Predictably Irrational

by Dan Ariely

  The ads we posted around Sproul Plaza read as follows: “Wanted: Male research participants, heterosexual, 18 years-plus, for a study on decision making and arousal.” The ad noted that the experimental sessions would demand about an hour of the participants’ time, that the participants would be paid $10 per session, and that the experiments could involve sexually arousing material. Those interested in applying could respond to Mike, the research assistant, by e-mail.

  For this study, we decided to seek out only men. In terms of sex, their wiring is a lot simpler than that of women (as we concluded after much discussion among ourselves and our assistants, both male and female). A copy of Playboy and a darkened room were about all we’d need for a high degree of success.

  Another concern was getting the project approved at MIT’s Sloan School of Management (where I had my primary appointment). This was an ordeal in itself. Before allowing the research to begin, Dean Richard Schmalensee assigned a committee, consisting mostly of women, to examine the project. This committee had several concerns. What if a participant uncovered repressed memories of sexual abuse as a result of the research? Suppose a participant found that he or she was a sex addict? Their questions seemed unwarranted to me, since any college student with a computer and an Internet connection can get hold of the most graphic pornography imaginable.

  Although the business school was stymied by this project, I was fortunate to have a position at MIT’s Media Lab as well, and Walter Bender, who was the head of the lab, happily approved the project. I was on my way. But my experience with MIT’s Sloan School made it clear that even half a century after Kinsey, and despite its substantial importance, sex is still largely a taboo subject for a study—at least at some institutions.

  IN ANY CASE, our ads went out; and, college men being what they are, we soon had a long list of hearty fellows awaiting the chance to participate—including Roy.

  Roy, in fact, was typical of most of the 25 participants in our study. Born and raised in San Francisco, he was accomplished, intelligent, and kind—the type of kid every prospective mother-in-law dreams of. Roy played Chopin études on the piano and liked to dance to techno music. He had earned straight A’s throughout high school, where he was captain of the varsity volleyball team. He sympathized with libertarians and tended to vote Republican. Friendly and amiable, he had a steady girlfriend who he’d been dating for a year. He planned to go to medical school and had a weakness for spicy California-roll sushi and for the salads at Cafe Intermezzo.

  Roy met with our student research assistant, Mike, at Strada coffee shop—Berkeley’s patio-style percolator for many an intellectual thought, including the idea for the solution to Fermat’s last theorem. Mike was slender and tall, with short hair, an artistic air, and an engaging smile.

  Mike shook hands with Roy, and they sat down. “Thanks for answering our ad, Roy,” Mike said, pulling out a few sheets of paper and placing them on the table. “First, let’s go over the consent forms.”

  Mike intoned the ritual decree: The study was about decision making and sexual arousal. Participation was voluntary. Data would be confidential. Participants had the right to contact the committee in charge of protecting the rights of those participating in experiments, and so on.

  Roy nodded and nodded. You couldn’t find a more agreeable participant.

  “You can stop the experiment at any time,” Mike concluded. “Everything understood?”

  “Yes,” Roy said. He grabbed a pen and signed. Mike shook his hand.

  “Great!” Mike took a cloth bag out of his knapsack. “Here’s what’s going to happen.” He unwrapped an Apple iBook computer and opened it up. In addition to the standard keyboard, Roy saw a 12-key multicolored keypad.

  “It’s a specially equipped computer,” Mike explained. “Please use only this keypad to respond.” He touched the keys on the colored pad. “We’ll give you a code to enter, and this code will let you start the experiment. During the session, you’ll be asked a series of questions to which you can answer on a scale ranging between ‘no’ and ‘yes.’ If you think you would like the activity described in the question, answer ‘yes,’ and if you think you would not, answer ‘no.’ Remember that you’re being asked to predict how you would behave and what kind of activities you would like when aroused.”

  Roy nodded.

  “We’ll ask you to sit in your bed, and set the computer up on a chair on the left side of your bed, in clear sight and reach of your bed,” Mike went on. “Place the keypad next to you so that you can use it without any difficulty, and be sure you’re alone.”

  Roy’s eyes twinkled a little.

  “When you finish with the session, e-mail me and we will meet again, and you’ll get your ten bucks.”

  Mike didn’t tell Roy about the questions themselves. The session started by asking Roy to imagine that he was sexually aroused, and to answer all the questions as he would if he were aroused. One set of questions asked about sexual

  preferences. Would he, for example, find women’s shoes erotic? Could he imagine being attracted to a 50-year-old woman? Could it be fun to have sex with someone who was extremely fat? Could having sex with someone he hated be enjoyable? Would it be fun to get tied up or to tie someone else up? Could “just kissing” be frustrating?

  A second set of questions asked about the likelihood of engaging in immoral behaviors such as date rape. Would Roy tell a woman that he loved her to increase the chance that she would have sex with him? Would he encourage a date to drink to increase the chance that she would have sex with him? Would he keep trying to have sex after a date had said “no”?

