Why We’re Bad at Spotting Lies

Part 1 of 8 in a series on lies and deception at work.

Most of us think we can tell when someone’s lying. The research says we’re barely better than a coin toss, and the signs we trust most are the ones that don’t work.

About three thousand years ago, someone wrote down how to recognise a man who means to poison you. He doesn’t answer questions, or answers evasively. He talks nonsense. He rubs his big toe along the ground and shivers. His face changes colour, and he rubs the roots of his hair with his fingers.

Swap the toe for a jiggling knee and you have the modern office version. In 2006 about 90 researchers calling themselves the Global Deception Research Team asked people in 58 countries how you can tell when someone is lying. The most common answer, top of the list in 51 of those countries, was that liars won’t look you in the eye.

They will, as it happens. Looking away is one of the weakest signs of lying ever measured, if it’s a sign at all. We’ve had three thousand years to get this right, and we’ve mostly swapped one foot for another.

This essay is about why. It isn’t about how clever liars are, though some are. It’s about why the rest of us are so confident about something we’re so bad at.

When I photograph events, I spend much of the evening stalking the room for people deep in conversation. When I find a good pair, I shoot in bursts, several frames across a couple of seconds. Back home at the computer I pick the best frame from each burst, because a single shot is a gamble. Catch someone mid-word and the shape of their mouth makes them look daft. Catch them just before a sneeze and they’re pulling a face they’d never forgive you for. Even a big laugh rarely flatters anyone. Clients want everyone looking their best, so I give myself choices.

Street photography is the opposite. Wandering around London, I look for abstract compositions and unusual interactions, and I do a lot of people-watching. I’ll find somewhere busy and wait. Someone is singing along to their headphones, or lost in a book, or in their own thoughts, or rushing to get somewhere. On the Underground people retreat into themselves altogether. Either way they’re in their own world, with no idea a street photographer is studying them.

Between the two, I’ve looked at an awful lot of faces frozen in a fraction of a second. A single frame can make anyone look daft or shifty. It can’t tell you what they were thinking.

More than a decade ago, I also worked in the criminal courts for the Ministry of Justice. That’s about as close as working life gets to a room built for deciding who’s lying, and I’ll come back to it.

Slightly better than a coin toss

In 2006 Charles Bond and Bella DePaulo pulled together 206 research documents in which 24,483 people tried to judge whether someone was lying. Average accuracy was 54%. A coin would get you 50.

The detail is more interesting than the headline. People correctly spotted 47% of lies, which is worse than guessing. They correctly believed 61% of truths. We’re not really lie detectors. We’re truth assumers who occasionally get lucky. Psychologists call this lean towards belief the truth bias, and it matters enough to get its own section later.

Professionals didn’t escape. Police officers, judges, customs officers and psychiatrists all turn up in this research, and they do only slightly better than everyone else, if at all.

In 2008 Bond and DePaulo asked whether some people are simply gifted. Across 247 samples, the differences between judges were tiny. The best performers did no better than chance would produce. If you’ve met someone who says they can always tell, statistically they’re the same as you, only louder.

Confidence doesn’t help much either. In 1997 DePaulo and colleagues pooled 18 studies and found that the link between how sure people felt and how often they were right was close to zero. A 2019 paper offered a slightly kinder picture: a person’s most confident calls are somewhat more likely to be right than their hesitant ones. But feeling sure doesn’t make anyone a good lie detector, and the people who are most certain are no better than the rest.

Why we believe what we believe

If the eye-contact rule doesn’t work, it’s odd that nearly every culture holds it. The Global Deception Research Team’s explanation has nothing to do with accuracy. They argued the stereotype has a moral root. A liar ought to feel ashamed, and shame looks like a lowered head and averted eyes. So the stereotype describes what a liar should feel, not what a liar does. It works less as a lie detector than as a warning to keep people honest.

I was told this myself, more than once. When I’d been naughty and been caught trying to get away with something, Mum and Dad always told me to look them in the eyes. I can still picture what my body was doing. My shoulders dropped, my eyes went to the floor, and I was as evasive as I could manage. I remember how uncomfortable it was to look Mum or Dad in the eye and try to lie to them.

I wonder whether that’s where the whole idea comes from: generations of parents watching small children squirm. There’s research suggesting it might be. In 2009 Anjanie McCarthy and Kang Lee asked children aged 7 to 15, and adults, to answer some questions truthfully and some untruthfully, and filmed where they looked. The 7- and 9-year-olds broke eye contact much more when they lied. The older children and adults didn’t. Knowledge of the “liars look away” rule grew with age, and the more a child knew about it, the less their eyes gave them away.

