Friday, June 22, 2007

Blink by Malcolm Gladwell - a review

A colleague who favors instinctive decisions urged me to read Blink! I decided to overcome my stodgy resistance to pop-psychology books and see if I should be paying more attention to unreasoned impulses. I was surprised to discover that the book is rather more ambivalent than my colleague thought (did she read the whole book?), though it is also a big muddled. Here's my Amazon review ...
Blink: The Power of Thinking Without Thinking: Books: Malcolm Gladwell

The strangest thing about Blink! Is the contradiction between the contents and the cover. Did any of the people quoted on the back cover actually read the book? (Hint: Blurb writers rarely read the books they comment on.) The quotes rave about the power of snap decisions, but Gladwell is much more ambivalent. He's particularly concerned with how racial stereotypes misinform judgments, so much so that Gladwell finds he has an intuitively negative opinion of African Americans -- despite being a black man. I suspect he spent some time thinking about when he was going to introduce his Jamaican mother; the book itself is an experiment in the power of framing and bias.

Contrary to the back cover, and the subtitle of "the power of thinking without thinking", this is a book about both the power and the treachery of the unspeaking mind. On the one hand we have powerful non-verbal detection of deception and emotional context, on the other hand we have unconscious bias based on height and ethnicity, the election of George ... err .. Warren Harding and the shooting of Amadou Diallo.

I'd been expecting a superficial justification of impulsive thinking, so I was pleasantly surprised to find Gladwell's Blink! is a much deeper wrok. On the other hand, the book is also somewhat muddled. Once Gladwell moved away from domains in which natural selection has build powerful non-verbal tools, such as deception, mating and eating, his examples of good impulsive analysis became much less persuasive. The "red team" commander's success seemed to owe much more to correct and measured analysis than impulsive decisions (I'd read the story before by the way. The US military needs to retire a lot of generals.). The Cook County MI algorithm story is about the superiority of an analytic decision tree over both analytic human reasoning and non-verbal impressions. I couldn't tell what the Diallo story was trying to communicate, I think he was saying that under high stress situations human reasoning collapses (the autism connection is highly speculative and has no biological foundation). That's certainly true, but hardly novel.

In other cases I had a "Blink" type suspicion that we was cherry-picking and shading anecdotes. I'm particularly suspicious that there was more to the Cook County story than we were told -- it would be very odd for a test to be so sensitive and specific that prior probability of disease was irrelevant. In an afterword he introduces new research findings that contradict the simplistic models in the early book; that's commendable but it doesn't make the book more cohesive.

I think Gladwell lost out by omitting an evolutionary context to human thinking -- a choice that may reveal his biases. An evolutionary approach to cognition explains why the "silent mind" can do so well with decisions lizards, birds and primates evolved around, such as mating, eating, fighting and deceiving. It also explains why the non-verbal mind can make terrible mistakes when evaluating CEOs, presidents, or cell phones. He could have connected the evolution of mind with his thesis experts do best when they combine the silent mind with formal symbolic analysis (words).

I did learn one or two new things. I was impressed by the research on how easy it is to alter emotional state through priming methods. Maybe those days when everyone around me seems to be driving badly are the results of some particularly noxious talk radio show.

Ultimately Gladwell comes across, to me, as suspicious of the intuitive mind. I think he decides that non-experts should "trust" their intuition in domains where natural selection operates, but that even there they need to identify and adjust for bias based on appearance, gender, ethnicity, race, etc. Domain experts do best when they combine non-verbal intuition with analytic reasoning; they can use the intuitive input as a guide to developing a rational and defensible decision (something non-experts are said to have great difficulty with).

That sounds like a plausible path. When making expert decisions in the non-primeval world, write down that initial "Blink!" impression -- but don't trust it. Adjust for bias and use it as the basis for a time limited and bounded analysis by translating it into a defensible rationalization. Then attack the rationalization. If it survives, then credit the silent mind. If it dies, recognize the failure of the paleolithic mind in a technocentric century.

No comments: