Strangers to Ourselves

by Timothy D Wilson

  When researchers tested all the children again at the end of the year with an actual I.Q. test, the students who had been labeled as bloomers showed significantly higher gains in their I.Q. scores than the other students did. The teachers had treated the bloomers differently, in such a way that made their expectations come true.

  The teachers' expectations about their students were conscious, but the way in which they made their expectations come true was not. When the teachers expected their students to do well, they unknowingly gave them more personal attention, challenged them more, and gave them better feedback on their work. Myra and David Sadker suggest that a similar self-fulfilling prophecy, operating at a nonconscious level, influences the relative performance of boys and girls in American classrooms. At a conscious level, most teachers believe that girls and boys are receiving equal treatment. In one study, the Sadkers showed teachers a film of a classroom discussion and asked who was contributing more to that discussion-boys or girls. The teachers said that the girls had participated more than the boys. Only when the Sadkers asked the teachers to watch the film and count the number of times boys and girls talked did the teachers realize that the boys had outtalked the girls by a factor of three to one.

  At a nonconscious level, argue the Sadkers, teachers often treat boys in more favorable ways than girls, thereby causing boys to do better in their classes. The nonconscious mind can jump to conclusions quite quickly ("the boys in my math class are smarter"), leading teachers to treat boys in preferential ways-even when they believe, consciously, that they are treating everyone the same."

  It is fair to say that the tendency for the adaptive unconscious to jump to conclusions, and to fail to change its mind in the face of contrary evidence, is responsible for some of society's most troubling problems, such as the pervasiveness of racial prejudice (discussed in Chapter 9). Why would an adaptive unconscious lead to such erroneous inferences? Again, the fact that mental processes have conferred a survival advantage does not mean that they are error free; in fact the advantages they bring (e.g., quick appraisals and categorizations) often have unfortunate byproducts.

  DOING BEFORE KNOWING

  Children are especially likely to act on automatic pilot, with their adaptive unconscious guiding their behavior in sophisticated ways before they are aware of what they are doing or why they are doing it. Nonconscious skills such as implicit learning and implicit memory appear early, before children have the ability to reason consciously at a very sophisticated level. Infants have the ability to remember things implicitly (nonconsciously) at birth or even before (in utero), whereas the ability to remember things explicitly (consciously) does not begin to develop until the end of the first year of life. Further, the parts of the brain that appear to be involved in explicit memory develop later in childhood than the parts of the brain that are involved in implicit memory.'Z

  Adults are often in the same quandary: they have no access to their nonconscious minds and have to rely on their conscious interpreters to figure out what is going on inside their own heads. Adults, at least, have a sophisticated, clever interpreter that often constructs an accurate narrative. Children are especially likely to be in the dark, because their conscious interpreter develops more slowly and does not yet have the sophistication to guess what the nonconscious mind is doing.

  This predicament creates a dilemma for psychologists interested in the development of the mind. One of the easiest ways of assessing what people are thinking is to ask them, and many studies of cognitive development rely on children's self-reports. Because the conscious system develops more slowly than the nonconscious one, relying solely on these reports can yield a misleading answer about the age at which a specific skill or trait develops. This error has been made in some well-known areas of developmental research.

  When do children learn the discounting principle? Both Suzie and Rosemary practiced the piano for half an hour. Suzie's mother gave her an ice cream cone for practicing the piano, whereas Rosemary practiced without receiving an ice cream cone. Who liked playing the piano more? Most adults say that Rosemary did, assuming that Suzie might have been motivated in part by the reward. Because Rosemary practiced without receiving any reward, she probably was motivated more by the intrinsic joy of playing. This is known as the discounting principle, the tendency to lower our estimate of the causal role of one factor (intrinsic interest in piano playing) to the extent that other plausible causes are present (the ice cream cone).

  Developmental psychologists have been interested in the age at which children begin to use the discounting principle. In the typical study, children listen to stories like the one about Suzie and Rosemary and report who liked the activity more. Before the age of eight or nine, children seem to use an additivity principle, whereby they think that people who performed activities for a reward like it more (assuming that intrinsic interest + a reward = greater intrinsic interest). By the age of eight or nine, children begin to use the discounting principle, assuming that people who do things for rewards like them less than people who do not (e.g., intrinsic interest + a reward = less intrinsic interest).

  But studies that rely on what children do instead of what they say show that children can use the discounting principle at a much earlier age than eight or nine. In these studies, children are given a reward for performing an attractive activity themselves, and their subsequent interest in the activity is measured by observing how much they choose to engage in it. For example, Mark Lepper, David Greene, and Richard Nisbett asked three- to five-year-old preschool children to draw with felt-tip pens, which at the time was a novel, fun activity for young children. Some of the kids were rewarded with a "Good Player Certificate" for drawing with the pens and some were not.