  A third set of questions asked about Roy’s likelihood of engaging in behaviors related to unsafe sex. Does a condom decrease sexual pleasure? Would he always use a condom if he didn’t know the sexual history of a new sexual partner? Would he use a condom even if he was afraid that a woman might change her mind while he went to get it?*

  A few days later, having answered the questions in his “cold,” rational state, Roy met again with Mike.

  “Those were some interesting questions,” Roy noted.

  “Yes, I know,” Mike said coolly. “Kinsey had nothing on us. By the way, we have another set of experimental sessions. Would you be interested in participating again?”

  Roy smiled a little, shrugged, and nodded.

  Mike shoved a few pages toward him. “This time we’re asking you to sign the same consent form, but the next task will be slightly different. The next session will be very much the same as the last one, but this time we want you to get yourself into an excited state by viewing a set of arousing pictures and masturbating. What we want you to do is arouse yourself to a high level, but not to ejaculate. In case you do, though, the computer will be protected.”

  Mike pulled out the Apple iBook. This time the keyboard and the screen were covered with a thin layer of Saran wrap.

  Roy made a face. “I didn’t know computers could get pregnant.”

  “Not a chance,” Mike laughed. “This one had its tubes tied. But we like to keep them clean.”

  Mike explained that Roy would browse through a series of erotic pictures on the computer to help him get to the right level of arousal; then he would answer the same questions as before.

  WITHIN THREE MONTHS, some fine Berkeley undergraduate students had undergone a variety of sessions in different orders. In the set of sessions conducted when they were in a cold, dispassionate state, they predicted what their sexual and moral decisions would be if they were aroused. In the set of sessions conducted when they were in a hot, aroused state, they also predicted their decisions—but this time, since they were actually in the grip of passion, they were presumably more aware of their preferences in that state. When the study was completed, the conclusions were consistent and clear—overwhelmingly clear, frighteningly clear.

  In every case, our bright young participants answered the questions very differently when they were aroused from when they were in
a “cold” state. Across the 19 questions about sexual preferences, when Roy and all the other participants were aroused they predicted that their desire to engage in a variety of somewhat odd sexual activities would be nearly twice as high as (72 percent higher than) they had predicted when they were cold. For example, the idea of enjoying contact with animals was more than twice as appealing when they were in a state of arousal as when they were in a cold state. In the five questions about their propensity to engage in immoral activities, when they were aroused they predicted their propensity to be more than twice as high as (136 percent higher than) they had predicted in the cold state. Similarly, in the set of questions about using condoms, and despite the warnings that had been hammered into them over the years about the importance of condoms, they were 25 percent more likely in the aroused state than in the cold state to predict that they would forego condoms. In all these cases they failed to predict the influence of arousal on their sexual preferences, morality, and approach to safe sex.

  The results showed that when Roy and the other participants were in a cold, rational, superego-driven state, they respected women; they were not particularly attracted to the odd sexual activities we asked them about; they always took the moral high ground; and they expected that they would always use a condom. They thought that they understood themselves, their preferences, and what actions they were capable of. But as it turned out, they completely underestimated their reactions.

  No matter how we looked at the numbers, it was clear that the magnitude of underprediction by the participants was substantial. Across the board, they revealed in their unaroused state that they themselves did not know what they were like once aroused. Prevention, protection, conservatism, and morality disappeared completely from the radar screen. They were simply unable to predict the degree to which passion would change them.*

  IMAGINE WAKING UP one morning, looking in the mirror, and discovering that someone else—something alien but human—has taken over your body. You’re uglier, shorter, hairier; your lips are thinner, your incisors are longer, your nails are filthy, your face is flatter. Two cold, reptilian eyes gaze back at you. You long to smash something, rape someone. You are not you. You are a monster.

  Beset by this nightmarish vision, Robert Louis Stevenson screamed in his sleep in the early hours of an autumn morning in 1885. Immediately after his wife awoke him, he set to work on what he called a “fine bogey tale”—Dr. Jekyll and Mr. Hyde—in which he said, “Man is not truly one, but truly two.” The book was an overnight success, and no wonder. The story captivated the imagination of Victorians, who were fascinated with the dichotomy between repressive propriety—represented by the mild-mannered scientist Dr. Jekyll—and uncontrollable passion, embodied in the murderous Mr. Hyde. Dr. Jekyll thought he understood how to control himself. But when Mr. Hyde took over, look out.

  The story was frightening and imaginative, but it wasn’t new. Long before Sophocles’s Oedipus Rex and Shakespeare’s Macbeth, the war between interior good and evil had been the stuff of myth, religion, and literature. In Freudian terms, each of us houses a dark self, an id, a brute that can unpredictably wrest control away from the superego. Thus a pleasant, friendly neighbor, seized by road rage, crashes his car into a semi. A teenager grabs a gun and shoots his friends. A priest rapes a boy. All these otherwise good people assume that they understand themselves. But in the heat of passion, suddenly, with the flip of some interior switch, everything changes.

  Our experiment at Berkeley revealed not just the old story that we are all like Jekyll and Hyde, but also something new—that every one of us, regardless of how “good” we are, underpredicts the effect of passion on our behavior. In every case, the participants in our experiment got it wrong. Even the most brilliant and rational person, in the heat of passion, seems to be absolutely and completely divorced from the person he thought he was. Moreover, it is not just that people make wrong predictions about themselves—their predictions are wrong by a large margin.

