Daniel Kahneman, the cognitive psychologist who with Amos Tversky won the Nobel Prize for Economics, has a major book out. Thinking, Fast and Slow is basically an effort at a popular synthesis of the state of research into human cognition and specifically into the heuristics-and-biases approach he and Tversky founded. It’s also a memoir of Tversky, a look over the debates between the cognitive psychologists and the economists, between the heuristics-and-biases school and their rivals (who Kahneman thinks have a point), and an attempt to operationalise some of this stuff.
This could easily have been a book that would have been worth mocking. A worse writer could have made it an annoying exercise in Enlightenment fetishism, an airport brick about how you too can get smarter by being more like Homo economicus through these three simple techniques everyone should know. I should know – I bought it in an airport, on a business trip, and read it on a plane. Fortunately, though, rather than one of Robin Hanson’s pals at a Koch Industries economics department near you, it’s by a proper scientist in a hard science, where being wrong is something that actually matters.
Wrongness is a major theme in the book, although not necessarily in the way you’d expect. Kahneman and Tversky’s research was based on the notion that people are wrong in regular and predictable ways, and studying these predictable errors of reasoning would cast light on the nature of intelligence, as well as perhaps helping to avoid them. These are the famous cognitive biases they listed in the classic article Judgment under Uncertainty, which is reprinted as an appendix to the book (it would be worth the price of the book in itself). On the other hand, the naturalistic decision-making school around Gary Klein argued that although people may well be reliably wrong in some ways, they make many more right decisions than wrong ones, and it ought to be worth knowing how. (This reminded me of Johan Galtung’s crack that plenty of people study the causes of war, but perhaps they ought to look at the causes of peace.)
The two schools of thought developed a hearty loathing, distinctive tone and style, and distinctive methodologies – Kahneman, whose training in psychology was in the highly experimentalist and numerical field of human performance and perception, tended to rely on carefully designed experiments with endless students staring into high-speed cameras while doing mental arithmetic against the clock, while Klein and his colleagues were keen on anthropological fieldwork, shadowing fire chiefs around New York City in an effort to document how experts made intuitive judgments of complex problems.
Thinking, Fast and Slow is a shot at synthesis between the two. Kahneman argues that although we are dogged by the reliable biases of intuition, we also tend to undervalue the power of the quick, parallel, associative, social, physically embodied, and effortless computation going on in what he (quoting Canadian psychologists Keith Stanovic and Richard West) calls System One. System One, among other things, is the home of genuine expertise – the coup d’oeil Klein’s firemen had gained through long practice and nasty surprises. It also knows how to empathise with others, operate the human body and every tool we use to extend it, and monitor our environments for the unusual with uncanny attention.
Unfortunately, it’s also systematically dreadful with numbers, in subtle and annoying ways (statistics in general stump it, but it likes to work with averages, its assessments of probability overweight both rare and certain outcomes, while also ignoring quite big differences in likelihood in the middle of the range), it hoovers up every scrap of information it can regardless of source or plausibility and works it into its judgments, and the cornucopia of processing power its massive parallelism provides means that it will work out all the possible answers with the information it has, for good or ill. It’s a sucker for a good story, and it has a tendency to answer the questions it can rather than the ones it’s asked.
System Two is the unified, serial, individual, disembodied, algorithmic mind we call “I”. It can do sums and apply general intelligence to new problems, and it can intervene in the activity of System One. However, it’s not much for the physical world, and its resources are strictly limited. Using it rapidly burns up actual physical fuel, running down the stock of glucose in the blood – as we tire, we rely more on System One, and we make more mistakes. Judges are more likely to grant prisoners parole if they’ve just had their breakfast. (If you want an airport factoid, Kahneman says eat a good breakfast.) Of course, if he’s right, this says some worrying things about the quality of the original judgments.
Thinking, Fast and Slow also says a lot of worrying things about the quality of economics, going right back to the 1950s. Kahneman provides a fantastic anecdote about how Paul Samuelson and Kenneth Arrow (among an impressive list of economists) were fooled by the Allais paradox, an experiment which reliably makes a large majority of people behave not just irrationally in terms of expected value, but inconsistently – they or rather we act in one question as if we were picking the biggest potential pay-off, but in the second as if we were opting for certainty above all. Its inventor, Maurice Allais, apparently expected the demonstration to collapse the whole project of rational choice economics, but the economists simply confessed their bafflement and proceeded to completely ignore the whole issue.
Kahneman is also pleasantly tough on some of his own ideas. In the 1950s, he was assigned by the Israeli army to devise psychometric assessments to choose potential leaders from the annual classes of conscripts. The ones they were already using, based on a structured interview with psychodynamic principles, had turned out to be completely unpredictive of future performance. He invented a system based on a list of factual questions designed to score the recruit against a number of qualities. The interviewers were outraged, feeling themselves deskilled and insulted. As a sop, he included a new question which asked them to dismiss the recruit, shut their eyes, and give a purely subjective rating on a scale of 1 to 10.
The subjective score turned out to be as good as any of the factual questions in predicting their later performance, and in the final version, it was assigned an equal weighting with them. You don’t often meet a book about being more rational that is sceptical of classical rationality; this one is, and as such it is worth reading. Some readers will note an echo of the Freudian view of the mind in the two-systems concept, and I think they would be right. Psychoanalysis may be a shadow-influence on the whole book, in fact – Kahneman is clear that part of his purpose is to provide language, a form of words, to express these problems in conversation. Jacques Lacan argued that the unconscious has the structure of a language, and it is worth noting that it is System One that deals in words. However, I’m not convinced by the talking-points that close each chapter, probably where the book gets closest to its friends on the airport bookshelf.
It is probably worth mentioning, as a finish, the fascinating idea that there may be a difference between intelligence and rationality – that the two are independent vectors. Kahneman gets this from Stanovic and West, who theorise that as well as intelligence in the raw-smarts sense, there is a further independent quality of reason that denotes executive function, kinaesthetic skill, and meta-cognition, the awareness of one’s own thinking and its limits (the inverse of the famous Dunning-Kruger effect). If this is so, we might imagine four groups of people – those unfortunate enough to end up without much of either, who are stupid but so much so that they can’t do too much damage, the geniuses blessed with both, the great competent majority with more rationality than intelligence, and a fourth category, those people who are impressively intelligent but somehow manage to be terribly wrong when it matters and a real pain in the arse to boot.
These men are dangerous – they think too much, or perhaps they reason or feel or whatever the right word is too little. And their intelligence means that their errors can be unusually disastrous, especially as it can lead them into positions of power and responsibility. Further, they are easily mistaken for eccentric geniuses, just as snobbery and racism lead people to confuse groups three and one.
The best thing I can say about Thinking, Fast and Slow is that a Nobel Prizewinner in Economics has successfully explained Larry Summers.