The Enigma of Reason, by Hugo Mercier and Dan Sperber (Harvard UP, 2017), 296 pp.
The Enigma of Reason, written by Hugo Mercier and Dan Sperber, is interesting and informative, packed with valuable insights on reason from new research in cognitive science and behavioral psychology. But it is also limited and reductionistic.
The enigma of reason, as Mercier and Sperber see it, is actually a double enigma: Reason appears to be a uniquely human “superpower”; yet it is also a “flawed superpower.” The point is that reason theoretically allows us to do much but in fact all too often disappoints us. (4) The book explores the uses and misuses of reason against the backdrop of contemporary research and debates.
One of the book’s main claims is that the fundamental purpose of reason has traditionally been misunderstood. Its goal, we have thought, has been to facilitate the expansion of knowledge and better human decision-making (4). In fact, we should now see, its function is to “justify our thoughts and actions to others and to produce arguments to convince others to think and act as we suggest” (7). The authors see two main functions of reason related to this: 1) It serves “to provide tools for the kind of rich and versatile coordination that human cooperation requires” (8). 2) It serves “to make communication effective even when the communicators lack sufficient credibility in the eyes of their audience to be believed on trust” (9).
While I will spend most of my time in this review looking further into the evidence offered for the thesis and exploring the author’s critique of predominant views of cognitive science and behavioral psychology, I think it important to note that the reductionist thesis is the fundamental weaknesses of the book. Mercier and Sperber set up a false dichotomy early in the book—between reason that would facilitate an expansion of knowledge and better decision-making, and reason that facilitates social belonging—and follow this through to the end. Reason is supposed to facilitate our fitting into social groups, not to solve problems and aid decision-making. The dichotomy is odd, precisely because it is so unnecessary and implausible; and the many wonderful points about how reason facilitates social acceptance are not strengthened by Mercier and Sperber’s claim that they have discovered in this reason’s true function.
As an overview of recent work on cognition from behavioral psychologists and cognitive scientists and a corrective to much work in that area, the book succeeds. But as a theory of reason, more generally, the book fails also in other ways. It does not reflect insights even of classical statements of reason as having theoretical and practical functions, such as we find in Kant, let alone more recent treatments of reason outside of the boundaries of cognitive science and philosophy of mind like Habermas’ reflections on knowledge and human interests or communicative action, which discuss reason’s various purposes: to facilitate instrumental control of the physical world; and to coordinate social interaction; to aid us in achieving freedom. What Kantians or Habermasians might characterize as these various functions of reason are not thematized but instead all these functions are reduced to functions oriented toward social acceptance and social belonging.
That said, the book does much well. One of the things it does well is provide an overview and critical assessment of current debates on reason in cognitive science and behavioral psychology. In Part I, “Shaking the Dogma,” the authors highlight the views of critics of reason and of our reasoning ability, at least contextualizing the present debate in the context of the debate’s history. Among the traditional critical voices is Martin Luther, who noted, “Reason is by nature a harmful whore. But she shall not harm me if I only resist her” (16). But the focus of the chapter and later chapters is on more recent critics who are less concerned with reason’s challenges to religious faith than with errs in judgment in everyday life that occur when employing traditional reasoning processes.
A present challenge to traditional views of reason, known as the “dual process view,” is based on but critical of groundbreaking work of Kahneman and Tversky. Kahneman and Tversky’s early work identifies reason with what have become known as “type 2” cognitive processes associated with traditional logic and argumentation. Their work has highlighted that the failings in these processes are often due to non-rational “type 1” intuitive processes that bias us in some ways, shutting down reasoning processes. Newer work in behavioral psychology questions whether Kahneman and Tversky are correct in maintaining that such “type 1” intuitions are actually non-rational or, for example, whether they might better be viewed as rational and supplying us with “simple heuristics that make us smart,” to quote the title of Gerd Gigerenzer and Paul Todd’s famous book on this topic. David Over was an early theoretician to introduce the “dual process” view that classified “type 1” process as rational. Mercier and Sperber want to avoid directly entering the dual process debate, but their book is “to offer something clearly better” (48). What they do offer is a critique of standard views of inference and reasoning that involves many of what have been viewed as “type 1” processes.
