On television, it all looks so simple. For a fraction of a second, the suspect raises the corner of his mouth. He is happy because he thinks the investigators are wrong about where he planted the bomb. But when his interrogator mentions the correct place, the terrorist’s face betrays a flash of rage. And he shrugs his shoulders as he pronounces his innocence. The evidence is open-and-shut as far as the expert is concerned: The suspect’s body language contradicts his words. He is lying.
The expert on microexpressions in the TV series Lie to Me is the alter ego of Paul Ekman, age 86, a world-renowned researcher of lying and emotion. He not only advised the creators of the program but has also been called upon by numerous U.S. agencies, such as the FBI and CIA. Ekman’s credo is that the truth is written on our face.
This idea has a long tradition. An ancient Indian text from about 900 B.C. describes the behavior of an attempted poisoner as follows: “He does not answer questions, or they are evasive answers; he speaks nonsense, rubs the great toe along the ground, and shivers; his face is discolored; he rubs the roots of the hair with his fingers.”
On supporting science journalism
If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
In commenting on a case at the beginning of the 20th century, Sigmund Freud wrote, “He that has eyes to see and ears to hear may convince himself that no mortal can keep a secret. If his lips are silent, he chatters with his finger-tips; betrayal oozes out of him at every pore.” Since the middle of that century, security experts in the U.S. have been trying to separate truth from fiction by using lie detectors. Among other things, so-called polygraphs register how sweat production, heart rate and breathing change when certain questions are asked. But in everyday life we use the same instruments as our ancestors in distinguishing truth from falsehood: our eyes and ears.
From 1971 to 2004 Ekman was a psychology professor at the University of California, San Francisco, where he is now emeritus. Starting before that time, he became the first researcher to examine, on a large scale, how observable changes in the face and body reflect truth telling or lying. In the 1960s he formulated his theory of universal facial expressions for the basic emotions: anger, disgust, enjoyment, fear, sadness and surprise. Ekman categorized the facial muscles involved in producing these expressions in what he called the Facial Action Coding System. He and his co-author Wallace V. Friesen laid the groundwork for Ekman’s popular theory of lies in their 1969 paper “Nonverbal Leakage and Clues to Deception,” which dealt with patients’ nonverbal signals. The core idea: emotions that one seeks to conceal are sometimes betrayed by facial expressions and movements of the arms, hands, legs and feet. A prime example is a momentary facial expression that lasts no longer than a quarter to a half a second and is virtually invisible to an unpracticed observer.
Paul Ekman. Credit: Steven Dewall Getty Images
Such microexpressions that reveal concealed emotions do not, however, occur all that often, according to Ekman. We are more apt to observe emotions that are broken off or incomplete. For example, if we try to fake fear or sadness, the characteristic creases on our forehead may not show. And the eye muscles may not be involved in a false smile. Ekman does not believe that such discrepancies are proof of falsehood. He merely thinks they are indications that something might be off. That is why repeated and varied clues are necessary; one is not enough. In his book Telling Lies, Ekman claims that, in laboratory experiments, truth and lies can be told apart by facial expression alone with an accuracy of more than 80 percent—and that the figure reached 90 percent when factors such as facial and body movements, voice and language were all included in one analysis.
But these statistics may be misleading. According to Maria Hartwig of the John Jay College of Criminal Justice, such claims are “simply implausible.” The research literature, by contrast, suggests that success rates are generally barely above chance. Even when Ekman requires extensive training of testers, he has apparently not published a single study that confirms his figures.
Legal psychologist Kristina Suchotzki of Johannes Gutenberg University Mainz in Germany says, “Many researchers don’t take Ekman’s idea of using microexpressions to uncover deception especially seriously.” And it is not only because of a lack of empirical evidence. The theory is itself inadequate. “Just because someone is afraid during an interrogation doesn’t mean they’re lying,” she says. “You cannot infer a deception from an emotion.”