Children pick up the belief early, too. In 2008 Shiri Einav and Bruce Hood found that children as young as six were already more likely to say someone was lying if that person looked away.

So parents aren’t imagining it. The cue is real in young children, and then children learn the rule and hide it. Some adults go further. In a 2012 study of 338 airport passengers, Samantha Mann and colleagues found that liars made more deliberate eye contact than truth-tellers. A follow-up study asked liars why. They wanted to convince the interviewer, and to check whether they were being believed.

Even with children, the rule doesn’t make adults good lie detectors. A 2017 review of 45 experiments found adults judged children’s truthfulness correctly 54% of the time, the same as with adults, and caught their lies no better than chance. The rule may have been learned in the nursery, but children learn to beat it long before they leave school.

There’s one more reason the belief survives: it’s almost never tested. You decide a colleague is lying because they looked at the floor, and you rarely find out whether you were right. If you do, it’s weeks later. By then the hunches that came good are memorable and the ones that didn’t have quietly gone. Nothing corrects you, so nothing improves.

What liars actually do

In 2003 DePaulo and her colleagues published the biggest review of liars’ behaviour to date: more than 100 studies and 158 possible cues. Most of those cues had no meaningful link to lying. Eye contact was one of them. So was fidgeting.

A few cues showed small differences. On average, liars said a little less and came across as slightly more tense. Their stories were a touch less convincing. The key word is small. These are differences you find by averaging hundreds of people, not ones you’d spot across a desk in a single meeting. They were a little clearer when people were highly motivated to be believed, or were covering up something they’d done wrong, but even then they stayed modest.

The obvious response is to fix the checklist: stop watching eyes and start watching for tension. In 2011 Maria Hartwig and Charles Bond tested whether that would help. They found people’s judgements already follow the behaviours that genuinely differ between liars and truth-tellers. We aren’t being fooled by the wrong signs. The right signs are just faint, and you can’t read a clue that’s barely there.

Timothy Luke went further in 2019, in a paper called “Lessons from Pinocchio”. Studies that find something tend to get published, while studies that find nothing sit in a drawer. Luke showed that with even a modest amount of that publication bias, the research on behavioural cues looks exactly as it would if there were no useful cues at all. He didn’t prove there are none. He showed we can’t rule it out, which after decades of research is quite an admission.

Why liars should give themselves away, and why they don’t

The theory behind all this is sensible. In 1981 Miron Zuckerman, Bella DePaulo and Robert Rosenthal set out why a liar might leak. Lying can stir up guilt or the fear of being caught. It can be harder work than the truth, because you have to build a story and keep it straight. A liar who knows they’re being watched will also try to control their behaviour, and control can look stiff.

Every one of those can happen to an honest person. Someone telling the truth who fears they won’t be believed feels that fear. Someone recalling a messy, half-remembered week works hard to get it straight. Someone who knows they’re under suspicion stiffens up. The signals don’t separate liars from truth-tellers. They separate people under pressure from people who aren’t.

Most everyday lies don’t create much pressure anyway. When researchers had 632 people log their lies for three months, around nine in ten were white lies: the “fine, thanks” and the “lovely haircut”. Nobody’s pulse races over telling a colleague their presentation went well. The bigger lies tend to come from a small group of people, and I’ll look at them in the fourth essay.

The face

If one idea keeps the lie-spotting business going, it’s the microexpression.

Researchers first described fleeting facial expressions in 1966, and in 1969 Paul Ekman and Wallace Friesen linked them to deception. The theory is that a concealed emotion can flash across the face too fast for the liar to stop it, in as little as a twenty-fifth of a second. Ekman built an enormously influential career on the face, including a system for scoring every muscle movement in it. The idea even got its own television drama, Lie to Me, with Ekman as scientific adviser.

In 2008 Stephen Porter and Leanne ten Brinke went looking for microexpressions frame by frame. They showed people disturbing and pleasant images, and asked them sometimes to react honestly and sometimes to fake or hide what they felt. Then they examined 104,550 frames across 697 expressions. They found 14 microexpressions. None covered the whole face, and some turned up when people weren’t faking anything.

To be fair to the face, it isn’t a perfect poker player. Hidden emotions produced more inconsistent expressions and more blinking, and every participant leaked something at least once. But untrained observers watching the footage did only slightly better than chance. Something leaks. It leaks rarely and ambiguously, and not in a form anyone can reliably read in real time.

