Book Review - The Enigma of Reason: A New Theory of Human Understanding, by Hugo Mercier and Dan Sperber
Reason holds a unique place in Western folk psychology. Originating with the philosophers of ancient Greece, reason has, in various forms, been identified as a, or the, defining feature of humanity. Yet, from the very start, agreement on the nature and function of reason has remained elusive. For Plato, reason served a primarily argumentative function. As demonstrated by Socrates, reason enabled extended discussion and engagement with interlocutors that, over time, led to the refinement, synthesis and extension of arguments. Knowledge accumulated slowly through accretion around germs of truth that emerged from the social process of reasoning. Aristotle, shortly thereafter, transformed the study of reason by developing the science of logic, which came to be understood as the description of correct reasoning in the way that a grammar defines a language. Like ungrammatical language, illogical reasoning is possible, but erroneous. Mercier and Sperber, cognitive scientists based in France, wade into this debate with their instant classic The Enigma of Reason. This underrated work provides a groundbreaking new approach to reason with tremendous explanatory and predictive power. It deserves to be much more widely read and understood.
The Intellectualists
The tight coupling of logic and reason, originating with Aristotle, is the germ of what Mercier and Sperber call the “intellectualist” tradition. For intellectualists, the epitome of reason was not the Socratic figure in the public square, but rather the reclusive intellect, heroically and independently reasoning his way to knowledge. A stylised assessment of academia today suggests both views continue to have proponents. The Socratic method is widely used as a pedagogical tool in fields such as law and the humanities, while the Aristotelian notion of reason has influenced the experimental study of reason in psychology and cognitive science.
But the debate did not end in ancient Greece. Enlightenment thinkers Rene Descartes and David Hume developed contrasting theories of reason. For Descartes, whose influence continues to strongly influence folk psychology, reason was embodied in his attempt to relentlessly shed himself of opinions and knowledge not grounded in and built up from logically sound reasons. Descartes remains the ultimate intellectualist. Hume rejected this view, arguing that it was simply not the case that humans (nor animals) navigate their environment by attending to reasons. Rather, he said,
“Nature must have provided some other principle, of more ready, and more general use and application; nor can an operation of such immense consequence in life, as that of inferring effects from causes, be trusted to the uncertain process of reasoning and argumentation.”
For Hume, the process of reasoning was ill-suited to the way in which animals and humans navigate the world, which requires continuous production and updating of expectations to guide behaviour. Rather than reason, it was a capacity for inference, the unconscious forming and updating of expectations based on detecting dependable regularities in the world, that characterised most behaviour. The success of individuals, and the success of species on evolutionary timelines, is contingent upon those inferences being mostly correct and so animals have inferential mechanisms highly evolved for their ecological niche. Reason, in this view, is but one form of inference, in which expectations about the future are informed by attending to reasons. But, crucially for Hume, reason was much less reliable than the unconscious and intuitive inferential mechanisms that inform most behaviour. Social norms and customs, developed over generations, were therefore necessary to keep reason in check.
Like many great academic debates, this one is perhaps best explained by Star Trek. Spock is the epitome of the rational thinker, able to dispassionately weigh reasons in his pursuit of truth. Kirk, to the contrary, is driven by intuition, inferring information about his environment and updating his actions on the fly, without attending to reasons. For Mercier and Sperber, it is no coincidence that Spock is only part human.
Dual Processes
While the intellectualist tradition continues to hold sway in our folk psychology, the academic understanding of reason has progressed considerably. Since the mid-20th Century, experimental psychologists have been accumulating growing evidence that our reasoning capacity is riddled with biases and flaws, leading us to systematic illogical conclusions. Perhaps the best known, and widely acclaimed, attempt to understand these observations is “Dual Process Theory”, popularized by Daniel Kahneman’s Thinking, Fast and Slow. The insight developed by Kahneman and his collaborators was that much of what people do when solving reasoning tasks is not explicit, conscious reasoning but, instead, a faster, more intuitive form of cognition. Yet, he argues, we retain the potential to engage in more deliberate and logical reasoning when appropriate. Dual Process Theory identifies multiple cognitive “systems” that mediate these competing processes: intuition (system 1) and reasoning (system 2). System 1 is fast and effortless, while System 2 is slow and effortful.
Kahneman’s framework has significant virtues. It explains the decades of research demonstrating our systematic tendency to reason poorly by locating the source of these flaws in our intuition. It is not that reason is failing, but that individuals are not reasoning at all in those instances. At the same time, it preserves the unique place of reason in human cognition by assigning an entire system (system 2) to that capacity. In so doing, it preserves the intellectualist assumption that reason, when properly applied, is fundamentally logical and that flaws arise from our failure to reason.
