The Enigma of Reason: A New Theory of Human Understanding

by Dan Sperber

  The mismatch between the proper and the actual domain of cognitive modules can be exploited in interaction across species (such as in prey-predator interaction, where mimicry in one species results in false positive for another species, or camouflage, which results in false negatives). It can also be exploited in social interaction within a species. Many aspects of culture are based on such exploitation. Belief in supernatural agents, for instance, may be rooted in a disposition to overdetect agency and intentionality to the point of attributing mental states to nonagents such as the sun or a mountain, and to seeing in natural patterns the effect of the actions of an imaginary agent.35 Mismatches between the actual and proper domains of modules are a bonanza for the development of cultural ideas, practices, and artifacts.

  The proper domain of reasoning is disagreements between oneself and others—clashes of ideas. Reasoning aims at reducing these disagreements by producing arguments to convince others or by evaluating the arguments of others and possibly be convinced oneself. The actual domain of reasoning, the kind of input that triggers its operations, is, we have argued in Chapter 9, the detection of a clash of ideas. Most clashes of ideas are detected in discussion with others. This is not surprising; most clashes of ideas emerge in clashes of people. In these cases, the proper and the actual domains of reasoning overlap.

  Notwithstanding, the overlap between the proper and the actual domains of reasoning remains partial. There are typical false negatives: people in a dominant position or in the vocal majority might pay little attention to the opinion of subordinates or minorities and fail to detect disagreements. There are also false positives: either clashes of ideas that occur between third parties with whom we are not even in a position to interact, as when we read ancient polemics or watch a debate on television, or clashes of ideas within oneself. Clashes of ideas within oneself may be internally generated dilemmas. We cannot make up our minds between two opinions or two possible decisions that have comparable intuitive support. Clashes of ideas within oneself may also consist of adopted artifacts: culturally produced brainteasers for entertainment, or paradoxes for flexing one’s philosophical muscle.

  Why should solving puzzles, thinking about paradoxes, or watching other people argue be, for some people at least, an enjoyable experience? Here again, evolutionary considerations are relevant. The operation of many evolved mechanisms requires energy, time, and effort to fulfill their function, but the benefit achieved may not be perceived by the organism, or at least not sufficiently to motivate the effort. The main biological benefit of sex is reproduction, but animals (including, for most of their history, humans) are not aware that sex produces offspring. Strong desires and sexual pleasure have evolved to motivate animals to mate.

  Once sex comes with hedonistic rewards, these rewards may be achieved or enhanced not just by basic mating but by means of various forms of sexual activity that may or may not contribute to reproductive success. Moreover, among humans, the procurement of these rewards to others may bring economic and social benefits and develop in a creative variety of sexual practices.

  Reasoning is a relatively high-investment cognitive activity bringing indirect fitness benefits. It is ultimately beneficial to one’s fitness, we claim, to overcome the limits to communication imposed by insufficient trust and, in so doing, to be better able to influence others and to accept the influence of others wisely. Such perspectives, however, are unlikely to motivate much investment in reasoning here and now. So, we suggest, the very performance of reasoning provides some hedonistic reward. This may not be true for all humans to the same degree or in the same way. Those who, because of their social position or their personal disposition, are reluctant to argue may nevertheless enjoy watching others do so, and they may use reasoning for evaluation rather than for production of arguments.

  Competitive debates, for instance, spread because they hijack reasoning, putting it to uses it did not evolve for. The audience listens to arguments not to acquire sounder beliefs but for the sheer pleasure of watching a competition and of processing very good arguments. The debaters produce arguments not to convince an audience but to enjoy and display their reasoning skills. Ancient philosophical quarrels, controversies between bloggers on abstruse topics, and formal debates do not belong to reasoning’s proper domain—reasoning did not evolve to process such stimuli—but they fall within its actual domain, and this is what explains their cultural success.

  An understanding of reasoning as a function of an evolved reason module need not conflict with a historical and anthropological interest in the remarkable variety of the use of reasons and arguments across cultures. In fact, we suggest, it helps explain it better and formulate testable hypotheses not just about the varieties of reasoning but also about its very variability.

  Early Reasoners

  Work with adult reasoners suggests that some traits of reasoning are universal. Striking results from developmental and educational psychology show that these traits emerge very early on. The acquisition of reason doesn’t depend on teaching institutions and a reason-oriented cultural environment—children spontaneously use reason to defend their ideas and their actions, and to evaluate the reasons offered by others.

  We start with a fictional character, the heroine of the Countess of Ségur’s Les Malheurs de Sophie (The Misfortunes of Sophie), a bestseller of nineteenth-century French children’s literature. Sophie’s very argumentative character aggravated her mother but delighted readers. Here is an example:

  Sophie: Mother, why don’t you want me to go see the stonemasons without you? And when we go, why do you always want me to stay with you?

  Mother: Because stonemasons throw stones and bricks that could hit you, and because there is sand and lime which could make you slip or hurt you.