There’s a deeper problem. Reading microexpressions assumes that a face reliably shows what a person feels. In 2019 Lisa Feldman Barrett and four colleagues reviewed more than 1,000 studies and found that assumption much weaker than most people think. People scowl when angry only about a quarter of the time, and they also scowl when concentrating or when they have stomach ache. Other researchers argue that faces carry more meaning than that, and the debate is far from settled. But if a scowl doesn’t reliably mean anger, a scowl lasting a twenty-fifth of a second certainly doesn’t mean a lie.

Ekman and his colleague Maureen O’Sullivan also claimed to have found a rare group of “truth wizards”, people who could spot liars with 80% accuracy or better. Charles Bond and Ahmet Uysal re-ran the numbers in 2007 and showed that chance alone could explain the results. Test enough thousands of people and a few will score brilliantly by luck, the same way somebody always wins the office sweepstake.

A $900 million experiment

None of this stopped the idea being sold, and nowhere was it sold harder than at American airports.

From 2007 the US Transportation Security Administration ran a programme called SPOT, Screening of Passengers by Observation Techniques. Officers watched travellers for behaviours meant to signal stress, fear or deception, and the TSA said Ekman had helped develop it. By 2013 it had cost around $900 million and employed about 3,000 officers at 176 airports.

That year the US Government Accountability Office reviewed the programme and found no scientifically validated evidence that the behaviours it relied on could identify threats. The TSA disagreed and carried on, renaming the programme in 2016.

In 2020 more than 40 researchers signed a paper warning about pseudoscience in security and justice. They singled out a commercial industry happy to sell lie-detection training to anyone with a budget.

It depends who you’re looking at

When Bond and DePaulo looked at what actually drives a lie-detection judgement, the biggest factor wasn’t the skill of the judge. It was the person being judged. Some people look honest whether they’re lying or not. Others look shifty while telling the plain truth. Tim Levine, who has spent decades studying deception, found the same in experiments with his colleagues: how believable someone seems has little to do with whether they’re actually honest.

Ekman saw this coming. In his 1985 book Telling Lies he named it the Brokaw hazard, after the American news anchor Tom Brokaw, who said he watched interviewees for evasive, roundabout answers. Ekman’s point was that any behaviour that sometimes goes with lying will be completely normal for someone. Meet that person for the first time and you’ll misread them.

That someone is often a person who is already different from the people judging them. In a study published in 2022, researchers at Flinders University in Australia filmed 30 autistic and 29 non-autistic adults in interviews, then asked 1,410 people to watch and judge them. The autistic adults were rated as more deceptive and less credible, even when they were telling the truth. Language and culture have the same effect, as the courts have recognised.

Being dyslexic and having ADHD can cause exactly this kind of problem. I worked in IT for years and know Windows administration well. At a job interview in London, I was asked whether I was familiar with Microsoft Intune, Microsoft’s device management service. I’d used it, so I said yes. Then they asked me to talk about it, and my mind went blank. I stumbled over my words, and it gave the interviewers the impression I was trying to blag it.

I wasn’t. I just looked exactly like someone who was. On the way home the answer came to me, and I was so annoyed with myself. Perhaps it was the pressure of the interview combined with the way my brain works. Either way, all the interviewers saw was the stumble.

Ekman’s second warning was the Othello error. In Shakespeare’s play, Othello watches Desdemona’s distress and takes it as proof that she’s been unfaithful. She’s distraught because she can’t make him believe her. The fear of not being believed looks almost exactly like the fear of being caught, and the person watching usually sees whichever one they were expecting.

What the courts worked out

Courtrooms are the one place where deciding who’s telling the truth is the actual job. English judges reached the psychologists’ conclusion by their own route, some of them before the research caught up.

In a 1973 article the judge Sir Brian MacKenna asked whether a witness who looks you straight in the face is more likely to be truthful than one who looks at the floor out of shyness. His answer: “For my part I rely on these considerations as little as I can help.” Lord Bingham quoted him in a 1985 paper on how judges decide facts. Of the ways a judge can test a witness’s account, Bingham put demeanour last.

In 2018 the Court of Appeal put it more bluntly. In a case called SS (Sri Lanka), Lord Justice Leggatt said it had increasingly been recognised that judging truthfulness from a witness’s demeanour is “usually unreliable and often dangerous”. He added that the danger grows when the witness is of a different nationality from the judge, speaks English as a foreign language or gives evidence through an interpreter. The only reliable approach, he said, is to look at what a witness actually says and whether it fits the other evidence and the known facts.

I saw that danger for myself. A courtroom can be a frightening place for a witness, for all sorts of reasons. Giving evidence against someone accused of violence, with them watching you from the dock, is daunting. I saw many witnesses look worried while giving evidence that was completely solid.

I also saw the reverse: witnesses who gave a confident performance, only for evidence such as CCTV to contradict them. Anyone judging by demeanour alone would have got both the wrong way round.