Part II of the book takes this up, focusing on “Understanding Inference.” Here, Mercier and Sperber embrace a Humean position that rejects the identification of inference and reason and also rejects that reason is more reliable than other forms of inference. As they summarize this: “What Hume implied was that reasoning was only one way of performing inferences, and not such a reliable way at that. We agree” (51). Mercier and Sperber thus (at least initially) reject both the broad spread and Non-Humean traditional view that all inference is to be identified as reasoning, juxtaposed to what we have called “type 1” processes and the view that reasoning, as classically understood, should be expanded to include inferences that occur in “type 1” processes that occur in nonverbal practices of people, as well as in children who have not acquired language ability and in animals.
In their analysis of this, Mercier and Sperber specify the following three processes of all cognitive systems: 1) Cognition is fundamentally a means developed in organisms to make their way around and get along in an environment. 2) Cognition involves more information than comes from perception alone. 3) Cognition is performed by specific mechanisms that manage one specific task. While we might think of vision, audition, and olfaction as merely assimilating facts, in fact they are in part constructive. We automatically see in a contextual framework, without thinking about it or willing it—and that sometimes leads us to see mistakenly. We experience visual illusions, such as evident in the picture “Monsters in a Tunnel” (see Link) where images of the same size appear to be of different sizes because of depth perception. This work of contextualizing is an inferential process, yet one in which we typically are not even aware that we are engaged in inference. (see p. 58-60) In “Monstors in a Tunnel,” we assume the monster in the foreground is smaller, but it isn’t. We assume the apparently smaller monster is running from the apparently larger one. Perception alone does not convey those things and they might not be true. Memory, drawing on analogous past experience, is also involved with inference. The looks of horror on the former monstor’s face remind us of the horror we have seen in similar settings. Yet, it could be that both monstors in this picture are running from some other creature still, or from a fire, or flood. Memory in any case also isn’t reliable. Study after study show that memory is reconstructive and we routinely remember things we haven’t seen. Mercier and Sperber argue from the use of inference in these varying areas to a modular view of mind that we have not some overriding power of inference-making, but varying “specialized inferential mechanisms” (64). They argue that our inferences exist on a continuum from those of which we are conscious to those of which we are wholly unconscious. Many of our inferences are simply intuitive. But for that, they argue against the view that we have general power of intuition or inference that is responsible for all of the specific inferences.
The view of inference that they support is modular, but they define this in a way to eschew some debates in philosophy of mind surrounding Jerry Fordor’s influential position on the “modularity of mind.” They think that recent cognitive science supports the view not that inference occurs because of some general character or reason but because of specific mechanism with special functions. They define their view as “modular” without thereby buying into all of Fodor’s arguments that various “input systems” are modular but that reason’s “central processes” are not (74-5).
Inferences, even where they have been viewed as occurring in basic cognitive processes, have traditionally been modeled on logical inference. Pavlov’s dog, in line with this view, would be seen as inferring backwards in a kind of syllogistic manner. “If the bell rings, then there is food.” Mercier and Sperber, however, argue that what is going on is more primary than this and in each case is the result of a specialized module, one that takes advantage of regularities in our environments. (84-86) In line with this, they argue that various modules are “task specific, problem specific, or opportunity specific” (92). In these inferential processes “logic plays a marginal role” (107)
While the authors’ agreement with Hume (51) indicate that they do not think inference making is rational, they are not entirely consistent on this point. Somewhat surprisingly, they note: “What makes agents rational…isn’t a general mechanism or dispositions to think and act rationally, but a variety of inferential mechanisms with different inferential specializations” [my emphasis] (96). The surprising part here isn’t that Mercier and Sperber reject the idea that there is a general mechanism responsible for all inferences in favor of the view that there are specialized mechanisms for specific kinds of inferences. What surprising is that despite earlier statements on the difference between reason and inferences, specified mechanisms for inferences taken together apparently make us rational after all. They underline this point later in the book noting “One of the main claims of this book is that reasoning is not an alternative to intuitive inference; reasoning is a use of intuitive inferences about reasons [emphasis in the original] (133).
