Book Review: "The Enigma of Reason" by Hugo Mercier and Dan Sperber

Why do Humans Have so Many Cognitive Biases?

In The Enigma of Reason, Mercier and Sperber argue that human reasoning is not ultimately based on strict logic and that it evolved for social reasons. They argue that reason is lazy and biased but works well when used in argumentation.

Overall Rating: 4.5/5.0

Why do humans have well over one hundred cognitive biases? This is a question I asked while being interviewed for a hedge fund whose founder was obsessed with psychological and personality tests. Despite their strong interest in the subject, the only answer I got was that sometimes evolution gets minor things wrong, like the continued presence of an appendix in humans. Like the authors of this book surely would, the answer struck me as inadequate. Over one hundred cognitive biases do not indicate some minor imperfection that can be dismissed with a hand wave brushing them away.

In this book, Mercier and Sperber argue that the whole notion of human reasoning and why it evolved is off. Rather than being a bug, the over one hundred cognitive biases are a feature. Since my favorite books not only challenge ideas so widely accepted that they are rarely questioned but also offer credible alternatives, this book absolutely fulfills the criteria and makes my list of important reads ever. 

The Wason Tasks

Mercier and Sperber's central thesis is that human reasoning is not ultimately based on a general reason module that functions by employing rules of logic to solve problems in general. They cite numerous experiments showing just how poorly people tend to perform tasks easily solvable by employing elementary rules of logic. A prime example cited is the so-called "Wason Tasks," an example of which is:

Given cards that are known to have a letter on one side and a number on the other and which, face up, show "EK72," which cards would you turn over if asked to evaluate the claim "If 'E' appears on one side, then '2' appears on the other?" Only 10% of the general population can solve this problem, and the statistic rises to only 20% even when limited to a sample of students in top universities. Furthermore, training does not improve results on similar tasks, nor do monetary rewards. Turning the problem into something more "real world" only improves the results slightly.

Evolutionary Psychology 

The Wason Tasks are just one example of extensive empirical research the authors cite which casts doubt on whether human reasoning is ultimately based on employing rules of logic. Once presented, the authors then discuss why reasoning based on a general module for applying rules of logic seems unlikely evolutionarily. Nature, they argue, tends to prefer specialized modules which are more efficient. This, they argue, seems like it would certainly be the case unless the conjectured general logic module applied to something essential to flourishing across a wide range of essential tasks routinely encountered in the environment of evolutionary adaptedness. The authors, however, believe logic is most suitable only for what would have been primarily recreational puzzles. Instead of making inferences based on logic, they argue, our inferences are based, instead, mostly on regularities we have noticed in our environment and which we, thus, expect to repeat.

If not Logic, What?

If not based on logic, what then is going on? The authors argue that we possess a general module to produce and evaluate reasons. This, they argue, is vital to our hypersocial species. When we do something, we prepare reasons justifying it to others. At a social level, this allows others to know the norms we operate on, which we reject, and what to expect from us in the future. 

The authors argue that the reasons, however, are not why we do what we do. That, instead, is based on intuitions that tell us what we want to do. The reason module then attempts to find reasons as rationalizations. If we can find ones that seem like they would be acceptable when explained to others, we go ahead with what we want to do; if not, we may decide not to. Provocatively, the authors argue that we do not understand the reasons for our own intuitions regarding what we want to do any better than we understand why others do what they do. Thus, they posit the same module exists both to attribute reasons to ourselves and to others. (This is not to say that we will attribute to everyone the same reason for doing something; individual variables such as what we imagine their competence to be are considered.)

The authors believe that another key function of the general reason module is to evaluate the reasons given by others. They even go so far as to claim that the reason module only indirectly allows us to reason about the world as a side effect of its ability to produce and evaluate reasons. We are evaluating reasons, they argue, not the underlying things the reasons are about!

The module also exists to produce reasons to try to convince others to take courses of action that we favor. In particular, the authors strongly believe in an "argumentation model" in which humans are best able to reason. Here, people are not reasoning on their own in pursuit of abstract, objective truth but are reasoning in small groups through argumentation with each other to reach conclusions. The authors present considerable empirical evidence to back this claim; for example, that performance on the Wason task rises to as high as 80% when performed by small groups, which come to their conclusion through discussion and argumentation.

Writing Style

Mercier and Sperber are excellent writers. Or, perhaps, more consistent with their theory, they have produced an excellent piece of writing by working together and through civil argumentation. Although the subject matter is on the cutting edge of philosophy and psychology, they present it in a way that is accessible to motivated non-professionals. They primarily do this by keeping it grounded in real-world examples. 

Starting about 30% of the way in, the book gets into some rather abstract and technical discussions. For example, they discuss reasoning about representations of representations. The examples keep it both accessible and engaging, however. That is not to say that the book is a breeze. Some parts will need to be reread. As such, although I usually prefer to listen to audiobooks, the denser parts made this unfeasible for about 25% of this book because of the need to reread. Once about two-thirds through, however, it became suitable for audio listening again. 

Evaluation of Arguments

The book makes a strong case logically, or, if you prefer, after reading it, in terms of reasons in addition to the empirical evidence it presents. I did, however, find a couple of problems:

First, the authors only seem to discuss the alternatives of a generalized reason-producing and evaluating module versus a general logical inference module. I believe there are at least a couple of other explanations for all the biases worthy of discussion, even if not ultimately correct. One is that human biases tend to be in the direction of overconfidence. It could be that confidence is just a trait selected for sexually because it is attractive even to the point of losing grounding with whether the confidence is optimal in terms of truth-knowing. Another logical possibility for all the Myside biases is that humans fall broadly into "leaders" and "followers." For the latter, it might be best to just not generally question what leaders say or what the social norms are and, instead, accept any reason from them as good enough.

The second weakness I detected with the book is that it tends to see the world of less formal, further away from rigorous logic through rose-colored glasses. For example, the authors argue for interpreting language in terms of common sense instead of literal and strict logical meaning. The problem is that, in practice, people often conflate something with an informal, well-understood, and uncontroversial meaning with something more controversial if applied literally and then make conclusions from there based on the latter. For example, "a woman has the right to do whatever she wants with her own body" is undoubtedly not controversial if an implicit "in general" or "nearly always" is assumed. Conflating the informal with the strict meaning and then concluding the latter trivially proves that abortion is justifiable is an example of a non-sequitur arising from a motte-and-bailey that many are prone to.

Conclusion

The Enigma of Reason presents a convincing case that human cognitive biases are not a mistake of evolution but something that exists, or at least existed, for good reason. Although many have speculated that social considerations are part of the explanation, the authors present a detailed conjecture about what the exact social explanations are.

The same hedge fund I asked why the cognitive biases exist later gave me a psychological test showing that I was in the bottom 1% regarding trusting my intuitions. Although this was presented positively as being in the top 1% in terms of logical thinking, I have since learned that downplaying one's intuitions at critical times can be costly. This book helped elaborate on why that is and, thus, hit quite close to home. Even if it then goes slightly too far in making a case for intuition over logic, it does not throw logic out and indeed praises it as one of humanity's greatest achievements. Given the previous historical lopsidedness of philosophers arguing for logic as the ultimate basis of human reasoning, this book is bound to be tremendously enjoyable to those with a serious interest in philosophy, psychology, and the intersection of the two.