Andrew Fish's Reviews > Thinking, Fast and Slow

There's something almost amusingly ironic about a book which purports to explain how the brain works written by someone clearly lacking in self-awareness.

Having watched Horizon's documentary: "How You Really Think" and finding the basic idea interesting, but the detail somewhat weak, I thought maybe the book would be better. I was wrong.

The premise of the book, that our brain is divided into two systems: system 1 for fast, intuitive thinking; system 2 for detailed, analytical thinking is both attractive and superficially reasonable. Unfortunately, it's also too simple and the author's mistake is that he takes it as almost a complete blueprint for human beings, not just failing to take into account the impact of learning on intuition, but rejecting it outright. This leads him to some improbable claims: for example, treating the association of tattoos to the lower orders as if it were some kind of genetic programming, when anyone with a rudimentary knowledge of cultural history would be aware the tattoos started as a trapping of a societal elite and only gained their current associations gradually.

And it's clear that Kahneman himself isn't entirely confident in his theory. Not only does admit that it isn't widely accepted in the field, at various points he attempts to bludgeon the reader into accepting as gospel some wild piece of speculation by accusing doubters of being everything from depressed to autistic. And yet the book is filled with demonstrably flawed research and spurious conclusions. Some statements, like the idea that people struggle to hold six-digit numbers in short term memory (something I do every day as part of my job) or that it's not possible to walk quickly and think at the same time, are just the author projecting his own limitations onto his audience, others show more serious failings.

So, for example, in a section on how a bias can be created by "priming" the subject with information, he talks about a paper which demonstrates how people exposed to money are apparently likely to be mean and then congratulates the author of the paper for not hammering the obvious conclusion home (giving the impression he has merely been primed to come to that conclusion himself). At another point, talking about sample sizes, he makes a sweeping statement that even statisticians are likely to underestimate the right sample size for a statistical study to be valid - his grounds, a straw poll of handily available statisticians with no information on the sample size involved. Another study which "proves" that people don't come to the aid of a man having a seizure if they think someone else is around fails to take into account that people might not know how to deal with a seizure and the results might differ if all participants had first-aid training. The author leaps to conclusions like a frog to lily-pads, only with less grace. The language of the book is desperate, pleading to be taken seriously, angry when it's clearly not going to happen. The strange business-speak quotes at the end of each section seem pointless and irritating, the tests presuppose the inability of a participant to admit they don't know an answer rather than simply guessing.

The second half of the book relegates the system 1/2 division to the status of an accepted norm as it talks about a concept called prospect theory, which supposedly regulates our ability to judge risky decisions. Again, the idea is interesting, but there are too many claims made without apparent evidence and even some of the basic statements don't square up. That any theory of risk can fail to take into account the marginal utility of the prospective losses and gains seems over-simplistic. That it dismisses the risk aversion which would make many - myself included - reject almost any bet is unforgivable. I've no doubt there are basic rules which influence the way people judge risk, but to take them out of context and assume they are universally applicable is too naive to be credible.

Part of the problem with any study of psychology, of course, is that the selection of test subjects is invariably less than scientific. Where political opinion polling has evolved to the point where a well-selected cross-section of the population can give a reasonably accurate indicator (although one which can still be completely wrong - see the 1992 UK election), psychological studies are invariably stuck with whoever will volunteer. This means that Kahneman's theory may hold for the kind of people who get involved in studies, but not for the population as a whole. Even amongst a self-selecting sample it's too simple and unscientific to lump everyone together.
Fundamentally, I think this is the problem with this book. A good psychology book presents the reader with tools to understand why a particular person behaves in the way they do; this book suggests there are tools to predict that behaviour. Classic behavioural psychology would, for example, pick up on the scattered references to marriage in this book and see a pattern where an author who is married to a woman whom he clearly sees as an intellectual inferior nonetheless finds that she doesn't always accept his opinion as fact, leaving him a lot less happy than he was on his wedding day. Kahneman, meanwhile, thinks he can tell you how you'll vote in the presidential election based on photographs of the candidates. When the winner of the presidential election is a tall, square-jawed strongman then it may appear that Kaheman has a point, but ask yourself this: why doesn't he ever get 100% of the vote?