The Enigma of Reason: A New Theory of Human Understanding

by Dan Sperber

  Some basic perceptual inferences, we hope you now agree, are not guided by psychological reasons. But does the argument extend to less automatic, more interpretive aspects of perception and to intuitive inference generally? When participants in the Latané and Rodin experiment heard various noises coming from the next room, they recognized some of these as noises of impact, others as words spoken in English. This recognition was, presumably, quite automatic: mentally represented reasons had no role to play in the process (even though, of course, it was an inferential process). Putting these perceptions together and inferring that the woman next door had fallen and hurt herself involved more interpretation, understanding that she might need help even more so. Does this mean that the participants had to have explicit reasons to come to these further more interpretive conclusions or that the mechanisms involved had to have unconscious reasons?

  Any socially competent person familiar with the kind of modern environment where the experiment was taking place would have developed a modularized capacity to make the relevant inferences. They would spontaneously recognize some noises not just as noises of impact but, using subtle acoustic features, as noises of a crash involving both hard pieces of furniture and a softer body such as a human body falling on a hard surface. We suggest that the inferential perception procedure that permits such recognition exploits correlations between features and probable causes of noises without representing these correlations, let alone representing them as reasons.

  When the participants heard the words, “Oh, my God, my foot … I … I … can’t move it. Oh … my ankle! … I … can’t get this … thing … off me,” they spontaneously understood “it” to refer to the speaker’s foot, they understood “this … thing” to refer to some large solid object like a piece of office furniture that had fallen on the person’s foot, and in so doing they were going well beyond what “it” and “this thing” linguistically encode. To perform these pragmatic inferences, participants used a modular comprehension procedure that exploits, without representing them, relevance-based regularities in verbal communication.11 They homed in on one of many linguistically possible interpretations without representing, either consciously or unconsciously, reasons to do so.

  Having understood in this intuitive manner that the woman next door had hurt herself and was communicating about her predicament, most participants would spontaneously conclude that she needed help. Again, we suggest that they could come to that conclusion without having to mentally represent (consciously or unconsciously) a reason for the inference. In fact, if someone had to mentally represent such a reason in order to realize that a person in pain and complaining might need help, this absence of spontaneous empathy might be taken as a symptom of impaired cognitive and social competence.

  Still, there is no doubt that reasons did play a role in Latané and Rodin’s experiment. The question is when reasons appeared in the course of events. After the crash in the next office, some participants stood up and went to see whether they could help. Others remained seated. Was their decision either to try to help or to do nothing a spontaneous decision arrived at without considering reasons? We would suggest that unless they showed some hesitation, there is no strong ground to assume that they had been guided by a conscious or unconscious reason represented in their mind. Be that as it may, a few minutes later, when the experimenter asked them what had gone through their mind, they readily provided reasons that, they claimed, had motivated their action or inaction.

  Some participants, especially those who had been helpful, may have come up with reasons just when they were asked. For them, finding reasons was effortless: in circumstances where it is obvious that someone may need you to help, wanting to help is an obviously good reason. Those who had remained seated and had done nothing had had more time to think about reasons they might invoke to justify themselves. They may also have felt a greater need to think of such a justification. Still, even for them, devising a plausible justification was not that difficult. In both cases, thinking about personal reasons for one’s decision typically came after having made the decision itself.

  If, as we suggest, the point of reasons isn’t to guide the formation of beliefs and the making of decisions, then what are reasons for?

  Reasons Are for Social Consumption

  Whatever humans do is likely to contribute for better or worse to the way they are seen by others—in other words, to their reputation. These indirect reputational effects may turn out to be no less important than the direct goal of their action, whatever it is. Socially competent people are hardly ever indifferent to the way their behavior might be interpreted. By explaining and justifying themselves, people may defend or even improve their own reputation. By failing to do so, they may jeopardize it.

  Thinking about good reasons for their actions is something that people often do proactively, anticipating that they may be called upon to explain or justify themselves. The minute you have engaged in a course of action that may have reputational costs—and sometimes even before, when you are merely considering it—a different mental mechanism may start working. Its function is to manage your reputation and for this, to provide an explanation that will justify your behavior. Participants in the Latané and Rodin experiment who ignored the events taking place in the next room may have been aware that their passivity might be open to criticism. This may have prompted them to think of reasons they might provide to justify themselves.

  The reputation management mechanism acts like a lawyer defending you, whatever you have done. Still, given the opportunity, a lawyer may advise a regular client against a course of action that would be hard or impossible to defend. What happens when the reputational mechanism cannot come up with a good narrative? Then the course of action considered (and possibly already undertaken) turns out to have costs that weren’t initially taken into account in the decision process. In such a case, the failure of the reputational mechanism to produce an adequate narrative may have a feedback effect and cause the initial decision to be rescinded, or at least revised. It is possible, for instance, that a few of the participants in the Latané and Rodin experiment who had first dismissed the idea of helping the person in distress in the next room then realized that they would be hard put to justify their passivity; hesitated, that is, compared the ease with which they could justify their two possible responses; and chose to do something after all.