In Section II, Rethinking Reason, the authors explore how reasoning is used (Chapter 7), whether reason is a module (Chapter 8), intuition and reflection (Chapter 9) and the purpose of reason (Chapter 10). One of Mercier and Sperber’s main points is that reasoning is more than anything involved in “after-the-fact rationalizations” (109). Or as they succinctly put this: “The main role of reasons is not to motivate or guide us in reaching conclusions but to explain and justify after the fact the conclusions we have reached” (112). In the context of this discussion, they propose the term “epistemic luck.” This describes a situation in which individuals by chance do something that there are good reasons for but they were not motivated in their action by those reasons, and in some cases they are also not able to provide reasons for their epistemically justifiable behavior after the fact. Generally, Mercier and Sperber portray reason as inept, arguing that rather than using reasons to guide our own actions, we provide reasons to guide or motivate others (113). In fact, humans are largely motivated by intuitions. They are not conscious of reasons for their actions (114). The post-facto justifications often serve to bolster our reputations. Failing to provide reasons for our activity would jeopardize them, so we formulate reasons proactively, anticipating that we might be called on to justify ourselves (123). The authors do mention research supporting to views that the reasons we provide a public narrative we try to live up to (124). In my view, though, they short-change such work as they focus on showing the ineptness of reason.
They conclude chapter 10, on how humans use reason, summarizing their views about the main ways people are wrong about how implicit reasons in fact have informed their decision making. 1) We often pretend to have made decisions based on reasons when in fact we reached them intuitively. 2) We may be mistaken about the information we were responding to in making our decisions. 3) Sometimes people are inhibited by reasons that they have considered and move to a different decision than they had originally pursued, but without those reasons still being the ultimate grounds for the revised decision (124-5).
Despite disarming reason, Mercier and Sperber do still see a positive social role for reason-giving. In giving reasons we justify ourselves to others. We also commit to forms of conduct. As they summarize from a more sociological perspective:
Reasons are social constructs. They are constructed by distorting and simplifying our understanding of mental states and 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…plays a central role in guiding cooperative or antagonistic interactions, in influencing reputations, and in stabilizing social norms. Reasons are primarily for social consumption (127).
One need not doubt that “reasons are primarily for social consumption” to still question the extreme way that Mercier and Sperber downplay its importance in expanding knowledge and decision-making. There is much more to say than the authors do say about how committing to certain courses of action plays a strong role in identity formation that does in fact influence our decision-making in dramatic ways.
One of the main topics of the book are intuitions. An intuition is viewed as one form of inference, “a thought that, you feel, you may assert on your own authority, without an argument or an appeal to the authority of a third-party” (135). Mercier and Sperber note that an intuition “is a social move” that can make us vulnerable, since it puts others in place of accepting or rejecting it. They do go on to argue that giving reliable information becomes important for reputational gains. Certain types of reasons will improve one’s reputation. The trust of certain authorities will be accepted and the trust of others will not. As they note, appearing objective becomes important for such reputational gains. Yet their discussion of these issues remains truncated. This is unfortunate, since it ties into one of their early controversial claims in the book—that the traditional view of reason is flawed in seeing its purpose as facilitating the expansion of knowledge and better decisions. For if there is a reason that being viewed as reliable becomes important for reputational gain it is because as groups we also benefit from having more accurate knowledge and reliable information facilitates better decisions. As individuals our own correct assessment of situation also often makes the difference between our own success and failure. So in many contexts, it will be important to surround ourselves with people who are well-informed about the state of the world.