Suchotzki is currently the German researcher most active in the field of lying. She focuses on evidence of mental effort that may be associated with false statements. It is simply not easy to lie. One must make an effort to hide the truth, come up with a plausible alternative story, put oneself in the interrogator’s shoes and keep a tight rein on feelings that could give up the game—while seeming authentic the whole time. “Up to now emotions and cognitions have been studied separately,” Suchotzki says. “I want to bring both together and bring clarity to what happens in the mind when one is lying.” And she does not think that using microexpressions to recognize deception is an especially promising approach. “There are simply no studies that support Ekman’s claims,” Suchotzki says.
One of the few independent studies on this theme was conducted by psychologists Stephen Porter and Leanne ten Brinke in 2008. Their test subjects were asked to conceal their true feelings upon viewing sad, fear-inducing and joyous pictures. If they tried to mimic a different emotion, their facial expressions were more often dissonant or incongruous. Microexpressions were seen in 2 percent of all snapshots. They occurred in 22 percent of all test subjects—though not only when they tried to cover up their feelings.
There is one thing that Ekman and his critics agree on, however: humans are generally very poor lie detectors. The most cited hit rate comes from a meta-analysis and is based on about 25,000 test subjects. They guessed right in only 54 percent of cases—just barely better than chance. For audiotapes alone, the success rate was 63 percent, which means it was significantly higher than it was for videotapes with or without sound. Apparently, the image distracts the viewer from noticing relevant clues. And it does not matter that a professional—a police officer, judge or psychiatrist—has more frequent occasion to deal with lying: the so-called experts did no better than the proverbial person on the street.
But what happens when an individual knows someone as well as their own child? A Canadian experiment studied whether parents can recognize their children’s lies better than other parents or undergraduates. All three study groups looked at videos in which eight- to 16-year-old kids and teenagers told the truth or lied about whether they had peeked at answers to a test. Parents looking at their own children were no better at distinguishing the truth from a lie than were other parents or undergrads. Participants in all three groups might as well have tossed a coin, although they tended to trust their own judgment—and, in particular, the parents assessing their own children tended to believe them.
One of the study’s co-authors, Kang Lee, a psychologist at the University of Toronto, could not let this subject go. During a TED Talk, he presented a photograph of his son lying. Lee used a method called transdermal optical imaging, which measures blood flow in the skin, to see what was behind his son’s neutral facial expression. He calls what he discovered the Pinocchio effect: during a lie facial, blood flow decreases in the cheeks but increases in the nose.
In response, however, Suchotzki notes, “The idea that blood perfusion could be an indicator of lying is absurd. Such claims are dangerous because they suggest that such practices might be useful in public spaces, such as airports.” This kind of effect might be evident in a controlled laboratory experiment, she says, but no technology can resolve the fact that characteristics of lying can be observed in suspects who are telling the truth. “There is no such thing as a clear-cut sign of lying—only indications that may allow us to conclude that a lie may have been told,” Suchotzki adds.
In a meta-analysis performed by a team led by psychologist Bella DePaulo, 14 of 50 nonverbal characteristics were observed to be more frequently associated with lying, most especially dilated pupils and tenseness. But the most telling was the impression made by the statements themselves. False statements tended to be hesitant, ambivalent and unsure. A German meta-analysis of 41 studies found something different, however. Psychologists at Justus Liebig University Giessen found that lying was particularly associated with evidence of self-control: fewer movements of the hands, legs, and feet and less head nodding.
“The effects are so minute and unstable that they cannot help us to identify lying in practice,” Suchotzki says. Linguistic characteristics have been shown to be more telling. “But these effects are not large, and the findings do not justify optimism,” she adds.
Psychologist Aldert Vrij of the University of Portsmouth in England, one of the most active researchers of lying, does not think much of nonverbal characteristics of deception either. In an overview study, he, Hartwig and their colleague Pär Anders Granhag of the University of Gothenburg in Sweden wrote that such signals are “faint and unreliable.” The researchers are pinning their hopes more on linguistic clues—although these are hardly more associated with lying than nonverbal ones. Yet they can be induced and strengthened by questioning techniques, as several experiments (including those conducted by Vrij’s team) have demonstrated. Such extensive research on nonverbal characteristics does not exist.
