People Are Pretty Bad at Reading Faces

Why humans are quick to judge expressions—and often get them wrong

Marco Goran Romano

The truth was written all over her face. The eyes are the window to the soul. From our clichés, you would think that we could read faces like they were … well, open books. In fact, the skill has more in common with dancing, or writing confessional poetry: People tend to overestimate their ability to do it.

Most of us can’t distinguish between certain expressions without contextual clues. In one study, participants were unable to tell whether faces in photos were showing pain or sexual pleasure about a quarter of the time [1]. In another, when people watched silent videos of the same person experiencing pain and faking pain, they couldn’t tell which was which. A computer was correct 85 percent of the time [2]. Computers were also better at telling that a person was smiling out of mild frustration rather than genuine delight [3].

And yet, as bad as we are at reading expressions, we jump to all kinds of conclusions based on people’s faces. We might scoff at the ancient Greek belief in physiognomy—assessing character on the basis of facial features—but we unwittingly practice it daily. Recent research shows that while there’s practically no evidence that faces reveal character, we nonetheless behave as if certain features signal certain traits [4]. People with stereotypically “feminine” facial features seem more trustworthy; those with lower eyebrows appear more dominant [5]. In another study, people were ready to decide whether an unfamiliar face should be trusted after looking at it for just 200 milliseconds. Even when given a chance to look longer, they rarely changed their mind [6].

Such judgments can defy logic. Subjects playing a trust game invested more money with a player who had a trustworthy face than with one who didn’t—even when the two players had the same reputation [7]. Another study reported that jurors needed less evidence to convict a person with an untrustworthy face [8]. And a researcher focusing on the Israeli-Palestinian conflict found that a Palestinian peace offering was more likely to be accepted by Jewish Israeli respondents if it was attributed to a politician with “babyfacedness” (big eyes, plump lips) [9].

Which brings us to a contradiction. A person’s face may not reflect her nature, and yet research finds that specific facial features do seem to influence futures [10]. U.S. Army War College graduates with dominant-looking faces are more likely than their peers to become generals [11]; people whose faces appear competent are more likely to become CEOs of successful companies [12]. This makes a certain sense. If everyone assumes strong-chinned Stanley is an assertive person, he’s more likely to become one. Perhaps by treating others as though their face reveals their character, we prompt them to become the people we assumed them to be.

The studies:

[1] Hughes and Nicholson, “Sex Differences in the Assessment of Pain Versus Sexual Pleasure Facial Expressions” (Journal of Social, Evolutionary, and Cultural Psychology, Dec. 2008)

[2] Bartlett et al., “Automatic Decoding of Facial Movements Reveals Deceptive Pain Expressions” (Current Biology, March 2014)

[3] Hoque et al., “Exploring Temporal Patterns in Classifying Frustrated and Delighted Smiles” (IEEE Transactions on Affective Computing, July–Sept. 2012)

[4] Todorov et al., “Social Attributions From Faces” (Annual Review of Psychology, 2015)

[5] Oosterhof and Todorov, “The Functional Basis of Face Evaluation” (Proceedings of the National Academy of Sciences, Aug. 2008)

[6] Todorov et al., “Evaluating Faces on Trustworthiness After Minimal Time Exposure” (Social Cognition, Dec. 2009)

[7] Rezlescu et al., “Unfakeable Facial Configurations Affect Strategic Choices in Trust Games With or Without Information About Past Behavior” (plos One, March 2012)

[8] Porter et al., “Dangerous Decisions” (Psychology, Crime & Law, May 2010)

[9] Maoz, “The Face of the Enemy” (Political Communication, July 2012)

[10] Olivola et al., “Social Attributions From Faces Bias Human Choices” (Trends in Cognitive Sciences, Nov. 2014)

[11] Mazur and Mueller, “Channel Modeling: From West Point Cadet to General” (Public Administration Review, March–April 1996)

[12] Graham et al., “A Corporate Beauty Contest” (Management Science, forthcoming)