Yet Mercier and Sperber do not focus enough on the social-historical context of that acceptance and rejection of intuitions. In a racist or sexist society, for example, a racist or sexist intuition (a non-justified visceral sense of things) is of course much more likely to be met with acceptance than in a society that is not racist or sexist. It is a pity that they do not more clearly take up a discussion of such concrete details, for it might well be that an exploration of such ideas will unveil ways that reason giving that is not in fact so reflexive and after-the-fact as they suggest. Public statements go into the creation of identities. Evaluations of fairness become expressed in laws. These interact with an affect our own subjective assessments and the kinds of intuitions that are considered legitimate. Evaluations more in this direction direct us to an assessment of whether some laws, social institutions and the like are in fact more rational than others. But a discussion of such matters would take the authors into a direction where it’s necessary to reflect on rationality not only in reference of individual subjective experience and decision making but also in relation to legal systems, social organizations, forms of expression in the lifeworld. The fact that they do not go far enough in this direction is reflective of one of the main problems with Mercier and Sperber’s perspective—namely that it mirrors the lack of reflection on social, historical, and general material place of our cognition that is so typical of work in cognitive science. Their overly keen focus on the retrospective use of reason for personal gain falls short.
Mercier and Sperber do, nonetheless, in a more truncated manner, offer some thoughts of the value of prospective use of reason (Chapter 9). But they are not convinced that reasons power is great. They (almost reluctantly) note that over time reasoning over time leads to a certain convergence in science, but note the lack of such convergence in practical affairs (172). Certainly, they are correct that reason does not do all we might hope and all that its early Englightenment proponents thought it could. Yet even in practical reason, we have seen increasing convergence, at least in political systems, that something of a basic rights regime is better than not. I would maintain, in fact, that the general framework within which we disagree about moral questions in everyday situations often has some parameters within which acceptable debate occurs that are not universally but still very generally accepted. Few would make an argument today that slavery should be allowed. Few would argue that we should reintroduce child labor. There may be more consensus than Mercier and Sperber see. Mercier and Sperber seem far from a position that would allow us to begin adequately to evaluate the rationality of such social systems.
In chapter 10, Mercier and Sperber propose their evolutionary approach in which they aim to show that “the main function of reason is social” (176). They do acknowledge the traditional evolutionary view that “Reason is a means for individuals to acquire superior knowledge and to make better decisions” but falsely attributes to evolutionists that view that the use of reason “elevates humans above all other animals” (179). This latter point is at least less than universally accepted, as Darwin himself has conflicting statements in which he calls into question that humans are the pinnacle of evolutionary development and views the human niche as one among so many others. Mercier and Sperber argue that evolution has lead to various specialized modules of the mind, with reason as one of them. But they again repeat their argument against the view that the function of reason is to facilitate greater knowledge and better decisions (182). Reason, they argue “evolved as a response to problems encountered in social interaction rather than in solitary thinking. Reason fulfills two main functions. One function helps to solve a major problem of coordination by producing justifications. The other function helps to solve a major problem of communication by producing arguments” (183).
Two points seem particularly odd about their argument here. One is that there seems no reason to connect the view that reason facilitates the growth of knowledge and better decisions with the view that it evolved for solitary thinking. What compelling reason is there to accept that the growth of knowledge and better thinking is achieved alone rather than in community? The other peculiarity is, again, the already mentioned false dichotomy: Why would reason have to evolve either to solve problems in the growth of knowledge and to facilitate better decisions or to coordinate social interaction? Why not both or the two in concert?