Training can make it worse

If ordinary people are bad at this, training seems the obvious fix. The research here is uncomfortable.

In 1999 Saul Kassin and Christina Fong trained students in the Reid technique, a widely used American police interrogation method that includes reading body language. The students then watched videos of people denying a mock crime, some guilty and some innocent. The trained students were less accurate than untrained ones. They were also more confident, and gave more reasons for their answers.

In a later study, Kassin and colleagues asked students and police investigators to judge recorded confessions from prison inmates. Half were true and half had been made up for the study. The students were more accurate. The investigators were more confident, and more inclined to decide people were guilty.

Put those findings together and you get the combination to worry about at work: someone fresh from a half-day course on body language, certain and wrong.

How lies really get found out

In 2002 Hee Sun Park, Tim Levine and colleagues tried something refreshingly practical. Instead of testing people in a lab, they asked them to remember a time they had actually caught someone lying, and how they did it.

Very few had spotted it at the time from the liar’s face or manner. Lies came out later, often much later. Someone else told them. They found physical evidence. The facts didn’t match what they already knew. Sometimes the liar simply owned up. Later studies found the same pattern, and a review of four of them found that more than four in five of the clues people reported had nothing to do with behaviour.

There’s a fair counterpoint. A 2018 study found that when people were asked what first made them suspicious, rather than what proved the lie, behaviour came up much more often. Other researchers have since questioned how that study’s data was analysed. Still, the distinction is useful, and it’s about the most that can be said for watching faces. An odd moment might make you look harder. It’s the looking harder that finds the lie.

Why being bad at this is mostly fine

Tim Levine argues that our default belief isn’t a flaw. His truth-default theory says we accept what people tell us unless something forces us to doubt it. Most of the time that’s the right call, because most people are honest most of the time.

The lab numbers hide this. In most experiments, half the statements are lies, because that’s the tidiest way to measure accuracy. Real life isn’t like that. Levine and his colleague Hee Sun Park showed that accuracy depends on how common lies are. When lies are rare, the truth bias that makes us look hopeless in the lab makes us right most of the time.

Here’s what that means at work. Suppose nine in ten things your colleagues tell you in a week are true. If you simply believed everything, you’d be right nine times out of ten. You’d also catch no liars at all, and that’s the price. Someone who treated every statement with suspicion would catch more lies, wrongly doubt a lot of honest people, and be exhausting to share an office with.

The cost of believing by default is that we’re open to the occasional skilled deceiver. Levine thinks that trade is worth making, and so do I. Doubting everyone would mean nothing ever got done.

Suspicion has costs of its own. A workplace where everyone watches everyone else for leaks doesn’t catch more liars. It punishes the nervous and the socially awkward, and teaches everyone that looking relaxed matters more than being honest. I’ll come back to what that costs in the seventh essay.

What to do instead

Stop watching faces. Listen to what people say, then check it against what you can verify. I saw this for myself in the criminal courts, and I’ll come to that in the second essay.

That poisoner from three thousand years ago was never going to be caught by his toe. He was going to be caught by whoever sold him the poison.

Sources and further reading

Barrett, L. F., Adolphs, R., Marsella, S., Martinez, A. M. & Pollak, S. D. (2019). Emotional expressions reconsidered: challenges to inferring emotion from human facial movements. Psychological Science in the Public Interest, 20, 1–68.

Bingham, T. (1985). The judge as juror: the judicial determination of factual issues. Current Legal Problems, 38, 1.

Bond, C. F. & DePaulo, B. M. (2006). Accuracy of deception judgments. Personality and Social Psychology Review, 10, 214–234.

Bond, C. F. & DePaulo, B. M. (2008). Individual differences in judging deception: accuracy and bias. Psychological Bulletin, 134, 477–492.

Bond, C. F. & Uysal, A. (2007). On lie detection “wizards”. Law and Human Behavior, 31, 109–115.

Denault, V. et al. (2020). The analysis of nonverbal communication: the dangers of pseudoscience in security and justice contexts. Anuario de Psicología Jurídica, 30, 1–12.

DePaulo, B. M., Charlton, K., Cooper, H., Lindsay, J. J. & Muhlenbruck, L. (1997). The accuracy–confidence correlation in the detection of deception. Personality and Social Psychology Review, 1, 346–357.

DePaulo, B. M., Lindsay, J. J., Malone, B. E., Muhlenbruck, L., Charlton, K. & Cooper, H. (2003). Cues to deception. Psychological Bulletin, 129, 74–118.

