First impressions: how fast they form, how accurate they are and the 7-38-55 rule

“A first impression is made in a few seconds”, “93% of what you say comes from your face and voice, not your words”: you have probably heard lines like these. Worrying about first impressions before a blind date, a job interview or the first day of term is natural too. So do first impressions really form in seconds, how accurate are they, and can they change once they are made? This guide answers those questions with research.

The short answer is that first impressions really do form fast. After seeing an unfamiliar face for a tenth of a second, people settle on impressions such as attractiveness or trustworthiness, and those judgments correlated highly with judgments made by people who looked without a time limit (Willis & Todorov, 2006). Fast does not mean accurate, though. From a brief look, people judged traits that show readily, such as extraversion, better than chance, but traits that show less, such as agreeableness, needed a longer look (Borkenau & Liebler, 1992; Carney et al., 2007). Impressions from facial appearance are shared from one person to the next, yet the claim that they reveal what someone is really like has been judged greatly overstated (Todorov et al., 2015). What we hear first does colour how we read what comes after, but that effect can be small (Sullivan, 2019), and new information that makes the earlier facts look different has reversed impressions (Mann & Ferguson, 2015). And the “words are only 7%” rule attributed to Mehrabian comes from narrow experiments on conveying liking and disliking, so it does not apply to conversation in general (Lapakko, 2007).

This guide covers how many seconds a first impression takes, how accurate first impressions are and which traits show, the traps of judging by the face alone, how long impressions last and when they change, what the 7-38-55 rule actually covers, and how to handle first impressions realistically. The research on reading personality from facial features is covered in more detail in the guide to the science of face reading. The findings here are averages across many people, so they cannot settle whether one particular first impression of one particular person is right or wrong.

The most quoted figure is a tenth of a second. In five experiments published in 2006, researchers varied how long they showed photos of unfamiliar faces and asked for judgments of attractiveness, likeability, trustworthiness, competence and aggressiveness. Judgments made after 0.1 seconds (100 milliseconds) correlated highly with judgments made by people who looked without a time limit, and longer exposure did not significantly raise those correlations. More time mainly boosted confidence in the judgments and made the impressions a little more differentiated (Willis & Todorov, 2006). In another study, where people judged whether neutral faces looked threatening, consistent judgments formed from whatever could be seen within the first 39 milliseconds, while judgments of intelligence under the same conditions were less consistent (Bar et al., 2006). In these experiments, though, judgments being consistent or highly correlated means the judgments resembled one another, not that they matched the real personality of the person in the photo.

Other studies showed brief stretches of real behaviour rather than face photos. Researchers call these short observations “thin slices”. A 1992 meta-analysis gathered 38 results in which observations of expressive behaviour lasting under 5 minutes were used to predict various objective outcomes, and the overall effect size for the accuracy of those predictions was r = .39. When studies were compared with each other, predictions based on less than half a minute of observation did not differ significantly from those based on 4 to 5 minutes, and whether the ratings were based on the face, speech, the body or the tone of voice was not related to accuracy (Ambady & Rosenthal, 1992). The following year, strangers’ ratings of silent clips under 30 seconds long of college teachers in class significantly predicted students’ end-of-semester evaluations of those teachers, and for high school teachers they predicted a principal’s ratings. Ratings of even shorter 6-second and 15-second clips were strongly related to these criteria, while ratings of the teachers’ physical attractiveness were not as strongly related (Ambady & Rosenthal, 1993).

So there is no single answer to “how many seconds does a first impression take?”. Figures such as 3 or 7 seconds circulate online, but the times measured in research range from tens of milliseconds to several minutes, depending on what was judged (how threatening a face looked, or how someone teaches) and what was shown (a photo or a video). Predicting teaching evaluations is also not the same as reading personality. That a quick impression is shared by many people and that it matches someone’s real personality are separate questions, and the next section looks at the difference.

How well people read the personality of someone they have just seen is studied at “zero acquaintance”, when people know nothing about each other. In a 1992 study, 100 adults were videotaped entering a room, walking through it, sitting down, looking into the camera and reading a standard text, and then rated themselves on five personality factors. A group of 24 strangers who had never seen them rated their personality from just one of four kinds of information: a film with sound, a silent film, a still photo or an audio recording. Those ratings matched the people’s own ratings best for extraversion and conscientiousness, and how valid they were depended on which information the strangers had (Borkenau & Liebler, 1992). A 1994 review of 32 studies, covering people from strangers to close acquaintances, found that agreement among different people rating the same person (consensus) ranged from zero to a correlation of about .3 and was highest for extraversion, with no evidence that it rose as people got to know each other (Kenny et al., 1994). Consensus means that ratings resemble each other, not that they are accurate.

A 2007 study varied the viewing time directly. Videos of 30 college students each talking freely with another participant for 5 minutes were cut into slices of 5, 20, 45, 60 and 300 seconds and shown to 334 judges, and the personality benchmark was the average of personality questionnaires answered by the students themselves, their peers and their parents. Negative affect, extraversion, conscientiousness and intelligence were judged moderately well after just 5 seconds, but positive affect, neuroticism, openness and agreeableness needed more time to reach similar accuracy, and agreeableness was the hardest to judge. Overall, accuracy rose with viewing time, and 60 seconds gave the best ratio of accuracy to slice length. Slices from the third or fifth minute of the conversation were more accurate than the first minute, which the researchers suggested may be a nervous phase of introductions. In the researchers’ words, first impressions can be “remarkably right or dead wrong”, depending on the circumstances (Carney et al., 2007).

Why do traits differ? According to Funder’s Realistic Accuracy Model (RAM) of personality judgment, an accurate judgment needs behavioural cues that are relevant to the trait, those cues being available to the person judging, that person detecting them, and using them correctly (Funder, 1995). A trait that shows in behaviour, such as extraversion, offers many cues even in a short meeting, while traits experienced mostly on the inside offer few. A 2010 meta-analysis of 263 samples covering 44,178 people likewise found that meeting more often improved accuracy, but that a close relationship was needed for substantial gains, especially for traits low in visibility such as emotional stability (Connelly & Ones, 2010).

Making a good first impression and being a good person are not the same thing either. In a study in which 73 psychology freshmen rated one another after brief self-introductions (2,628 pairs), people higher in narcissism were more popular at first sight, and the aspects that are most harmful in the long run, exploitativeness and entitlement, looked most attractive at first (Back et al., 2010). This is an average tendency: making a good first impression is not a sign of narcissism. In discussion groups of 4 to 6 people who met 7 times, those with a strong tendency to self-enhance seemed agreeable, well adjusted and competent at the first meeting but were rated negatively after 7 weeks (Paulhus, 1998). The research on narcissism is covered in the guide to narcissism.

Many first-impression studies use photos of faces. That impressions from faces are shared from one person to the next has been known since the early 20th century, and recent studies have shown that surprisingly little time and effort are needed to reach that agreement. A 2015 review of this research presents evidence that impressions from faces predict important outcomes, such as investment decisions and the choice of leaders, but argues that the claim that these impressions correctly diagnose people has been greatly overstated in the research literature (Todorov et al., 2015).

How face impressions creep into real choices showed up in elections too. Judgments of which candidate in US congressional elections looked more competent, made from their faces alone, predicted the results better than chance, matching 68.8% of the 2004 Senate races, and such competence judgments emerged within a 1-second look at the faces. The researchers read this as a sign that quick, unreflective impressions can play a part even in voting, which is widely assumed to rest on rational, deliberate thought (Todorov et al., 2005). What this showed is not that faces reveal competence, but that impressions from faces can sway choices.

What is more, a single photo does not represent a person. A 2014 study that compared impressions from several photos of the same people found that for a range of social judgments, the differences between photos of the same person were as large as or larger than the differences between people, and that the photos people picked changed depending on whether they were for online dating or a political campaign (Todorov & Porter, 2014). This is one reason a single dating-app profile picture or ID photo is a poor basis for judging someone. Whether a face can reveal personality, and what to make of AI face reading, is covered in detail in the guide to the science of face reading.

The idea that first impressions stick goes back to a 1946 experiment by Solomon Asch. One group (34 people) heard a person described as “intelligent, industrious, impulsive, critical, stubborn, envious”, and another group (24 people) heard the same six words in reverse order. Although the words were identical, most of those who heard the first order pictured an able person with some shortcomings that did not overshadow their merits, while most of those who heard the reverse order pictured a “problem”, someone whose abilities were held back by serious difficulties. Words such as impulsive and critical were read in a positive or a negative light depending on the order. Asch explained this as a primacy effect, in which the first words set a direction and later words are read to fit it, but noted that it did not appear in every participant (Asch, 1946). The experiment looked at word order within one list heard at once, though, not at whether an impression lasts over time.

Later studies found the order effect again, but it can be smaller than its reputation. Someone described as “fun, witty and vicious” is typically rated more favourably than someone described as “vicious, witty and fun”: the same words in a different order change the impression, and that is the primacy effect. Five experiments with 2,882 people in total confirmed it once more, but found the effect to be small (Sullivan, 2019). Asch’s experiments are also often cited as evidence that warmth matters more than competence in first impressions, but a 2014 replication found no such effect: the role of warmth depended heavily on context, and competence was at least as important (Nauts et al., 2014).

