WHEN AN informant tells you that six people he knows are planning on detonating a dirty bomb in Boston, should you believe him? That is no hypothetical question -- which is why psychologists who study lie detection, like Paul Ekman, a retired psychologist at the University of California at San Francisco, are getting more calls from the government than they used to.
Ekman's specialty is the human face and the emotion it reveals. How good is he at spotting lies? In Malcolm Gladwell's new book "Blink" -- an exploration of the brain's ability to make good snap judgments -- Ekman is quoted as saying that he knew President Clinton was lying the second he saw the videotape of the president's infamous finger-wagging statement about Monica Lewinsky. Ekman saw giveaway flashes on Clinton's face -- a "hand-in-the-cookie-jar, love-me-Mommy-because-I'm-a-rascal look," he says.
Reviewing the book in The New Republic, however, the federal appellate judge Richard Posner dismisses Gladwell's "credulous" account of Ekman's claims that careful analysis of facial expressions can tell us whether a person is lying. So where does the science of lie detection stand? Does Ekman -- or anyone -- have the goods?
Most studies, as it happens, find that the average person using just his five senses does little better at spotting a liar than chance would predict. But Ekman has long argued that the main problem with those studies is that there's nothing at stake -- like, say, jail time -- in lies told in university laboratories. When you offer significant rewards and punishments, as Ekman does in his experiments, you create more realistic "high stakes" lies that mirror real life.
"I always make a point of saying there are no absolutely reliable signs of lying," Ekman says. "There are signs of strong emotional or cognitive loads -- which means that at a particular moment a person is feeling very strong emotions that are not being expressed in words."
In a typical Ekman experiment, two volunteers walk through a room in which sits a $50 bill. One of them has to pick it up. After the "crime," Ekman interrogates them on videotape; then other volunteers have to pick out the liars on the tapes. (The students who successfully lie get to keep the money; those who are caught lose not only the $50 but also the token fee they receive for volunteering.)
The success rate of most observers hovers just above 50 percent, or chance levels. But across different experiments Ekman finds that certain people consistently pick out the liars roughly 75 or even 90 percent of the time. That suggests they are seeing something. The question is what -- and can we learn from them?
Ekman thinks the key to the "tell," as a poker player would put it, is small flashes of emotion lasting as little as 5th of a second. These "micro-expressions" might express anger (at the questioner) or disgust (at oneself). Using something called the Facial Action Coding System, which he developed in the 1970s, Ekman and his colleagues go back over videotaped interviews, recording on paper every lip curl, nostril flare, and eye glare of the subjects.
Ekman himself claims to have absorbed, to the tune of 80 percent accuracy, how to spot micro-expression tip-offs. But critics say his papers describe his techniques only vaguely and offer no advice on whether an angry curl of the lip represents guilt at being caught or righteous anger at being doubted.
"Nowhere in his research does he describe the methods by which he works," complains Jennifer Vendemia, an assistant professor of psychology at the University of South Carolina and a specialist in deception.
Vendemia herself remains a fan of the old-fashioned polygraph, which she says can come close to 90 percent accuracy when dealing with clear-cut questions like, "Did you take the money?" (A National Academy of Sciences panel in 2003 was somewhat more skeptical, putting the accuracy at merely "well above chance.")
With the help of a $5 million grant from the Department of Defense, Vendemia has lately been giving laboratory liars MRI scans, hoping to learn more about how the brain resolves internal conflict while prevaricating. And other researchers report some success at using thermal-imaging devices to unobtrusively measure subtle blushing effects around the eyes of deceivers.
Some of the most interesting findings have to do with what liars don't do. On average, they apparently do have more dilated pupils than truth tellers, according to a review of the literature published in 2003 in Psychological Science, and they are less pleasant to their interrogators. But they don't fidget. They don't blink. And despite near universal belief -- firmly held in at least 75 countries, according to a survey organized by a Texas Christian University psychologist -- their eyes don't wander. Even mediocre liars can look you right in the eye, as they slip in the shiv.
Christopher Shea's column appears in Ideas biweekly. Email firstname.lastname@example.org.