  Most of the time, according to the results of the study, Roy is smart, decent, reasonable, kind, and trustworthy. His frontal lobes are fully functioning, and he is in control of his behavior. But when he’s in a state of sexual arousal and the reptilian brain takes over, he becomes unrecognizable to himself.

  Roy thinks he knows how he will behave in an aroused state, but his understanding is limited. He doesn’t truly understand that as his sexual motivation becomes more intense, he may throw caution to the wind. He may risk sexually transmitted diseases and unwanted pregnancies in order to achieve sexual gratification. When he is gripped by passion, his emotions may blur the boundary between what is right and what is wrong. In fact, he doesn’t have a clue to how consistently wild he really is, for when he is in one state and tries to predict his behavior in another state, he gets it wrong.

  Moreover, the study suggested that our inability to understand ourselves in a different emotional state does not seem to improve with experience; we get it wrong even if we spend as much time in this state as our Berkeley students spend sexually aroused. Sexual arousal is familiar, personal, very human, and utterly commonplace. Even so, we all systematically underpredict the degree to which arousal completely negates our superego, and the way emotions can take control of our behavior.

  WHAT HAPPENS, THEN, when our irrational self comes alive in an emotional place that we think is familiar but in fact is unfamiliar? If we fail to really understand ourselves, is it possible to somehow predict how we or others will behave when “out of our heads”—when we’re really angry, hungry, frightened, or sexually aroused? Is it possible to do something about this?

  The answers to these questions are profound, for they indicate that we must be wary of situations in which our Mr. Hyde may take over. When the boss criticizes us publicly, we might be tempted to respond with a vehement e-mail. But wouldn’t we be better off putting our reply in the “draft” folder for a few days? When we are smitten by a sports car after a test-drive with the wind in our hair, shouldn’t we take a break—and discuss our spouse’s plan to buy a minivan—before signing a contract to buy the car?

  Here are a few more examples of ways to protect ourselves from ourselves:

  Safe Sex

  Many parents and teenagers, while in a cold, rational, Dr. Jekyll state, tend to believe that the mere promise of abstinence—commonly known as “Just say no”—is sufficient protection against sexually transmitted diseases and unwanted pregnancies. Assuming that this levelheaded thought will prevail even when emotions reach the boiling point, the advocates of “just saying no” see no reason to carry a condom with them. But as our study shows, in the heat of passion, we are all in danger of switching from “Just say no” to “Yes!” in a heartbeat; and if no condom is available, we are likely to say yes, regardless of the dangers.

  What does this suggest? First, widespread availability of condoms is essential. We should not decide in a cool state whether or not to bring condoms; they must be there just in case. Second, unless we understand how we might react in an emotional state, we will not be able to predict this transformation. For teenagers, this problem is most likely exacerbated, and thus sex education should focus less on the physiology and biology of the reproductive system, and more on strategies to deal with the emotions that accompany sexual arousal. Third, we must admit that carrying condoms and even vaguely understanding the emotional firestorm of sexual arousal may not be enough.

  There are most likely many situations where teenagers simply won’t be able to cope with their emotions. A better strategy, for those who want to guarantee that teenagers avoid sex, is to teach teenagers that they must walk away from the fire of passion before they are close enough to be drawn in. Accepting this advice might not be easy, but our results suggest that it is easier for them to fight temptation before it arises than after it has started to lure them in. In other words, avoiding temptation altogether is easier than overcoming it.

  To be sure, this sounds a lot like th
e “Just say no” campaign, which urges teenagers to walk away from sex when tempted. But the difference is that “Just say no” assumes we can turn off passion at will, at any point, whereas our study shows this assumption to be false. If we put aside the debate on the pros and cons of teenage sex, what is clear is that if we want to help teenagers avoid sex, sexually transmitted diseases, and unwanted pregnancies, we have two strategies. Either we can teach them how to say no before any temptation takes hold, and before a situation becomes impossible to resist; or alternatively, we can get them prepared to deal with the consequences of saying yes in the heat of passion (by carrying a condom, for example). One thing is sure: if we don’t teach our young people how to deal with sex when they are half out of their minds, we are not only fooling them; we’re fooling ourselves as well. Whatever lessons we teach them, we need to help them understand that they will react differently when they are calm and cool from when their hormones are raging at fever pitch (and of course the same also applies to our own behavior).

  Safe Driving

  Similarly, we need to teach teenagers (and everyone else) not to drive when their emotions are at a boil. It’s not just inexperience and hormones that make so many teenagers crash their own or their parents’ cars. It’s also the car full of laughing friends, with the CD player blaring at an adrenaline-pumping decibel level, and the driver’s right hand searching for the french fries or his girlfriend’s knee. Who’s thinking about risk in that situation? Probably no one. A recent study found that a teenager driving alone was 40 percent more likely to get into an accident than an adult. But with one other teenager in the car, the percentage was twice that—and with a third teenager along for the ride, the percentage doubled again.6