First impressions leave a trace in real meetings too. In a 2017 study, 636 adults in an online survey rated their liking for, and the personality of, four people they had never met, from photos, and 55 of them later met one of those people in the lab, at least a month later (about 4 months on average), solving a quiz together and talking for two 10-minute sessions, 20 minutes in all. Most of them (89%) did not realise they were meeting someone whose photo they had seen. Liking based on the photo predicted liking after the meeting, and personality judgments carried over too, more weakly (except for extraversion). Part of the reason was that people who liked the photo behaved more warmly in the meeting, which drew more warmth from the other person in return: a first impression that partly fulfils itself. Liking based on the photo explained about 35% of the variance in liking after the meeting (less for personality judgments), which the researchers read as considerable consistency but also revision (Günaydın et al., 2017).

So can an impression never be undone? It can. In a series of experiments published in 2015, participants heard that a fictional man had broken into the neighbours’ homes, rummaged through the rooms and caused damage, and formed a negative impression. When they then learned that the break-in had been a search for children in burning buildings, their automatic, gut-level evaluation fully reversed. This happened only when the new information led them to reinterpret what they had learned before, it required at least some capacity to think it over, and the revised evaluation was still there 3 days later (Mann & Ferguson, 2015). Research has also looked at the other direction. According to the university’s news release, even for a fictional person known for many good deeds, a single bad act that strongly revealed their true nature fully reversed the automatic evaluation, and one such piece of information was enough to turn a good impression bad, but not always enough to turn a bad impression good (Cone & Ferguson, 2015).

The “Mehrabian rule” often comes up alongside first impressions. The figures, that 7% of what gets across comes from the words, 38% from the voice and 55% from the face, come from two studies the psychologist Albert Mehrabian published in 1967 (Mehrabian & Wiener, 1967; Mehrabian & Ferris, 1967). The equation as Mehrabian presents it is “Total Liking = 7% Verbal Liking + 38% Vocal Liking + 55% Facial Liking” (Mehrabian, 1981).

Mehrabian attached a clear limit to it, though. This equation about the relative importance of verbal and nonverbal messages comes from experiments on communicating feelings and attitudes, that is, liking and disliking, and unless a speaker is talking about their feelings or attitudes, it does not apply (Mehrabian, 1981). The experiments were also very narrow. The participants in the main study were 37 female psychology students, and the verbal content was a single word, “maybe”, a design in which words could hardly matter much from the start. The figures were put together from two studies, and neither study compared words, voice and face at the same time (Lapakko, 2007).

Mehrabian has said that these findings are often misquoted and that it is absurd to suggest that words make up only 7% of all communication. When someone tells you that the eraser you are looking for is in the second right-hand drawer of a desk in a third-floor office, how could anyone claim the words are only 7% of that message (Lapakko, 2007)? Even so, a 2007 study of 79 websites among the top Google results for the figures found that 63 of them (80%) presented them as a general fact about communication, and only 1 correctly described them as findings about emotional meaning (Lapakko, 2007).

So the lesson of Mehrabian’s results is not that words do not matter. It is closer to this: under those experimental conditions, when the words and the voice and face conveying liking or disliking contradicted each other, listeners gave more weight to the voice and face. What you say in an interview or on a first meeting still matters, and if you want to get a feeling across, it helps when your words and your tone say the same thing.

First, when you meet someone, treat your first impression as a hypothesis rather than a verdict. Traits that show readily, such as extraversion, are read reasonably well in a short meeting, but agreeableness was the hardest to judge from a brief look (Carney et al., 2007), and traits low in visibility, such as emotional stability, were judged much more accurately only in close relationships (Connelly & Ones, 2010). Choosing or ruling out people by their facial features or a single photo in particular can turn a weakly grounded impression into a label (Todorov et al., 2015; Todorov & Porter, 2014). It also helps to remember that your first impression can change how you treat someone, and that how you treat them can in turn change how they respond (Günaydın et al., 2017).

Second, if you worry about the impression you make, start from the likelihood that the other person liked you more than you think. After talking with someone new, people consistently underestimated how much the other person liked them and enjoyed the conversation. This “liking gap” appeared among strangers meeting in the lab, first-year college students getting to know their dorm mates and members of the public meeting at a workshop; it persisted across conversations of different lengths and sometimes lasted for months (Boothby et al., 2018).

