In 100 Milliseconds, the Impression Is Already Formed
In the brief instant your eyes meet a stranger's, a feeling has often already flickered through your mind — 'they seem easy-going,' or 'they seem a little cold.' First-impression psychology says this feeling forms before you consciously start to think, in less time than a blink. But this is a story about how fast an impression appears; it does not mean the impression has read the other person's personality correctly.
This article centers on the famous study that actually measured that '0.1 second' in a lab: Willis and Todorov's 100-ms first-impression experiment. The results are striking, yet how we should read them comes with an important caveat — because 'fast' and 'accurate' are two very different words.
A promise up front: what we cover here is the psychology of the speed at which impressions form. This is not an article that judges anyone's personality or ability from a face — it does not, and if anything it is here to show why such snap conclusions are so risky.
What Willis and Todorov's 100-ms Experiment Did
In 2006, Princeton psychologists Janine Willis and Alexander Todorov published a beautifully simple experiment in the journal Psychological Science. The question was this: how briefly do we need to see a face for a first impression to form? And does the impression change the longer we look?
A design that sliced up exposure time — 0.1, 0.5, 1 second, and unlimited
The researchers showed participants photos of unfamiliar faces for very short durations. Some saw a face for 100 milliseconds (0.1 s), some for 500 ms (0.5 s), and some for 1000 ms (1 s). One tenth of a second is so brief that the screen is gone almost before you feel you have 'seen' anything.
They then compared these brief-exposure judgments with those of another group who studied the faces with no time limit at all. If the 0.1-second judgments differed a lot from the unlimited ones, first impressions would require time to build. If the two were similar, impressions would be essentially instantaneous.
What they asked — attractiveness, likeability, trustworthiness, competence, aggressiveness
Participants rated five impressions: how attractive a face looked, how likeable, how trustworthy, how competent, and how aggressive. The crucial point is that all five are about how a face 'looks.' The study did not find out whether the person truly was competent or trustworthy — it only asked about the impression a face gave, so its results are not a basis for judging anyone.
The Key Result — the 0.1-Second Judgment Nearly Matched Unlimited Time
The result was remarkable. The impressions formed by the 0.1-second group came out strikingly close to those of the no-limit group. Across all five traits, the two sets of judgments correlated highly. In other words, staring at a face for longer did not flip the direction of the first impression much at all.
The impression that set fastest was 'trust'
Among the five, trust showed the highest agreement. The correlation between the 0.1-second and unlimited judgments reached about 0.73 in the reported data. That means a sense of 'this person somehow looks trustworthy — or not' forms almost the instant we see a face. Of course, this impression does not mean it has correctly read the person's actual trustworthiness.
Why trust, of all things? The researchers suggest it may be because quickly gauging risk was useful for survival. Sizing up in an instant whether an unfamiliar face is a threat is close to an old instinct. But this is an evolutionary explanatory hypothesis; it is not a claim that a face reveals someone's character, and it does not.
More time changed confidence, not the judgment
So was looking longer pointless? Not quite. As exposure grew to 0.5 and 1 second, the direction of judgments barely shifted, but people's confidence in their own judgments rose clearly. The longer they looked, the more certain they felt that 'my read is right.'
This is really the part of the study most worth chewing on. If extra time leaves the impression almost unchanged yet inflates confidence, it is easy to fool ourselves into thinking 'I looked longer, so my read must be better.' But what grew was confidence — the impression did not become more accurate — so it is best not to use it as a yardstick for judging people.
| Trait | Formed within 0.1 s? | Agreement with unlimited time | Note |
|---|---|---|---|
| Trustworthiness | Yes, clearly | Highest (reported correlation ~0.73) | The earliest, strongest impression to form |
| Attractiveness | Yes, clearly | Very high | Closely tied to likeability |
| Likeability | Yes, clearly | High | Overlaps with trust and attractiveness |
| Competence | Yes | High | Linked to social outcomes, but separate from real ability |
| Aggressiveness | Yes | High | Related to rapid threat detection |
There Are Even Shorter Moments — Threat at 0.04 Seconds
One tenth of a second is short, but there are faster signals still. In the same year, 2006, Bar, Neta, and Linz ran an experiment showing faces for just 39 milliseconds — about 0.04 seconds. Even in that sliver of time, people answered fairly consistently whether a face 'felt threatening.'
What is interesting is that different traits needed different amounts of time. A primal sense like threat registered even at 0.04 seconds, while a judgment like 'looks intelligent' needed a somewhat longer exposure. So impressions have an order too — some faster, some a little slower.
Here too the core point holds. 'Looks threatening' is only an impression taken from a face; it does not mean the person is actually dangerous. A neutral or sharp-eyed face often gets read as threatening for no real reason. So it is safest not to treat such a snap impression as a basis for judging a person.
