It takes just 100 milliseconds to judge a stranger’s face. Then why Princeton psychologists warn against trusting it too much

A stranger’s face can trigger an impression almost instantly. Research by Princeton psychologists Janine Willis and Alexander Todorov found that people could form judgments about whether an unfamiliar face looked trustworthy, competent, likeable, attractive or aggressive after seeing it for just 100 milliseconds, which is one-tenth of a second.

What makes the finding more striking is that giving people more time to look at the same face did not significantly change how closely their judgments matched those made by people who had unlimited viewing time. However, longer exposure did increase their confidence.

The distinction matters. The study showed how quickly an impression can form, not that people can accurately read another person’s character from their face.

What happened in the Princeton experiment

Willis and Todorov’s 2006 study, published in Psychological Science, involved 245 Princeton undergraduate students across five experiments.

Participants were shown 66 frontal photographs of unfamiliar amateur actors. The faces were presented for either 100 milliseconds, 500 milliseconds or one second. Different experiments focused on one of five traits: attractiveness, likeability, competence, trustworthiness and aggressiveness.

The researchers first collected ratings from a separate group that could examine the photographs without a time limit. Those ratings were then used as a comparison for the rapid judgments.

The results showed that even the shortest exposure produced a strong relationship with the unrestricted ratings. For example, the correlation for trustworthiness was 0.73, while it was 0.69 for attractiveness, 0.59 for likeability, and 0.52 for both competence and aggressiveness.

Giving participants half a second or a full second did not significantly strengthen those correlations.

More time increased confidence, not necessarily accuracy

The study found that longer exposure was not completely irrelevant. When viewing time increased from 100 to 500 milliseconds, participants' judgments became somewhat more negative, their responses became faster and their confidence increased.

From 500 milliseconds to one second, most judgments changed little, while confidence increased for several traits.

This means that spending longer looking at a face could make people feel more certain about an impression that had already formed. It did not establish that the impression itself was a more accurate assessment of the person.

The researchers also found that longer exposure allowed impressions to become more differentiated. At very short exposures, several judgments appeared to be driven by a broad positive-versus-negative impression. With more time, distinctions between traits became clearer.

The biggest warning: agreement is not accuracy

One important limitation is often lost when the study is summarised as proof that people can judge character in a fraction of a second.

The 0.73 correlation for trustworthiness, for example, does not mean participants were 73% accurate at identifying trustworthy people.

It measured how closely their ratings matched those of another group looking at the same photographs without a time limit. Those unrestricted ratings themselves were based on facial appearance, rather than behavioural records or evidence about the photographed people's actual personalities.

Alexander Todorov has subsequently described facial first impressions as powerful but faulty. In a 2017 interview published by Princeton, he said people are often overconfident in such impressions and that they are generally inaccurate as stable predictors of how someone will behave across different situations.

Later research pushed the timing even lower

A 2009 study by Todorov, Manish Pakrashi and Nikolaas Oosterhof examined trustworthiness judgments using shorter exposure times and improved masking procedures.

The researchers found that agreement with unrestricted judgments was already above chance after 33 milliseconds. It improved with additional exposure but stopped improving after about 167 milliseconds.

The result suggests that appearance-based impressions can begin forming extremely quickly. But it still does not establish a precise biological deadline at which people can reliably determine another person's character.

Why a quick impression can become a problem

Todorov's later work highlights the broader concern, people may naturally form impressions from faces, but those impressions can influence real-world decisions even when they are not reliable measures of character.

Faces are especially powerful social cues, and people encounter them constantly. But an ordinary interaction contains far more information than a static photograph including behaviour, speech, circumstances and a person's actions over time.

That is why the Princeton research is better understood as a study of how quickly the mind forms an impression, rather than a demonstration that faces reveal who people really are.