Shifty eyes and fidgeting: what research says about spotting a liar
Shifty eyes, fidgety hands and a shaky voice are the classic signs of a liar, and researchers who tested them found only weak links to deception. A 2011 piece in the FBI Law Enforcement Bulletin by David Matsumoto, Hyi Sung Hwang, Lisa Skinner and Mark Frank opens with a detective questioning a suspect. As the suspect prepares to answer, he looks up and to the left, purses his lips, tenses his eyelids and brings his eyebrows down. The detective reads shifty eyes and gaze aversion as signs of lying and prepares to press harder. The authors say this investigator likely would be mistaken.
Twenty-three of 24 peer-reviewed studies in scientific journals that ran experiments on eye behavior as an indicator of lying rejected the idea, and no scientific evidence suggests that eye behavior or gaze aversion can gauge truthfulness reliably. They add that gaze aversion, fidgety feet or hands, voice stress analysis and body posture have all been tested and found only weakly associated with deception. A person with shifty eyes may simply be nervous, and a calm person may be lying.
A mistaken read carries a price: a witness, suspect or informant is judged to be lying when they are not, or truthful when there is more to the story. That breeds misplaced confidence about a case and sends investigators down dead ends, and a false read can have deadly consequences.
The shifty eyes stereotype is nearly universal. The Global Deception Research Team, with 90 members, published "A World of Lies" in the Journal of Cross-Cultural Psychology in 2006. Its first study covered 58 countries, with 20 men and 20 women who had lived there all their lives in each, 2,320 people in all. They answered the open question "How can you tell when people are lying?" and gave 11,157 responses, which coders sorted into 103 belief categories. Gaze aversion was the most common answer, given by 63.66 percent of participants. Nervousness came next at 28.15 percent.
A second study used a 10-item questionnaire in 63 countries with 40 respondents each. There, 71.5 percent endorsed gaze aversion, 65.2 percent posture shifts and 64.8 percent self-touching. In 61 of the 63 countries, respondents were likelier to say liars decrease eye contact than increase it. The authors read these stereotypes as social control, a way of promoting norms against lying, more than as accurate descriptions of behavior. Experimental research does not back them, and shifty eyes are no exception: a 2003 meta-analysis by DePaulo and colleagues found that stereotypic behaviors bear negligible relationships to deception, and gaze aversion occurs nearly as often in truth-tellers as in liars.
Ordinary judges do not do much better than chance. Bond and DePaulo's meta-analysis found that Western perceivers average roughly 54 percent accuracy at detecting lies from behavior, where guessing would give 50. A 2018 PLOS ONE study revisiting that meta-analysis reports that across 247 samples, 12,704 judges and 6,060 senders, the reliability of judge ability was near zero, with an alpha of .13, so the person lying or telling the truth shaped results more than the person judging.
Matsumoto and his colleagues argue that many laboratory studies miss real conditions. Participants often have no personal, financial or emotional investment in the lie and no fear of being caught, so nothing is at stake. Their own studies recruit people motivated to act against a person or group with a different ideology. Each participant chooses whether to commit a crime, such as stealing a check made out to the group they despise, and is then interviewed by a retired law enforcement officer. Anyone who is not believed faces detention, blasts of white noise, or having the stolen check donated to the hated group. Truthful people are often nervous in police interrogations because consequences follow judgments, not reality.
Sensors in these studies record facial behavior, gestures, voice, heart rate, pupil dilation and gaze direction. When motivated people lie and face consequences, clues leak through several channels: facial expressions, gestures and body language, voice, verbal style and the words themselves. The authors say it is not the mere presence or absence of gaze aversion or fidgeting that signals lying. What counts is how the cues change over time from a person's baseline and how they combine with the person's words. No single indicator exists, they say, or research would have found it by now.
Their latest study examined videos of members of ideologically motivated groups in two scenarios. In one, participants chose whether to steal $50 in cash from a briefcase and were interviewed about it. In the other, they chose to lie or tell the truth about their political beliefs. Anyone judged to be lying lost the participation fee and faced an hour of white noise. They chose 10 people from each scenario, half truthful and half lying. Liars showed significantly more nonverbal behavior that clashed with their words, such as denying a theft while looking afraid. Statement analysis also separated the groups: minimizing and editing adverbs and changes in nouns and verbs went with lying, while equivocation and spatial details went with truth telling. Inconsistent facial expressions combined with statement analysis correctly classified 90 percent of the participants.
The bulletin stresses limits on its tools. Microexpressions, flashes of concealed emotion that can last as little as one-fifteenth of a second, and statement analysis should never be treated as proof of lying by themselves. They help identify areas that need further probing, and the behaviors can have other causes, such as embarrassment or fear of retaliation.
- FBI Law Enforcement Bulletin, Evaluating Truthfulness and Detecting Deception (June 2011) — leb.fbi.gov
- Global Deception Research Team, A World of Lies (Journal of Cross-Cultural Psychology, 2006) — pmc.ncbi.nlm.nih.gov
- Law et al., It's the deceiver, not the receiver (PLOS ONE, 2018) — pmc.ncbi.nlm.nih.gov