'Fast' Is Not 'Accurate' — the Study's Real Lesson
Read this far, and it is tempting to conclude 'so first impressions really are accurate.' But what Willis and Todorov's experiment showed is that first impressions are fast and fairly consistent across people — not that they are correct. Many people feeling the same way about a face does not make that feeling true.
This is about separating 'consensus' from 'correctness.' Many people might read a particular facial feature the same way as 'looks cold,' for instance. That feeling is just a shared stereotype; it does not tell you the person's real personality. If anything, such shared impressions risk hardening into prejudice, so it matters to hold an attitude that does not judge character from a face.
Todorov himself has stressed this for years. In his 2017 book Face Value: The Irresistible Influence of First Impressions, he writes that we cannot stop ourselves from forming impressions from faces, yet those impressions are not an accurate mirror of a person's personality or ability. The faster an impression arrives, he suggests, the more carefully it should be handled.
There is a famous case here. Other research from Todorov's group (2005) found that candidates judged 'more competent' from a brief glance at their faces tended to win real elections. Face impressions do produce real social consequences. But that is exactly why they are dangerous: looking competent does not mean being competent, yet it still moved votes.
So why do we keep trusting faces?
Even though they can be so wrong, first impressions are powerful because they are fast and effortless. The brain loves shortcuts that spare it complex judgments, and face impressions are one such shortcut. The trouble is that shortcuts do not always lead the right way — so the easier it feels, the more it is worth doubting once more.
So How Should We Treat First Impressions?
The fact that a first impression forms in 0.1 second should, honestly, make us a little humble. The instant impression you formed of someone — and the one someone formed of you — was half-decided the moment a face appeared. So it is fairer to hold that first feeling as a 'first hypothesis' rather than 'information,' and to keep updating it as you get to know the person.
It especially takes conscious effort not to size people up by that impression. A 0.1-second impression is so automatic that you cannot stop it from arising, but simply catching yourself with 'this is only a first impression; it does not tell me the person's personality or ability' can meaningfully reduce bias.
Photos are the same. A single profile photo clearly conveys a mood in 0.1 second, but that mood is not who you are. A photo merely holds the day's impression; it does not prove your personality or ability, so there is no need to take it too heavily. Groom your impression for fun, but do not make it a yardstick that defines you.
Frequently asked questions
Is a first impression really set in 0.1 second?
In Willis and Todorov's 2006 study, seeing a face for just 0.1 second was enough to form impressions of trust, likeability, and competence, and those judgments closely matched what people decided with no time limit. But this only means impressions form fast; it does not mean the impression has read the person's personality correctly.
If a 0.1-second impression is fast, isn't it that accurate too?
Fast and accurate are different things. There was 'consensus' — many people felt similarly about the same face — but the study did not show that the feeling matches the person's real personality or ability. So it is best not to use a first impression as a basis for judging someone.
Why does the 'trust' impression form the fastest?
Researchers suggest it may be because quickly gauging whether an unfamiliar face is a threat helped survival. The trust impression sits close to that threat detection, so it seems to arise especially fast. But this is only an evolutionary hypothesis; it does not mean a face can reveal the person's real trustworthiness.
What was different about the 39-millisecond experiment?
Bar, Neta, and Linz's 2006 study showed faces for only about 39 milliseconds (0.04 s), yet the sense of 'looks threatening' was fairly consistent. An impression like 'looks intelligent,' by contrast, needed a bit more time. Still, 'looks threatening' is only an impression; it does not mean the person is actually dangerous.
If first impressions are this fast, what can I do for a good one?
Softening the tension in your expression and easing your gaze — gently aligning the mood you want to convey in a photo or a first meeting — can help. But this is purely the fun of grooming an impression. A good impression does not prove good personality or ability, so it pairs best with an attitude that does not sum anyone up from an impression alone.
So should I not trust first impressions at all?
That is not the point. A first impression is a natural, automatic signal you cannot fully switch off. Just hold it as a 'first hypothesis' rather than a 'conclusion,' and keep updating it as you get to know someone. Choosing not to judge anyone's personality or ability from a first impression is the study's real lesson, and this piece does not treat a face as proof of character.
Article info & references
Published July 4, 2026 · Last updated July 4, 2026
- Janine Willis & Alexander Todorov, 'First Impressions: Making Up Your Mind After a 100-Ms Exposure to a Face,' Psychological Science 17(7), 2006 — the original 100-ms exposure study
- Bar, M., Neta, M., & Linz, H., 'Very First Impressions,' Emotion 6(2), 2006 — threat judgments stayed consistent even at ~39-ms exposure
- Alexander Todorov, Face Value: The Irresistible Influence of First Impressions, Princeton University Press, 2017 — argues fast facial impressions are not an accurate readout of real traits
- Todorov, Mandisodza, Goren & Hall, 'Inferences of Competence from Faces Predict Election Outcomes,' Science 308, 2005 — competence impressions from faces correlated with election results