Third, trying to make a good impression is not necessarily putting on a mask. People who were asked to “put their best face forward” on meeting others were not only better liked than those who were not, but were also seen more in line with their own distinctive self-reported traits and their intelligence test scores. The researchers’ analysis suggested this may be because they engaged more in the conversation (Human et al., 2012). Among new classmates, more accurate impressions of each other predicted spending more time together over the semester and more liking by its end, and first impressions tilted in a positive direction also strongly predicted relationships growing closer (Human et al., 2013). Taking an interest in the conversation and letting yourself be seen fits the research better than impression tricks do.

Fourth, if a first impression went wrong, give some context at the next meeting. In the experiments, new information that made earlier events look different changed impressions (Mann & Ferguson, 2015). These were experiments about fictional people, so they do not transfer directly to real life, and an explanation such as “I was nervous that day and did not say much” was not itself tested in the studies, but an explanation that lets someone see your earlier behaviour in a new light is worth trying. If meeting new people frightens you so much that you keep avoiding it, or it makes daily life hard, the guide to social anxiety may help more than impression tips. mind.io’s Big Five personality test shows five personality traits from your own answers, not from a first impression.

Q. How many seconds does it take to form a first impression? A. Impressions such as attractiveness or trustworthiness from a face photo form within a tenth of a second (0.1 seconds), and looking longer did not significantly increase how well they matched other people’s judgments (Willis & Todorov, 2006). How threatening a neutral face looked was judged consistently from what could be seen in 39 milliseconds (Bar et al., 2006). The accuracy of personality judgments, however, generally rises with longer viewing: in a study using 5-minute conversation videos, 60 seconds gave the best accuracy for the time spent (Carney et al., 2007).

Q. Are first impressions accurate? A. It depends on the trait. When 24 strangers rated people from just a video, a photo or an audio recording, their ratings matched the people’s own ratings best for extraversion and conscientiousness (Borkenau & Liebler, 1992), and in a study using conversation videos of 30 people, agreeableness was the hardest to judge (Carney et al., 2007). Traits that show less, such as emotional stability, became much more accurately judged only in close relationships (Connelly & Ones, 2010). Being popular at first sight does not mean someone is a good person either: people higher in narcissism were more popular at first sight (Back et al., 2010). This is an average tendency: a good first impression is not a sign of narcissism.

Q. Can a first impression change? A. Yes. There is a primacy effect, in which the first information sets the direction of an impression (Asch, 1946), but in experiments with 2,882 people it was small (Sullivan, 2019). Liking based on a photo explained about 35% of the variance in liking after meeting the person more than a month later, so it carried over considerably but was also revised (Günaydın et al., 2017). New information that made earlier events look different reversed automatic evaluations, and the change was still there 3 days later (Mann & Ferguson, 2015).

Q. Is the Mehrabian rule true? A. The 7% words, 38% voice and 55% face figures come from two experiments in 1967, but Mehrabian has stated that they do not apply unless someone is talking about their feelings or attitudes (Mehrabian, 1981). The main study had 37 female students hearing a single word, “maybe”, and neither study compared all three channels at once (Lapakko, 2007). It does not mean that words make up 7% of a conversation or of a first impression.

Q. How can I make a better first impression? A. What research supports is closer to an attitude than a technique. People consistently underestimated how much their conversation partners liked them (Boothby et al., 2018), so you have likely made a better impression than you think. People who tried to show their best self were better liked and also seen more accurately, possibly because they engaged more in the conversation (Human et al., 2012). If a first impression went wrong, you can offer context that lets the other person see your earlier behaviour differently (Mann & Ferguson, 2015).

Q. Is it fine to judge first impressions by the face alone? A. It is not advisable. Impressions from faces are shared among people and influence decisions such as investments and the choice of leaders, but the claim that they correctly diagnose people has been greatly overstated (Todorov et al., 2015). The impression the same person gives also varied a lot from photo to photo (Todorov & Porter, 2014). The accuracy of face reading and of impressions from faces is covered in the guide to the science of face reading.

References: Willis, J., & Todorov, A. (2006), First impressions: making up your mind after a 100-ms exposure to a face, Psychological Science 17(7), 592-598, https://doi.org/10.1111/j.1467-9280.2006.01750.x; Bar, M., Neta, M., & Linz, H. (2006), Very first impressions, Emotion 6(2), 269-278, https://doi.org/10.1037/1528-3542.6.2.269; Ambady, N., & Rosenthal, R. (1992), Thin slices of expressive behavior as predictors of interpersonal consequences: a meta-analysis, Psychological Bulletin 111(2), 256-274, https://doi.org/10.1037/0033-2909.111.2.256; Ambady, N., & Rosenthal, R. (1993), Half a minute: predicting teacher evaluations from thin slices of nonverbal behavior and physical attractiveness, Journal of Personality and Social Psychology 64(3), 431-441, https://doi.org/10.1037/0022-3514.64.3.431; