A test of your deductive powers: on the table lie four cards, and you know that each has a letter on one side and a number on the other. Facing up you see E, K, 2 and 7. A rule has been proposed: “Each card with a vowel on one side has an even number on the other.” Which cards do you turn over to test the rule?
This is Peter Wason’s “selection task”, beloved of psychologists interested in human reasoning, and a persistent demonstration of human error. Reword the question, train the participants in logic, offer monetary incentives – and the majority of people still get the wrong answer. They neglect to inspect the vital 7 card, which could potentially falsify the rule, and instead choose the 2 card, which can only confirm the rule (even if it has a consonant on the other side, that doesn’t contradict the stated rule).
Here is Hugo Mercier and Dan Sperber’s enigma of reason: Reason is held up as a uniquely human ability, the thing that separates us from the animals, yet decades of research in psychology suggest that our powers of reason are contaminated by pervasive biases. Reason is a “flawed superpower”, unique and yet weirdly unfit for purpose.
The resolution of this enigma is to shift our construal of the “design problem” that reason solves. Most investigators of reasoning, as Mercier and Sperber portray them, subscribe to an “intellectualist” tradition that views reasoning as an engine for solitary ratiocination. On this view, our reason should ape the principles of deductive logic. Against this specification, reason arguably falls short. Experiment after experiment shows that even individuals with a training in logic often fail to apply logical principles, or apply them with wild inconsistency. Our decisions are systematically distorted by irrelevant factors, not least the widespread tendency to seek to confirm what we already believe to be true.
Mercier and Sperber place themselves at odds with this intellectualist tradition, proposing an “interactionist” view in which reasoning is primarily a social, rather than an individual, tool. Here the purpose of reasoning is not inquisitive, but instead justificatory – we provide reasons to other people, and we evaluate the reasons provided by other people. The niche of reasoning is in the highly social world of human cooperative groups, a niche where it is highly advantageous to be able to transfer information and trust between individuals who are not kin. This view puts the flaws of individual reasoning in a new perspective – the bias towards confirmatory enquiry is now a feature, rather than a bug. It allows efficient division of cognitive labour between interlocutors. While the interactionist view predicts that individuals will be biased and lazy in their reasoning about their own beliefs, it predicts the opposite when we are faced with arguments produced by others, expecting us to be sceptical, but not pig-headed. Mercier and Sperber introduce a tantalising array of evidence to show that this is the case. Not only are individuals far less gullible when evaluating arguments produced by people other than themselves, they are willing to abandon their beliefs when confronted with strong arguments.
All this runs counter to the popular view of human reasoning. Psychologists have been making hay out of human irrationality for decades. This has promoted the “nudges” of behavioural economics, decision-making by markets rather than deliberation, and a general scorn for the individual’s potential to meet standards of rationality. Collective decision-making has a similarly low reputation. Consideration of the social aspects of reasoning is as likely to evoke images of group-think and partisan bias as it is images of civil and productive discussion.
Mercier and Sperber’s book is timely in that it gives a counterpoint to this picture. Yes, individuals are biased, and group deliberation isn’t always successful, but here is important evidence on the conditions when exchange of arguments can lead groups towards the truth (or at least reduced error).
If small groups are given the selection task described above and a chance to discuss it, they convert majority failure to majority success – most groups get the right answer. Analysis of the transcripts shows that if one participant hits upon the right answer, they can generally convince the others. Argumentation can weave the truth out of flawed intuitions. Other experiments show that the necessary conditions for successful argumentation also apply when the right and wrong answers aren’t so clear cut (with strong arguments trumping weak ones). Even children as young as three prefer genuine to circular reasoning on the lines of: “The dog went this way because he went in this direction.”
The key insight is that by producing reasons, individuals commit themselves, both to their interlocutors and to the reasons themselves, staking their reputation on their claims. This creates a mutual space of reasons. Mercier and Sperber argue that such a space can be far more than a mere local consensus, but can be – and has been – a mechanism for collective truth-seeking, as deliberation sorts strong from weak arguments, truth from error.
Their account is impressively synthetic in the range of disciplines it draws on, from legal theory to anthropology. In this, it demonstrates the strength of a mature cognitive science approach that draws on scholarship across philosophy, linguistics, computer science and evolutionary theory. It is also refreshingly European in its sources at a time when accounts of human (ir)rationality can seem dominated by North American voices.
The interactionist account is an antidote to the dual-process models, which dominate most reasoning research and which are the inspiration for the title of Nobel prizewinner Daniel Kahneman’s famous Thinking, Fast and Slow. Mercier and Sperber show that the opposition of intuition (fast) and reason (slow) that defines dual-process accounts is unnecessary. Reason can be viewed as an instance of intuitive inference, rather than a novel mental mechanism that poorly implements deductive logic.
This lets us transcend the dual‑process accounts of reason, accounts that owe as much to religious metaphor as science (“rationality versus intuition” is really just “reason versus emotion”, which in turn is really just “spirit versus flesh”). The scientific sterility of dual-process accounts contrasts with the interactionist account of reason, which suggests rich fields of further investigation, both in direct tests of Mercier and Sperber’s particular account and in a renewed focus on reasoning as a social activity. This would explore mechanisms that allow social coordination, information transfer and the management of reputation and trust rather than the solitary deployment of abstract logic.
An account of why we reason, and how we can reason most successfully, is timely. Not only are homages to our irrationality scientifically sterile, but they are socially corrosive. The reality of our collective reason is threatened by our lack of faith in it but shored up by investment in it. We need a faith in reason, and this book provides strong arguments that such faith is reasonable.
Tom Stafford is senior lecturer in psychology and cognitive science at the University of Sheffield.
The Enigma of Reason: A New Theory of Human Understanding
By Hugo Mercier and Dan Sperber
Allen Lane, 416pp, £25.00
Published 5 April 2017
Although raised an atheist, he says, he was “also encouraged to try to understand the viewpoint of people with whom I might deeply disagree, including religious believers”. This led to him “wondering ‘why do humans disagree so much among themselves if they are all rational? What could be the function of reason, if it is not to put us on the path to knowledge?’ Collaborating with Hugo [Mercier] made me think we could actually answer these questions”.
Hugo Mercier is a researcher at the French National Centre for Scientific Research in Lyon. After growing up in Brittany and completing his first degree, he was “lucky to find universities with a cognitive science programme – first in Lyon, then in Paris, where I did my masters. These were small classes of mostly passionate students, and it was very intellectually stimulating”.
Mercier asked Sperber to become his PhD supervisor because he was impressed by how “Dan brilliantly applied [evolutionary psychology] to human reason, challenging the usual, rather unconvincing story about the evolution of reason.
“Among experts at least, it is becoming increasingly clear that simple dual-process models are conceptually problematic and do not fit with a wide range of data. At this stage, I think that the main reason they are still used is the lack of a consensual alternative. It’s too early to tell, but obviously I hope that our theory will become a part of this alternative.”