  Sophie: Oh! Mother, first I’ll be very careful, and also sand and lime can’t hurt me.

  Mother: You believe this because you are a little girl. But I, being older, know that lime burns.

  Sophie: But Mother …

  Mother, interrupting: So, do not reason so much, and be quiet. I know better than you what can hurt you or not. I do not want you to go in the yard without me.

  Sophie lowered her head and said nothing more; but she took on a morose air and said to herself: I’ll go anyway; this amuses me, I’ll go.36

  Although Sophie is only four years old, she is well able to understand her mother’s arguments and to reply to them. Is this a figment of a writer’s imagination? No, the Countess had been able to observe her eight children, and psychological studies have confirmed that children start to give arguments extremely early, sometimes as early as eighteen months. Three-year-olds exchange arguments routinely, using them in one-third of their (numerous) disputes.37

  Children’s reasoning shares two basic features of adult reasoning: myside bias and low evaluation criteria for one’s own reasons. Judy Dunn and Penny Munn observed that three-year-olds were fifteen times more likely to invoke their own feelings rather than their interlocutor’s feelings in their arguments.38 A three-year-old boy (Hugo’s son Christopher) who wanted to be left alone to climb some big stairs argued that he was a big boy. When a few minutes later he got tired and wanted to be picked up, he pointed out that he was a little boy. The inconsistency didn’t seem to bother him. He also once argued that he shouldn’t go to bed right away because he was too tired. Clearly, there is some room for improvement in how children use arguments (although, to be fair, his parents sometimes used similar arguments without much more regard for consistency).

  Parents will not be shocked to hear that children spontaneously produce many arguments—indeed, like Sophie’s mother, they might even find this skill a tad exasperating. But children do not only use reasoning to explain why they shouldn’t go to bed or why they are entitled to take their sibling’s toys. They also pay attention to others’ arguments and evaluate them.

  Psychologists have conducted many observations of parents interacting with their children in an effo
rt to document differences in parenting styles. When it comes to telling children what they should or shouldn’t do, some parents mostly rely on authority. Others have a more reasoned approach, explaining to the children why they should go to bed, take their bath, pick up their toys, stop bothering their sister, and so forth. The psychologists measured various traits of the children to see whether parenting style had an effect on children’s cognition and behavior.

  These studies showed a clear advantage of the use of reasons: the reasoned approach was “successful in promoting resistance to temptation, guilt over antisocial behavior, reparation after deviation, altruism, and high levels of moral reasoning.”39 It seems that the children had, at least in part, made the reasons given their own.

  An issue with these studies is that they only show a correlation. Parents with a reasoned style have children who do better on a variety of measures. This does not mean that the reasoned style caused the children to do better. Perhaps the same causal factors explain both why some parents are inclined to give reasons for their requests and why children born of these parents are better at resisting temptation or at feeling guilty over antisocial behavior. Even if this concern were dispelled, it could still be that children are influenced in their attitudes by the fact that they are given reasons, and not at all by the quality of the reasons given.

  To more properly test children’s sensitivity to argument quality, with Stéphane Bernard and Fabrice Clément we conducted a simple experiment with small children.40 Using Playmobil toys, the experimenter told a story with a little girl, Anna, who has a dog. One day, the dog has strayed away and the children are invited to help Anna look for it. To help the children find the dog, two new characters are introduced. The first points to the left and says, “The dog went this way because I saw it go in this direction”—a pretty strong argument. The second points to the right and says, “The dog went this way because it went in this direction”—a circular argument. Even three-year-olds were sensitive to the quality of the arguments: they were more likely to be swayed by the stronger one (and two-year-olds already start to display the same skills).41

  Conservation tasks—like the task used earlier in this chapter in which we dropped Play-Doh balls in water—were invented nearly a century ago by Jean Piaget to test children’s understanding of elementary physics and mathematics. You can find online many videos of young children performing these tasks. One such video pits a candid little girl against a wily experimenter. Having placed two crackers on her side and only one on the girl’s side, the experimenter asks, “Do you think that we share those fairly?” “No!” emphatically replies the little girl. The psychologist then breaks the girl’s cracker in two pieces and asks, “Now is it fair?” “Yeah.” “Why?” “Because we both have two!”42

  Conservation tasks can be used to con young children out of their crackers but also to study various aspects of developing cognition, including the importance of social interaction in fostering understanding. In the 1970s, several groups of psychologists—in Switzerland,43 in the United States, and in England, all influenced by Piaget—started asking six- and seven-year-olds to solve conservation tasks in pairs. To make things more interesting, the psychologists created as many pairs as possible in which one child could solve the conservation tasks on her own (“conservers”) and the other one couldn’t (“nonconservers”). The conservers proved very persuasive, being three times more likely to convince the nonconservers than the other way around.44

  The nonconservers changed their mind because the conservers had good arguments. They were not simply following the lead of more confident or smarter-sounding peers. In fact, when conservers and nonconservers discussed other topics, such as “What is the best TV show?,” they were equally likely to win the argument.45

  Typically, when two nonconservers discuss a conservation task, they don’t go anywhere, as they have no grounds for arguing. However, if they fail the task in different ways, a real understanding can emerge from their discussion. In one variant of conservation tasks, children are shown the same amount of water being poured into a thin tall glass on the one hand, and a short wide glass on the other. Fatima might believe there is more water in the thin tall glass, while Mariam believes there’s more in the short wide glass. Fatima can try to show Mariam the error of her ways, and vice versa. Through argumentation, they can come to accept that they were both wrong, and understand the correct answer: there is as much water in one glass as in the other.46

  From very early on, children are influenced by good reasons—not only adults’ reasons but also their peers’ reasons. Indeed, in some cases they seem to pay more attention to their peers’ reasons than to adults’. When a teacher spells out the solution to a math problem, students believe her on trust. They don’t need to pay much attention to her explanations, since they are willing to accept the conclusion anyhow. By contrast, if students disagree among themselves, they mostly change their minds when confronted with good reasons. So, when they do change their minds, they are more likely to understand why they did so. Students can be each other’s best teachers.

  Educators haven’t missed the pedagogical potential of argumentation. Beginning in the early twentieth century and gathering strength in the 1970s, research into cooperative, or collaborative, learning has become “one of the greatest success stories in the history of educational research.”47 Hundreds of studies have shown that when students discuss a task in small groups, they often reach a deeper understanding of the materials.48 Implementing cooperative learning in the classroom is not always easy. There must be disagreement, but not to the point of generating conflict. Letting the students talk things through takes time. Yet in spite of these practical obstacles, by the 1990s more than two-thirds of elementary and middle school teachers in the United States relied on cooperative learning, generally to good effect.49

  Learning to Argue Better

  Claiming that basic traits of reason are universal does not mean denying cultural variation. Similarly, drawing attention to the early developing character of argumentation does not mean denying the difference between the argumentation of a three-year-old and of an adult. If the basic skills that enable producing and evaluating reasons do not require learning, people can certainly get better—much, much better—at argumentation.

  The most basic way in which people become better arguers is by acquiring local knowledge relevant to persuading their audience. For instance, as you get to know your friends’ tastes in movies, you can convince them more effectively to go see a movie you think they’ll like. People can also learn what kinds of arguments are appropriate for a given audience. For instance, scientists know that arguments from authority carry little weight in academic articles. Learning when to argue is also critical. Figuring out the contexts in which argumentation is frowned upon or encouraged is at least as important as learning how to argue. But none of this really amounts to acquiring new reasoning skills or becoming a better reasoner.

  We saw in Chapter 12 that it often makes sense not to bother searching for the strongest arguments at the beginning of a discussion. Finding strong arguments is effortful and is not always necessary; a weak argument might convince the audience. Moreover, the counterarguments offered by our interlocutor help us understand better the opposition and find more appropriate arguments. As the conversation unfolds, people come up with arguments that do a better job at taking the audience’s point of view into account.

  Children offer an exaggerated version of this dynamic. When young children lie, they do a poor job at taking their interlocutors’ point of view into account—for instance ignoring the fact that their mother can see a trail of crumbs leading to the cookie thief. Similarly, young children’s reasons are too blatantly egocentric. Young children are apt to respond to their mother asking why they took away their sibling’s toy with a “Because I want it.” In doing this, they ignore the point of view of the sibling, who wanted the toy as well (and of the mother, who had already figured out as much
). They expose themselves to easy counterarguments: “He wants it too,” “He had it first,” and so on.

  When the child adjusts well to his mother’s counterarguments—by accepting them or refuting them with new arguments—this dynamic is the mark of a well-functioning reasoning mechanism. Some counterarguments, however, are deemed so easy to anticipate that people resent having to produce them. For instance, if the interaction between the child and the mother repeats itself, the mother is likely to become annoyed: “I’ve already told you, that’s not a good reason!” The mother expects the child to learn that his argument is a poor one, and to either admit he is wrong right away or move straight on to a better argument.

  In the preceding example, the child has two things to learn. The first is that there are counterarguments that make his “Because I want it” weak. The second is that failing to anticipate these counterarguments has adverse social consequences—in the form of an annoyed mother. It might take a few iterations, but the child will likely learn to argue better.

  Experiments have demonstrated how people learn to anticipate counterarguments on more complex topics. We reported in Chapter 12 a study by Deanna Kuhn and her colleagues in which discussion between peers enabled adolescents and adults to produce better arguments about capital punishment.50 Among the improvements was an ability to see both sides of the argument. This does not mean that people had necessarily changed their mind to adopt a middle-of-the-road position. Instead, they were able to anticipate some of the arguments for the other side and counter them. One of the participants, who was clearly opposed to capital punishment, cited an argument for the death penalty: “I could understand killing a repeat offender to stop the chain [of murders].” However, he mentioned this argument only to refute it in the next sentence, pointing out that life imprisonment would achieve the same goal.