  There is some fascinating experimental evidence that the search for reasons aimed at justification may, in fact, influence action (more about this in Chapter 14). Still, living up to the story you want to be able to tell about yourself isn’t quite the same thing as telling a true story.

  So, when people produce reasons to explain and justify their beliefs or actions, the narrative they come up with may be at odds with what really happened in their mind in three different ways, each interesting in its own right.

  First, people commonly present themselves as having considered reasons and been guided by them in the process of reaching a belief or a decision that, in fact, was arrived at intuitively. The error we all make here is to falsely assume that we have direct knowledge of what happens in our mind when we draw intuitive inferences and that we are guided by reasons of which we are conscious or that we can easily introspect. Contrary to such an assumption, in coming to an intuitive decision or belief, we are not guided by reasons (not even “implicit” ones).

  Even so, we often do correctly identify a piece of information that served as input to our inferential processes. When this happens, our mistake is just to describe this piece of information as a personal reason. Participants who went to see if they could help the person in the next room had mentally represented the fact that that person was in pain and needed help; what they hadn’t mentally represented at the moment was the higher-order consideration that the person needing help was a reason to go and help. This appropriateness of helping people in need (in certain circumstances at least) is so obvious that in socially competent people, it is expl
oited by a dedicated procedure, which works without representing it. For our social interactions, such a mistaken self-attribution of reasons doesn’t matter at all. In fact, it is less an individual mistake than a socially encouraged use of commonsense psychology. From a scientific point of view, on the other hand, this should be recognized as a misrepresentation: acting spontaneously to help others need not be guided by conscious or unconscious reasons any more than smiling at someone who smiles at you.

  Second, people may be mistaken about the information that served as input to their inferences. Participants who chose not to help were first and foremost influenced by the presence and passivity of the other person in the room, but they failed to acknowledge this crucial factor. In all likelihood, when they claimed that their decision had been motivated by the thought that nothing too serious had happened and that their help was not needed anyhow, they were just reporting thoughts that occurred to them only after they had made the intuitive decision to align their behavior to that of the other person in the room. In giving these reasons to the experimenter, they may have been sincere, even if self-deluded: people who think of themselves as nice and helpful and who realize they didn’t help when they should have may be puzzled by their own behavior. The easy, too easy, solution to this puzzle is to assume that they must have had what looked at the time like good reasons to think that their help wasn’t required.

  Third, there are people who fail to find good enough reasons for what they are about to do and who, as a result, waver and change or at least readjust their course of action. In such cases, the personal reasons people invoke have truly played a causal role in their final decision. These reasons didn’t directly shape the decision that was finally taken, but they inhibited the carrying out of the initial decision. Still, contrary to what these people may believe and claim, the true causal process didn’t go from reasons to decision, but from tentative decision to search for justification, failure of search, and revised decision aimed at a more easily justifiable course of action.

  What we have said so far implies that there is no role for unconscious reasons in human cognition. If unconscious reasons have no role to play, then they probably don’t exist, and if they don’t exist, then implicit reasons don’t exist either. When we attribute to ourselves an implicit reason, we are just interpreting our thoughts or actions in terms of newly constructed conscious reasons that we fictitiously represent as having been implicit before.

  Does all this suggest that psychological reasons, whether “implicit” or conscious, have no reality whatsoever, that they are a pure construction? No, reasons are indeed constructed, but under two constraints that ensure that they have some degree of both psychological and social reality. The reasons we invoke for justification have to make psychological sense. Talk of reasons need not—in fact, we have argued, cannot—provide an accurate account of what happens in our minds, but it tends to highlight factors that did play a causal role. Reasons are typically constructed out of bits of psychological insight.

  In order to fulfill their justificatory function, moreover, psychological reasons have to represent objective reasons socially recognized as good reasons. This social recognition may be biased by many social and cultural factors, but if it were a purely arbitrary affair, how would it hold any sway? For many beliefs and practices—and in particular for the use of reasons as justification or argument—genuine objectivity is, we maintain, a major factor in cultural success (a fact well explained by our account of the evaluation of reasons, as discussed in Chapter 13).

  To attribute reasons to oneself or to others is not so much to formulate a hypothesis about how things are as to construct a tool for social action: justification, evaluation, or, we will see in Chapters 8 through 10, argumentation. But what makes reasons a useful tool?

  When we give reasons for our actions, we not only justify ourselves, we also commit ourselves. In the first place, by invoking reasons, we take personal responsibility for our opinions and actions as described by us, that is, as attitudes and behavior that we had reasons to adopt. We thereby indicate that we expect others to either accept that we are entitled to think what we think and do what we do or be ready to challenge our reasons. When what we thought or did is unlikely to be approved, by giving reasons, we may indicate a line of defense: we had, if not good reasons, at least reasons that seemed good at the time. A defense based on reasons typically allows us to accept responsibility while denying guilt.

