ”If you win,” the man told Fisher, ”the crowd will think you were lucky to beat everyone else who bets on minor moves. But if my extreme call wins, they'll be dazzled.” The next day the Dow dropped 29 points. ”That afternoon, folks bombarded the winner for details on how he had foreseen the crash. He obliged them all, embellis.h.i.+ng his 'a.n.a.lysis' more with each telling. That night, when I saw him alone, he had convinced himself that he had known all along, and became indignant when I reminded him that his call was based on showmans.h.i.+p.”

Blame evolution. In the Stone Age environment in which our brains evolved, there were no casinos, no lotteries, and no iPods. A caveman with a good intuitive grasp of randomness couldn't have used it to get rich and marry the best-looking woman in the tribe. It wouldn't have made him healthier or longer-lived, and it wouldn't have increased his chances of having children. In evolutionary terms, it would be a dud. And so an intuitive sense of randomness didn't become hardwired into the human brain and randomness continues to elude us to this day.

The ability to spot patterns and causal connections is something else entirely. Recognizing that the moon waxes and wanes at regular intervals improved the measurement of time, which was quite handy when someone figured out that a certain patch of berries is ripe at a particular period every summer. It was also good to know that gazelles come to the watering hole when the rains stop, that people who wander in the long gra.s.s tend to be eaten by lions, and a thousand other useful regularities. Pattern recognition was literally a matter of life and death, so natural selection got involved and it became a hardwired feature of the human brain.

And not only the human brain. Birds and animals also benefit from spotting patterns, and thus their cognitive wiring makes them adept at seeing connections. Sometimes they are too good at it. When B. F. Skinner put pigeons in his famous ”Skinner box” and gave them food at randomly selected moments, the pigeons quickly connected the appearance of the food to whatever they happened to be doing when it appeared. A pigeon that happened to be thrusting its head into a corner, for example, ate the food, then went back to the lucky corner and resumed thrusting its head, over and over, expecting more food to drop. It would be nice to blame this behavior on the limited intelligence of pigeons, but they are far from the only species that draws false connections between unrelated events. Humans do it all the time. Skinner believed it was a root cause of superst.i.tion. ”The birds behaved as if they thought that their habitual movement had a causal influence on the reward mechanism, which it didn't,” wrote biologist Richard Dawkins. ”It was the pigeon equivalent of a rain dance.”

Rain dances are ineffective, but they aren't harmful. Someone who does a dance, gets rained on, and concludes that dancing causes rain has made a serious mistake, but he won't increase his chances of an early death if he dances when he wants rain. That's typical of false positives. Seeing patterns that aren't there isn't likely to make a big difference to a person's chances of surviving and reproducing-unlike failing to see patterns that do exist. This profound imbalance is embedded in our cognitive wiring. We consistently overlook randomness but we see patterns everywhere, whether they are there or not. The stars may be scattered randomly across the night sky, but people see bears, swans, warriors, and the countless other patterns we call constellations. We see faces in clouds, rocks, and the moon. We see ca.n.a.ls on Mars and the Virgin Mary on burnt toast. Of course, we also see a great many patterns that really are there, often with astonis.h.i.+ng speed and accuracy. But the cost of this ability is a tendency to see things that don't exist.

Although humans may share this tendency with other animals, at least to some extent, there is something quite different about the human quest to spot patterns. In a cla.s.sic experiment that has been conducted with many variations, people sit before a red light and a green light. The researchers ask them to guess which of the two lights will come on next. At first, there isn't much to go on. There seems to be no pattern. And indeed, there isn't a pattern. The flas.h.i.+ng of the lights is random, although the test subjects aren't told this. But as lights continue to flash and time pa.s.ses, it becomes apparent that there is one regularity: The red light is coming on much more often than the green. In fact, the distribution is 80 percent red, 20 percent green. Faced with this situation, people will tilt their guesses to match the frequency with which the lights are coming on-so they'll guess red about 80 percent of the time and green 20 percent. In effect, they are trying to match the ”pattern” of the flashes. But that's impossible because it's random. Needless to say, people don't do very well on this test.

But pigeons do. So do rats and other animals. Put to the same test, they follow a different strategy. Since there are more red flashes than green, they simply choose red over and over. That yields far better results. You might say it's the rational thing to do. But we rational humans don't do it.

Why not? That's the question University of California neuroscientist Michael Gazzaniga explored in a fascinating experiment. For decades, Gazzaniga has worked with people who have had the connection between the right and left hemispheres of their brain severed, usually as a form of treatment for severe epilepsy. These ”split-brain patients” function surprisingly well under most circ.u.mstances. But because the hemispheres control different aspects of perception, thought, and action, severing the two does produce some startling results. Most important for researchers is the fact that it is possible to communicate with one hemisphere of the brain-by revealing information to one eyeball but not the other, for example-while keeping the other in the dark.

When Gazzaniga and his colleagues put the left hemispheres of split-brain patients to the red-green test, he got the usual results: They tried to figure out the pattern and ended up doing poorly. But right hemispheres did something startling: Like rats and other animals, they guessed red over and over again and thus got much better results. For Gazzaniga, this was important proof of an idea he has pursued for many years. In the left hemisphere of the brain-and only the left hemisphere-is a neural network he calls ”the Interpreter.” The Interpreter makes sense of things. After the brain experiences perceptions, emotions, and all the other processes that operate at lightning speed, the Interpreter comes along and explains everything. ”The left hemisphere's capacity of continual interpretation means it is always looking for order and reason,” Gazzaniga wrote, ”even when they don't exist.”

The Interpreter is ingenious. And relentless. It never wants to give up and say, ”This doesn't make sense,” or ”I don't know.” There is always an explanation. In one experiment, Gazzaniga showed an image to the left hemisphere of a split-brain patient and another to the right hemisphere. An array of photos was spread out on a table. The patient was asked to pick the photo that was connected to the image they had seen. In one trial, the left hemisphere of the patient was shown an image of a chicken claw; on the table was a photo of a chicken. The right hemisphere was shown a snow scene; on the table was a photo of a snow shovel. When the patient's left hand-which is controlled by the right hemisphere-pointed to the shovel, Gazzaniga asked the left hemisphere why. It had no idea, of course. But it didn't say that. ”Oh, that's simple,” the patient answered confidently. ”The chicken claw goes with the chicken and you need a shovel to clean out the chicken shed.”

For humans, inventing stories that make the world sensible and orderly is as natural as breathing. That capacity serves us well, for the most part, but when we are faced with something that isn't sensible and orderly, it's a problem. The spurious stories that result can seriously lead us astray, and, unfortunately, more information may not help us. In fact, more information makes more explanations possible, so having lots of data available can actually empower our tendency to see things that aren't there. Add a computer and things only get worse. ”Data mining” is now a big problem for precisely this reason: Statisticians know that with plenty of numbers and a powerful computer, statistical correlations can always be found. These correlations will often be meaningless, but if the human capacity for inventing explanatory stories is not restrained by constant critical scrutiny, they won't appear meaningless. They will look like hard evidence of a compelling hypothesis-just as the apparent cl.u.s.tering of rocket strikes in London looked like evidence that the n.a.z.is were targeting certain neighborhoods in order to advance their cunning strategy.

We can all fall victim to this trap, but it's particularly dangerous for experts. By definition, experts know far more about their field of expertise than nonexperts. They have read all the books, and they have ma.s.ses of facts at their fingertips. This knowledge can be the basis of real insight, but it also allows experts to see order that isn't there and to explain it with stories that are compelling, insightful, and false. In a PBS interview, Jeff Greenfield, an American journalist, recalled how he and other pundits were tripped up during the presidential election of 1988. Vice President George H.W. Bush wouldn't win, they believed. The reason was an obscure fact known only to political experts: ”No sitting vice-president has been elected since Martin Van Buren.” Aha! A meaningful pattern! Or so it seemed. But as it turned out, Bush didn't lose and the pundits would have been better off if they had never heard of Martin Van Buren.

This is the quicksand that consumed Arnold Toynbee. His lifelong project began with an intuition-a ”flash of perception,” he called it-that the trajectory of Western history was following that of ancient Greece and Rome. After spotting that pattern, Toynbee elaborated on it and committed it to paper in 1921. When, in the course of writing A Study of History, Toynbee was confronted with information that didn't fit his tidy scheme-such as the sudden appearance of the Islamic ”universal state”-he was in the position of the hapless split-brain patient whose hand was pointed at a photo of a shovel for some reason. It didn't make sense; it didn't fit the pattern. So Toynbee's left hemisphere got busy. Drawing on his intelligence and his vast store of knowledge, Toynbee created ingenious stories that explained the seemingly inexplicable and maintained order in his mental universe.

Arnold Toynbee wasn't deluded despite his brilliance. He was deluded because of it.

ALWAYS CONFIDENT, ALWAYS RIGHT.

Is absinthe a precious stone or a liquor? You probably know the right answer. But how certain are you that your answer is right? If you are 100 percent certain, you are dead sure. There's no way you can be wrong. Ninety percent certainty is a little lower but still quite confident. Eighty percent a little lower still. But if you only give yourself a 50 percent chance of being right, it's a toss-up, a random guess, and you're not confident at all. In 1977, psychologists Paul Slovic, Sarah Lichtenstein, and Baruch Fischhoff used a series of questions and a rating system like this one in order to test the confidence people have in their own judgments. What they were looking for was not how many questions people got right or wrong. They were interested in calibration: When people said they were 100 percent confident, were they right 100 percent of the time? Were they right in 70 percent of the cases in which they gave a 70 percent confidence rating? That's perfect calibration-proof that they are exactly as confident as they should be.

The researchers found that no one was perfectly calibrated. In fact, their confidence was consistently skewed. When the questions were easy, people were a little underconfident. But as the questions got harder, they became more sure of themselves and underconfidence turned into overconfidence. Incredibly, when people said they were 100 percent sure they were right, they were actually right only 70 to 80 percent of the time.

This pattern has turned up in a long list of studies over the years. And, no, it's not just undergrads lacking in knowledge and experience. Philip Tetlock discovered the same pattern in his work with expert predictions. Other researchers have found it in economists, demographers, intelligence a.n.a.lysts, doctors, and physicists. One study that directly compared experts with laypeople found that both expected experts to be ”much less overconfident”-but both were, in fact, equally overconfident. Piling on information doesn't seem to help, either. In fact, knowing more can make things worse. One study found that as clinical psychologists were given more information about a patient, their confidence in their diagnosis rose faster than their accuracy, resulting in

<script>