There is a psychologist at Washington State University, John Gottman, who studies marriages in an attempt to predict divorce. Couples, chosen at random, are invited simply to talk to each other about their relationship in a “love lab”. They are wired up to a machine which measures their heartbeat and the clamminess of their hands.

The conversation is filmed and analysed using a complex system in which the emotions of each partner through the conversation are recorded. Twenty different emotional attitudes are coded, from disgust through contempt to spontaneous displays of affection. Three thousand couples have submitted themselves to this ordeal over 25 years, and Gottman has now found that, based solely on his analysis of one hour’s unstructured conversation, he can predict with 95 per cent accuracy whether that couple will still be married 15 years later. Furthermore, the success rate of his system is 90 per cent, even if only a 15-minute segment is used. And he and his researchers can now make these rapid judgments without recourse to the code.

This is just one of a series of intriguing tales that Malcolm Gladwell tells us in Blink. He is New Yorker-trained and has an easy storytelling style. It might be unkindly characterised as being from the “just fancy that” school of journalism, but that would be unfair. For Gladwell is not just telling tales. He has bigger fish to fry.

He has a Theory, with a capital Th. A second one, indeed, following on from his success with The Tipping Point a couple of years ago, which introduced a handy new phrase into the political lexicon.

This time, as best I can tell, his theory is that first impressions, snap judgments, can often be more accurate than those we reach after painstaking analysis, as long as we focus attention on the right things. So Gottman has discovered that certain facial expressions or conversational tics are reliable indicators of contempt — an emotion which is almost bound to be fatal to a marriage. If you focus on that, you can capture the health of a relationship more accurately than you can through a series of structured sessions with a shrink. Or so he asserts.

Of course, as Gladwell recognises, some first impressions are less reliable. Often they are based on prejudice, on grounds of race, age, sex and even size. Harvard’s Implicit Association Test shows that many people who are sure they are not racists find it hard to map positive associations on to blacks. Short guys, too, suffer badly in the US. Only 3.9 per cent of American men are 6ft 2in (1.88m) or taller: a third of Fortune 500 chief executives are. So it is vital to ensure that the early signals you encode are appropriate.

How to do that? Well, maybe Gladwell’s third book will help, because about halfway through this one it drifts away from the central theme and rambles on to the airport bookshelf, with some long and hackneyed war stories about major corporate disasters — New Coke, for example, and, yes, the Edsel, the catastrophic 1950s Ford motor car. (The American appetite for bowdlerised Business School case studies is inexhaustible.) The New Coke story as recorded here is particularly baffling.

Coca-Cola changed their celebrated secret formula on the basis of blind “sip tests”, ie, first impressions, where the market researchers told them they were beginning to fall behind sweeter Pepsi. Yet Coke drinkers typically do not just sip their drink, they down the whole can, and find that the extra sweetness cloys by the 330th millilitre. As I read it, this argues against Gladwell’s Theory. The sip test misled the company into thinking that it needed to change, and it had to revert to “Classic Coke” to stem falling sales.

But I am falling into precisely the trap that Gladwell cautions against — over-analysis based on a close reading of the text. I should stick to my first impression of Blink, which was of course formed before I opened it. The back cover is dominated by a photograph of Gladwell himself — who turns out to be a chap with an engaging smile, but sorely in need of a haircut. Blink is the same.

Explore Books

• Book Extracts

• Book Reviews

• Books Group

• Audio Books

Times Recommends

• The 100 Best Books of the Decade
• The Times picks the 50 best paperbacks of 2009
• The conversation: James Ellroy