In parts IV and V of the book, Mercier and Sperber set out to show how their interactionist approach illuminates the role of reason in human affairs (201). In Chapter 11, the first chapter of Part IV, the authors examine the biases of reasoning. Rather than viewing the heuristics as “quick-and-dirty,” they tend to share Gerd Gingerenzer and Gregory Todd’s view that they are often “fast and frugal” means of generating good decisions (208). Yet while noting this, they also underline that such fast and frugal, often biased, thinking is often wrong, and wonder why, if reason is to lead to gains in knowledge, we would so often fall victim to confirmation bias, the availability heuristic and the like. In some of this discussion the authors in fact don’t seem quite clear about whether they really believe that the heuristics are fast, frugal, and generally good, or not. They highlight mistakes made in reasoning with heuristics. Highlighting the ways in which they are good would seem, in contrast to what they do, to add support to the idea that the social function of reason is not at odds with good decisions and the growth of knowledge.
Mercier and Sperber do a fair job of outlining the means taken by those who are in favor of what they call the “intellectualist view of reason” that the function of reason is primarily to facilitate the growth of knowledge and better decisions. As they rightly point out, proponents of this view often indicate that the confirmation bias and other heuristics are simply not rational (217). What I wonder is why, if Mercier and Sperber disagree with this view, they so vehemently contrast the “interactionist view” and the “intellectualist” one. It seems more consistent with their position that if the heuristics generally work well, then they can well be seen as complimenting the growth of knowledge and decision-making, at least over the long haul.
Mercier and Sperber do spend much time outlining the weakness of reasoning. To sum up their arguments, the two main features of reasons are  “it is biased—people overwhelmingly find reasons to support their previous beliefs—and  it is lazy—people do not carefully scrutinize their own reasons” (247). In their summative statements of Chapter 13, the authors speak of other functions that reason has taken on beyond the interactionist ones that they believe comprise its fundamental purpose. They note, though, the problems that occur with such additional roles quoting Keynes: “It is astonishing what foolish things one can temporarily believe if one thinks too long alone” (250). In fact, this quote might have alerted the authors to one of the most serious oversights of their book—of contrasting the growth of knowledge and good decision-making with the interactionist view rather than linking them together. Reason cannot be adequately evaluated as a function of individual agents, even if evaluating how those individual agents use reasoning to gain social reputation and the like.
The authors do write of some of the social ways in which reasoning processes is embedded in social procedures, even in a cross-cultural framework, especially in the last portion of the book. But more thorough analysis of how reason is reflected in the design of social institutions would benefit their project.
Mercier and Sperber conclude their book explicitly indicating how they think they have solved the enigma of reason. “Reason is an adaptation to the hypersocial niche humans have built for themselves. First part of the enigma of reason solved” (330). Further: “Reason properly understood as a tool for social interaction is certainly not perfect, but flawed it is not. Second part of the enigma solved” (331). If, like me, you think Mercier and Sperber have failed in substantiated what reason’s proper role is, then you will also have to reject this conclusion–based on not well-substantiated premises. Further, though this is a banal point, just from a logical point of view, one might wonder about their definition of “flawed” in the above statement. The word, in some pretty standard definitions is described as “not perfect in some way.” So saying reason is imperfect but not flawed is ambiguous at best, and self-contradictory at worst.
Beyond this, Mercier and Sperber speak to a position that various of my comments have been hinting at. They reject that we should view this interactionist position as supporting group selection. The benefits they highlight are achieved in interaction but are, they insist, benefits for individuals not part of a group selection theory. This rather truncated discussion as the general conclusion remains a bit dissatisfying. It would at least be informative to see how groups that care about rational conditions for debate, that care about objectivity, tend to outperform those that do not. Such arguments, which I think might be found provide a possibility of linking epistemic gains over time with an interactionist position. It would involve us in a discussion about reason is connected to certain cares in the world—namely whether it has a demand to cultivate a care for the conditions of its own better use.
In the latter parts of the book the authors do analyze some of the ways reason exercised in society. However, they fail to clearly reflect the subtle ways that reason is socially embedded, as taken up in much sociology and Continental social theory. All that said, The Enigma of Reason is a learned and good book. It has a great many insights, indeed a great many more than I have been able to discuss here.