  Later the researchers put the pens in the classroom during a free-play period and measured how much time each child spent playing with them. As predicted, the children who had been rewarded earlier played with the pens significantly less than those who had not been rewarded. They seemed to have applied the discounting principle to their own behavior, concluding-not necessarily consciously-that if they played with the pens in order to get the Good Player Certificate, they must not have liked the pens very much."

  Why don't children use this same discounting principle when explaining other people's behavior until the age of eight or nine? Perhaps the adaptive unconscious learns the discounting principle earlier than the conscious interpreter. Young children act according to the discounting principle because their nonconscious inference system is driving their behavior (e.g., whether they play with the pens in the classroom). Interpreting behavior consciously and verbally reporting why it occurred, however, is the job of the conscious system, which takes longer to learn and apply the discounting principle.

  This schism between what people do and what they say persists into adulthood. On the basis of what they do, adults often seem to have discounted their interest in a rewarded activity. During unconstrained, free-time periods, those who have been rewarded for engaging in the activity (such as playing with puzzles) spend less time with the activity than do people who have not been rewarded for engaging in the activity. Given what people reported, however, they did not seem to have discounted their interest in the activity: they said they liked the activity as much as people who had not been rewarded.

  If there really are two systems implicated in these studies, a nonconscious one that determines what people do and a conscious one that determines what people say, are there ways of getting them more in synch? How can the conscious system do a better job of inferring what the nonconscious system already knows? Given that consciousness appears to take longer to learn the discounting principle, maybe it needs a little more of a nudge to apply it. That is, whereas the nonconscious sys tem discounts intrinsic interest in the presence of rewards quite readily, maybe the conscious system has to think about it a little more carefully. I tested this hypothesis with Jay Hull and Jim Johnson in a study in which college students were given a re
ward to play with an interesting puzzle. As in many studies of this type, the students' behavior indicated that the reward reduced their interest in the puzzle: they played with the puzzle less in a subsequent, free-time period than did unrewarded students.

  As is also common, however, the students did not report, on a questionnaire, that they disliked the puzzle-unless they had first been asked to think about the reasons for their actions. Whereas putting people in this reflective mode did not, for the most part, influence their behavior-they still engaged less in an activity if they had been rewarded for it-it did influence their reported liking for the activity. When in the reflective mode, people who were rewarded for doing the activity now reported that they liked it less. These results suggest that when people think about it carefully, they can apply the discounting principle, deducing that they must like an activity less if they were rewarded for doing it. If they are not thinking carefully about it, however, their conscious system fails to apply the discounting principle (which, after all, was learned rather late in development)-even though the adaptive unconscious already has."

  When do children acquire a theory of mind? At sonic point, people come to realize that they are not the only ones with a mind-other people have them, too. Because we cannot tell this directly by looking inside another person's head, we develop what psychologists call a theory of mind-the inference that other people have thoughts, beliefs, and feelings, just as we do. We believe that humans and inanimate objects are quite different (humans have minds, rocks do not), we often look where other people are looking (we want to learn what they are thinking that we are not), we can pretend to be someone else (by simulating their thoughts and feelings), and we often try to deceive other people (by encouraging them to develop false beliefs). All these are signs that we have a theory of mind. We rarely pretend to be a rock or try to deceive a tree, precisely because we presume that they do not have minds that contain beliefs, thoughts, and feelings.

  The prevailing wisdom is that a theory of mind develops around the age of four, as shown by children's performance in what is called the false-belief paradigm. In a typical study, children watch an actor place something in a hidden location. They might see Matt, for example, hide a piece of candy in a box and leave the room. Sally then enters the room, finds the piece of candy, and puts it in a basket a few feet away. When Sally leaves and Matt returns, the stage is set. Where will Matt look for the candy: in the box where he put it, or in the basket where Sally hid it? Most four-year-olds reply to this question by saying, "the box where he hid it." They recognize the seemingly obvious point that Matt still believes the candy is in the box because he did not see Sally put it in the basket. Most three-year-olds, however, say that Matt will look in the basket where Sally hid the candy. They seem unable to separate their own knowledge from another person's, assuming that because they know that the candy is in the basket, Matt knows this too. They do not yet have a well-developed theory of mind that tells them that other people can have different beliefs from their own.

  Or do they? Wendy Clements and Josef Perner performed an intriguing variation on the false-belief task that suggests that even threeyear-olds have a theory of mind, at least at an implicit or nonconscious level. Their study was very much like the one described above, except that in addition to asking the children where Matt would search for the candy, they also observed where the children looked when Matt returned to the room: Did they look in the location in which Matt had hidden it, or in the location where it had been moved by someone else? The researchers assumed that children would look first to the location in which they anticipated Matt would search for the candy. If they had a correct theory of mind, they should look where Matt thought the candy was, not where they knew it was. If they did not have a correct theory of mind, they should look where they knew it was, not where Matt thought it was.