It is no wonder that is the case: language is simply easier to record. Reliably capturing facial expressions and gestures requires specially trained observers or more complex wiring of the face and body. Researchers have only recently been increasingly experimenting with computer-assisted methods such as automatic facial recognition. This technology promises new understandings because it can process enormous data streams and identify complex patterns.
Vrij, Hartwig, and Granhag admit that more subtle nonverbal characteristics—qualities such as subcategories of facial expressions of the sort that Ekman had defined —have either been overlooked or ignored. If we look closer, we find, for example, that true statements are more often accompanied by demonstrative gestures and that lies are more frequently paired with metaphorical ones, such as a fist as a symbol of strength. Perhaps researchers will discover even more signs, or a combination thereof, when other methods are used.
When Hartwig and psychologist Charles Bond combined various behavioral characteristics in a meta-analysis with thousands of test subjects, they were able to identify about two thirds of the lies correctly. Most studies merely test selected characteristics. In general, laboratory experiments do not re-create realistic conditions. There is no genuine interaction between the investigator and the subject. Even more crucially, the deception is made on demand. And no one can say with certainty the extent to which, and under what, conditions laboratory findings can even be generalized to real offenses.
To make the test subjects feel they have some skin in the game, they are generally promised money if they are convincing. Suchotzki has tried out more challenging measures in the service of science: In one study with her colleague Matthias Gamer of the Julius Maximilian University of Würzburg in Germany, participants were questioned about a fake theft they had committed. The researchers told half of the subjects they would receive a weak electric shock if a computer deemed their statement to be unbelievable. In that group, Suchotzki and Gamer observed a slower pulse rate during untrue responses, along with increased sweating of the hands. Fear of potential consequences increased these differences.
Of course, Suchotzki’s lab subjects actually had nothing to fear if they did not sound convincing. The ramifications were much more serious for involuntary subjects of an investigation in a field study conducted by ten Brinke and Porter. The researchers analyzed videotapes of 78 individuals who turned to the public in their efforts to find a missing family member. About half of them were later found guilty of having killed the missing person.
The guilty and the innocent subjects did not differ in terms of body language, as a comparison of 75,000 still images showed. The authors reported, however, that the faces of the guilty ones exhibited more signs of concealed emotions, such as simulated happiness and sadness. Distressed individuals who were actually innocent conveyed “full-face sadness and distress,” ten Brinke and Porter wrote.
In another analysis, the guilty used twice as many vague wordings, such as “Somebody’s got to know something, somewhere. I think so. I think there’s somebody who’s got to be running scared, who knows what they’re doing.” Genuine appeals sounded clearer and more immediate: “You can’t imagine what Sarah means to us. We are a strong family, and we don’t survive well apart. We need her home now, today, quickly as we possibly can.”
But as impressive as such studies may sound, they still do not resolve the problems of research on lying. The differences are small; the indicators are ambiguous. These results only represent averages, and at best, they offer coarse potential indications in individual cases. A confidently spoken lie can seem more believable than a stuttered truth. That is because most people base their judgment on how confident, clear and unambiguous a statement seems, according to a meta-analysis performed by Hartwig and Bond. When individuals overlook a deception, it is not because they pay attention to the wrong signals. They mostly fail when a person who elicits trust lies or when a seemingly unbelievable person tells the truth.
We can pay a heavy price for not knowing what is going on in others’ mind. It would seem that evolution should have given us a good feel for the truth, and yet we are easily led astray. Perhaps that is the downside to coexistence in society. The harmless lies of everyday life have taught us credulity.
Still, why do so many people think that they can recognize lying? Let us turn the question around: How would it be if lies and truth looked the same, like two eggs? How would it be if the guilty got away, and the innocent paid the price in their stead? Hartwig finds the thought hard to bear. “We want to believe that liars give themselves away,” she says.
This article originally appeared in Spektrum der Wissenschaft and was reproduced with permission.