Einav, S. & Hood, B. M. (2008). Tell-tale eyes: children’s attribution of gaze aversion as a lying cue. Developmental Psychology, 44, 1655–1667.

Ekman, P. (1985). Telling Lies: Clues to Deceit in the Marketplace, Politics, and Marriage. New York: Norton.

Ekman, P. & Friesen, W. V. (1969). Nonverbal leakage and clues to deception. Psychiatry, 32, 88–106.

Global Deception Research Team (2006). A world of lies. Journal of Cross-Cultural Psychology, 37, 60–74.

Gongola, J., Scurich, N. & Quas, J. A. (2017). Detecting deception in children: a meta-analysis. Law and Human Behavior, 41, 44–54.

Haggard, E. A. & Isaacs, K. S. (1966). Micromomentary facial expressions as indicators of ego mechanisms in psychotherapy. In L. A. Gottschalk & A. H. Auerbach (eds), Methods of Research in Psychotherapy, 154–165. New York: Appleton-Century-Crofts.

Hartwig, M. & Bond, C. F. (2011). Why do lie-catchers fail? A lens model meta-analysis of human lie judgments. Psychological Bulletin, 137, 643–659.

Kassin, S. M. & Fong, C. T. (1999). “I’m innocent!”: effects of training on judgments of truth and deception in the interrogation room. Law and Human Behavior, 23, 499–516.

Kassin, S. M., Meissner, C. A. & Norwick, R. J. (2005). “I’d know a false confession if I saw one”: a comparative study of college students and police investigators. Law and Human Behavior, 29, 211–227.

Levine, T. R. (2020). Duped: Truth-Default Theory and the Social Science of Lying and Deception. Tuscaloosa: University of Alabama Press.

Levine, T. R., Park, H. S. & McCornack, S. A. (1999). Accuracy in detecting truths and lies: documenting the “veracity effect”. Communication Monographs, 66, 125–144.

Lim, A., Young, R. L. & Brewer, N. (2022). Autistic adults may be erroneously perceived as deceptive and lacking credibility. Journal of Autism and Developmental Disorders, 52, 490–507.

Luke, T. J. (2019). Lessons from Pinocchio: cues to deception may be highly exaggerated. Perspectives on Psychological Science, 14, 646–671.

MacKenna, B. (1973). Discretion. Irish Jurist (new series), 9, 1.

Mann, S., Vrij, A., Leal, S., Granhag, P. A., Warmelink, L. & Forrester, D. (2012). Windows to the soul? Deliberate eye contact as a cue to deceit. Journal of Nonverbal Behavior, 36, 205–215.

Mann, S., Ewens, S., Shaw, D., Vrij, A., Leal, S. & Hillman, J. (2013). Lying eyes: why liars seek deliberate eye contact. Psychiatry, Psychology and Law, 20, 452–461.

Masip, J. & Sánchez, N. (2019). How people really suspect lies: a re-examination of Novotny et al.’s (2018) data. Journal of Nonverbal Behavior, 43, 481–492.

McCarthy, A. & Lee, K. (2009). Children’s knowledge of deceptive gaze cues and its relation to their actual lying behavior. Journal of Experimental Child Psychology, 103, 117–134.

Novotny, E., Carr, Z., Frank, M. G., Dietrich, S. B., Shaddock, T., Cardwell, M. & Decker, A. (2018). How people really suspect and discover lies. Journal of Nonverbal Behavior, 42, 41–52.

Park, H. S., Levine, T. R., McCornack, S. A., Morrison, K. & Ferrara, S. (2002). How people really detect lies. Communication Monographs, 69, 144–157.

Porter, S. & ten Brinke, L. (2008). Reading between the lies: identifying concealed and falsified emotions in universal facial expressions. Psychological Science, 19, 508–514.

R (SS (Sri Lanka)) v Secretary of State for the Home Department [2018] EWCA Civ 1391.

Serota, K. B., Levine, T. R. & Docan-Morgan, T. (2022). Unpacking variation in lie prevalence: prolific liars, bad lie days, or both? Communication Monographs, 89, 307–331.

Smith, A. M. & Leach, A.-M. (2019). Confidence can be used to discriminate between accurate and inaccurate lie decisions. Perspectives on Psychological Science, 14, 1062–1071.

Trovillo, P. V. (1939). A history of lie detection. Journal of Criminal Law and Criminology, 29, 848–881.

US Government Accountability Office (2013). Aviation Security: TSA Should Limit Future Funding for Behavior Detection Activities (GAO-14-159).

Zuckerman, M., DePaulo, B. M. & Rosenthal, R. (1981). Verbal and nonverbal communication of deception. Advances in Experimental Social Psychology, 14, 1–59.