  On the standard measure of where children say Matt will look, the researchers found the same thing as previous studies: almost none of the very young children (those between the age of two years five months and two years ten months) got the question "right"; that is, almost all of them said that Matt would look for the candy in the basket, where they knew it to be-suggesting that they did not yet have a theory of mind. In the older groups, the percentage of children who gave the right answer steadily increased, such that by the age of four, most of the children gave the right answer.

  As for where children looked when Matt reentered the room, the youngest children's gaze was consistent with their verbal reports: they looked at the basket where they knew the candy was and said that this was where Matt would look. That is, both measures indicated that these children did not have a theory of mind. However, the two measures diverged dramatically in children right around three years of age. They looked in the correct location, even though they gave a different answer when asked where Matt would search for the candy. Judging by what these children did, they had developed a theory of mind earlier than revealed by what they said. The children who were three years eight months and older looked in the correct location and gave the correct answer when asked.''

  The best explanation of this and subsequent studies is that the looking and verbal measures reflect different kinds of knowledge that develop at different rates. The looking measure may have tapped a nonconscious, implicit type of knowledge-in my terms, knowledge acquired by the adaptive unconscious-whereas the verbal measure tapped a conscious understanding of the theory of mind that takes longer to develop.

  There is even evidence that nonhuman primates have a rudimentary theory of mind, judging by where they look during a false-belief task like the one described above. Thus, very young children, and possibly even nonhuman primates, may possess a nonconscious theory of mind that guides their behavior. This view is quite compatible with the developmental literature on children's understanding of the discounting principle. Developmental psychologists who rely too heavily on verbal measures may not be giving children their due. They are studying children's verbal, conscious system, which may develop more slowly than the adaptive unconscious.'

  Does the conscious system ever catch up? Perhaps people's conscious abilities are especially limited early in life, but when they reach adulthood they acquire a full-blown, conscious self and achieve greater insight into their adaptive unconscious. Although people's conscious theories and insights surely become more sophisticated as they age, there is reason to believe that people do not gain perfect insight.

  One example is people's ability to detect complex patterns in the environment. As we have seen, the nonconscious system is skilled at quick, accurate pattern detection. Recall the study by Pawel Lewicki, Thomas Hill, and Elizabeth Bizot mentioned in Chapter 2, in which people learned a very complex rule that predicted where the letter X would appear on a computer screen, as indicated by the fact that their performance improved over time and deteriorated when the rule was changed. None of the participants ever learned the rule consciously; the adaptive unconscious clearly outperformed the conscious system in this case.

  Numerous studies on covariation detection show that the conscious system is notoriously bad at detecting correlations between two variables (e.g., whether there is a relationship between people's hair color and their personalities). In order to detect such relationships, the correlation has to be very strong, and people must not have a prior theory that misleads them about this correlation. For example, many people persist in believing that they are more likely to catch a cold when they go outside without a coat on a winter day, even though there is no evidence that exposure to cold weather is related to catching a cold. Most people are unaware of the relationship between touching their noses and eyes with their fingers and catching a cold, even though there is good evidence that this is the main way in which rhinoviruses enter our bodies. The adaptive unconscious is not perfect and may not have recognized this covariation either. Or maybe it has, preventing us from touching our eyes even more than we do!''

  IS THE ADAPTIVE UNCONSC
IOUS MORE SENSITIVE TO NEGATIVE INFORMATION?

  Now we come to the most speculative point about differences between nonconscious and conscious processing: there may be a division of labor in the brain, in which the unconscious is more sensitive to negative information than the conscious self.

  As mentioned earlier, Joseph LeDoux has shown that animals and people possess preconscious danger detectors that size up their environments very quickly. The sensory thalamus evaluates incoming information before it reaches conscious awareness. If it determines that the information is threatening, it triggers a fear response. In evolutionary terms, it can be seen how adaptive it is for the brain to trigger a fear reaction to a dangerous (i.e., negative) stimulus as soon as possible.

  Recall also the experiment by Antoine Bechara and his colleagues, in which people developed gut responses signaling them which decks of cards had the better monetary payoffs-before they knew consciously which decks were the best. The cards in decks A and B resulted in large gains or losses of money, adding up to a net loss if played consistently. The cards in decks C and D resulted in small gains or losses of money, adding up to a net gain if played consistently. People quickly developed gut reactions (as indicated by their skin conductance responses) warning them that decks A and B were to be avoided.

  But how did their adaptive unconscious figure this out? One possibility is that it kept a mental tally of the different cards and figured out that on balance, decks A and B resulted in a net loss. It is also possible, however, that it had a simpler strategy: avoid big losses. If the nonconscious system is especially sensitive to negative information, it should focus on the large losses that sometimes came up in deck A. An intriguing implication of this finding is that the nonconscious system will not always make the correct choice. For example, if on balance decks A and B resulted in a higher payoff despite its occasional big losses, then the adaptive unconscious would shy away from the decks that would make the most money."