Mercier and Sperber, while acknowledging some successes of Dual Process Theory, remain unconvinced. Specifically, they argue that there is a fundamental unresolved tension, which they trace to the intellectualist assumption that the proper function of reason is the objective and logical assessment of reasons. This assumption forces intellectual contortions to integrate and account empirical evidence about how reason actually operates. Specifically, by presupposing that reason ought to be objective and analytical in the production and evaluation of reasons, intellectualists must interpret the ever-growing body of experimental psychology about our cognitive biases as evidence of reasons incapacity. Mercier and Sperber flip the analysis, grounding themselves in the principle that ruthless evolutionary pressures would not have produced a function that was so fundamentally dysfunctional. If reason is so central, and critical, how has evolution allowed us to rely on a seemingly flawed intuition system most of the time? Perhaps, they argue, it is not that we systematically fail to reason properly, but rather than we have been misunderstanding the function of reason all along.
The Interactionist Approach
Mercier and Sperber argue that the intellectualist tradition must be abandoned in favour of an “interactionist” approach. Their departure point is a rejection of the thesis that our use and application of systematically flawed. Theories, they argue, which assume the incapacity of the capacity under study should be inherently suspect from an evolutionary perspective. They have some fun suggesting that, perhaps, the centuries-long failure to reevaluate intellectualist assumptions about reason in the face of overwhelming contradictory empirical evidence is itself evidence that reason might not operate as an objective assessor of reasons.
The problem, they argue, is with the intellectualist assumption that reason functions to bestow individual knowledge seekers with a general purpose faculty for impartial and critical analysis. On the contrary, they argue that reason is but one type of inference among many that humans exercise to anticipate and navigate features of their world.
The central claim of the book is that, rather than a general purpose analytical engine, reason is a module of human cognition that produces intuitive inference about reasons.
First, following Hume, Mercier and Sperber distinguish “inference” and “reasoning”. Inference, as discussed above, is a universal process by which animals (and humans) derive information about the world to guide expectations and actions. Reason, in this argument, is a particular form of inference in which information is derived by attending to reasons.
Second, those inferences are intuitive because the judgments are made without knowledge of the reasons that justify them. Again following Hume, they argue that most decisions are not guided or motivated by our conscious attention to reasons. Rather, we rely on intuition, produced by a variety of adaptive inferential modules, the output of which may be conscious but the operation of which remains unconscious. Intuitions are, in other terms, “conscious conclusions arrived at through unconscious processes”. However, when an intuition is challenged, humans have the unique capacity to effortlessly generate second order (or “metarepresentational”) intuitions about the reasons for our first-order intuition. They provide the following example,
You arrive at the party and are pleased to see that your friend Molly is there too. She seems, however, to be upset. When you have a chance to talk to her, you say, “you seem to be upset tonight.” She replies, “ I am not upset. Why do you say that?” Just as you had intuited that she was upset, you now intuit reasons for your initial intuition. Here are what your two intuitions might be:
First-order intuition: Molly is upset.
Metarepresentational intuition about reasons for your first-order intuition: the fact that Molly isn’t smiling and that her voice is strained is what gives me reasons to believe that she is upset.
For the vast majority of decisions, only the first-order intuition counts because they are normally correct. But, there is a critical context in which metarepresentational intuitions become critical: social interaction. When interacting with others, our intuitively guided behaviour is subject to scrutiny and criticism. Preserving our reputation and status therefore requires the ability to justify and explain our actions and persuade others that they are sensible. This is a deep insight. Reason has been reframed as a module that evolved opportunistically to solve problems arising frequently in the context of human social interaction in which language is central.
The purpose of this specialized module of inference is the production of intuitions about reasons, which enables two main functions:
- Justification: enabling coordination and cooperation by producing justification and explanations for behaviour in the form of reasons; and
- Argumentation: enabling communication by evaluating reasons offered by others.
In their terms,
The ability to produce and evaluate reasons has not evolved in order to improve psychological insight but as a tool for defending or criticizing thoughts and actions, for expressing commitments, and for creating mutual expectations. the main function of attributing reasons is to justify oneself and to evaluate the justifications of others.
This explains why reason is a unique and defining feature of human interaction and cognition: it is an evolved response to humans’ rich and complex social dynamics. In a world where most action is driven by intuition, guided by a variety of inferential mechanisms, we need to be able to justify and explain our behaviour in a social setting. Reason is, therefore, a way to produce reasons for lower-order conclusions operating mostly at the unconscious level to enable communication and cooperation.
Predictions
The interactionist theory is superior to Dual Process theory in both its predictive and explanatory power. It also enjoys the virtue of being parsimonious. Rather than relying on a somewhat arbitrary definition of cognitive “systems”, reason is placed within a broader set of inferential mechanisms that are known to exist (even if we remain uncertain about their physical instantiations).