  By giving reasons, we also commit ourselves to a future line of thought and conduct. Invoking reasons as motivations of one’s past views and actions expresses a recognition of the normative aptness of these reasons and a commitment to being guided by similar reasons in the future. For our audience, this commitment to accepting responsibility and to being guided in the future by the type of reasons we invoked to explain the past is much more relevant than the accuracy of our would-be introspections. This is why we all pay attention to the reasons of others and why we produce our own.

  To put it in more sociological terms: Reasons are social constructs. They are constructed by distorting and simplifying our understanding of mental states and of their causal role and by injecting into it a strong dose of normativity. Invocations and evaluations of reasons are contributions to a negotiated record of individuals’ ideas, actions, responsibilities, and commitments. This partly consensual, partly contested social record of who thinks what and who did what for which reasons plays a central role in guiding cooperative or antagonistic interactions, in influencing reputations, and in stabilizing social norms. Reasons are primarily for social consumption.

  8

  Could Reason Be a Module?

  On a sunny day, an elder from the Dorzé tribe in southern Ethiopia was exhorting a group of idle young men: they shouldn’t smoke cigarettes—it was against their religion! The young men pointed out that the young French anthropologist (Dan) who had been living among the Dorzé for the past few months was sitting nearby, doing what? Smoking cigarettes! The elder turned to the foreigner and demanded an explanation: “Why do you smoke?” The anthropologist didn’t quite know what to say: “Well,” he mumbled, “my father smokes, my grandfather smoked …”

  The elder turned back to the young men: “See,” he exulted, “the foreigner shows respect for his father and his forefathers. They smoked, so he smokes. Follow his example, show respect for your forefathers: they never smoked a cigarette; you shouldn’t either!”

  As this example illustrates, the same reasons that can be used to justify oneself can also be used as an argument to convince others. (The extra twist here is that when combined with other considerations, the same reasons can be used to justify one individual’s practice and to argue against others adopting it: the import of reasons depends on the context.)

  Retrospective and Prospective Uses of Reasons

  Sometimes people use reasons to explain or to justify decisions already taken and beliefs already held. This is a retrospective use of reasons. Sometimes people use reasons as arguments in favor of new decisions or new beliefs. This is a prospective use of reasons. When reasons are used prospectively, it may be to answer a question of one’s own to which one doesn’t yet know the answer or to convince others of an opinion one already has. In the first case, it is inquisitive reasoning, and in the second case, it is argumentative reasoning. Figure 13 represents these uses of reason.

  Figure 13. Retrospective and prospective uses of reasons.

  The philosophical and psychological literature treats retrospective and prospective uses of reasons as if they were altogether different topics, with little or no attention to what they have in common. We, on the contrary, will argue that there is no dividing line between retrospective and prospective uses of reasons, and that there are many cases of overlap. The use of reasons in explanation is often treated as more important than their role in justification (or, at least, as equally important). We, on the contrary, will argue that the justificatory role of reasons is more important
than their explanatory role. The individual, inquisitive form of reasoning, aimed at answering questions on one’s own, is considered reasoning par excellence, while the communicative, argumentative form is considered secondary. We, on the contrary, will argue that argumentation is primary.

  Reasoning is a major topic in philosophy and psychology in its own right. The dominant view is that reasoning is aimed at truth and at good decisions and should be impartial and objective. The reasons used in reasoning should be impersonal “arguments” that owe their force to their formal properties (studied in logic and probability theory). The use of reasons to explain and justify ideas already adopted or decisions already made and acted upon have, on the other hand, a personal component and cannot be really impartial.

  How warranted is this contrast between impersonal reasoning and personal justification? Are personal considerations always irrelevant in good reasoning? In practical reasoning, at least, the relevance of personal considerations is generally recognized. Not surprisingly, philosophers working on practical reason and moral philosophy, in particular Joseph Raz,1 have been more open to a unified approach to what we call retrospective and prospective uses of reasons.

  There is, anyhow, a simple argument in favor of approaching in a unified framework retrospective and prospective reasons. The argument is that all reasons that can be used in justification can be used in reasoning and conversely. If you doubt this, imagine, for instance, a reason that would be good enough to retrospectively justify the opinion that no one should smoke. Then, surely, this reason could be used prospectively to convince oneself or others that no one should smoke. Conversely, imagine that a math teacher asks her pupils whether there is a greatest prime number and helps them reason and rediscover the proof that there is not. Then, surely, that very same proof, which caused the pupils in the first place to come to the conclusion that there is no greatest prime, becomes thereafter an impeccable justification for their belief, as it already was for the teacher.