Similarly, it is more evolutionarily sound because, rather than concluding that cognition is systematically flawed, Mercier and Sperber present a history, purpose and function for reason consistent with the centrality of social interaction to humans. Specifically, they predict that, because most behaviour is guided by intuition, and we only produce reasons for our behaviour when challenged in a social setting, we ought to expect that reason is lazy and biased in our own favour. Most reasons are post hoc rationalisations of decisions already made, so we should expect to be highly capable in the production of self-serving reasons for our behaviour. We should not expect those reasons, nor the decisions that they are produced to justify, to necessarily be logical.
But, at the same time, the theory predicts that we ought to remain critical and analytical in the evaluation of reasons by others. In a social setting, we must constantly evaluate the intentions and trustworthiness of others, so we are evolutionarily adapted to view the reasons of others with a skeptical eye.
Understanding that humans will be systematically biased and lazy in producing their own reasons but critical and analytical in evaluating the reasons of others offers key insights to extend and apply these insights.
Why, And How, To Argue
Henry Farrell, writing at Crooked Timber, recently offered a fascinating application of an interactionist understanding of reason.
The problem is that our individual reasoning processes are biased in ways that are really hard for us (individually) to correct. We have a strong tendency to believe our own bullshit. The upside is that if we are far better at detecting bullshit in others than in ourselves, and if we have some minimal good faith commitment to making good criticisms, and entertaining good criticisms when we get them, we can harness our individual cognitive biases through appropriate group processes to produce socially beneficial ends. Our ability to see the motes in others’ eyes while ignoring the beams in our own can be put to good work, when we criticize others and force them to improve their arguments. There are strong benefits to collective institutions that underpin a cognitive division of labor.
In other words, individual reason, while effective at the things it evolved to do in social settings, is hopelessly unsuited to the job to which it has been directed — dispassionate and neutral weighing of reasons — since antiquity. But, this does not provide grounds to argue that reason is flawed but rather that our understanding of reason is flawed. With an interactionist mindset, the collective process of production, critique and marginal improvement of reasons in social settings is the true purpose of reason. Blinded by the limitations of our own reason, we rely on the criticisms of others to point out our weaknesses, provided that we remain individually predisposed to consider those criticisms.This is a classic Hume conclusion, in which reason is a slave to the passions, but cultural and collective mechanisms are harnessed to orient it more effectively.
Farrell goes on to consider the obligation, difficulty and trade offs in seeking those criticisms,
Most obviously, serious criticism/disagreement is one of the most valuable things that we can do or we can get as public intellectuals (for values of public intellectual that mean no more and no less than someone who wants to think and argue in public). On average, our criticism of others is going to be closer to the truth than our own original thoughts. Furthermore, our original thoughts are likely to be valuable just to the extent that they’re responsive to serious criticism from others, and have been modified in response to previous rounds of criticism. More broadly, reasoning well will often be less about reasoning purely, than being reasonable (i.e. being open to others’ reasoning). […]
As a corollary, what may initially seem to us as trolling (and sometimes, what actually is trolling), may contain valuable criticisms that we may benefit from. The tradeoffs are that diversity of perspective is typically correlated with diversity of goals — someone who disagrees with how you see the world is also likely to want different things from it. But you should still push towards the margins of diversity as best as you can, since it is at those margins that you will get the most unexpected criticisms, even if some of those criticisms are irrelevant, since they presuppose that you should want different things than those that you do want. There are judgment calls as to where you stop — but you should do your best to be open to criticisms that are intelligent, clearly expressed, and plausibly constructive with respect to the goals that you want to achieve, rather than overtly destructive of them.
Farrell is absolutely correct. Adopting an individual stance that is open and receptive to criticism, even when biting, is the critical step in moving beyond the blinders imposed by reason. Producing reasonable criticisms of the reasons offered by others, particularly of those with whom you tend to agree, is one of the most fruitful avenues of intellectual investment. Fostering cultures in which the give and take of productive criticism flourishes remains a vital and underrated problem. None of us are well positioned to individually reason our way to truth, but institutions and practices that constrain criticism of thoughts and ideas are more likely to lead us astray.
But we should not be naive. Where ends are incompatible, the ability of interlocutors to engage and reason towards a more accurate outcome will be fundamentally constrained. We must, as Farrell argues, push towards the margins of diversity of perspective and expect to be uncomfortable in the process. But there will always remain difficult decisions about the margin at which criticism can be engaged with constructively. Internet culture today, I believe, is significantly defined by this question. There is not one correct answer and we should expect to observer many sub-cultures adopting different practices. But, over time, we should seek out and invest in those cultures which are successful at widening the field of constructive debate because they offer the most reliable route to well-reasoned outcomes.
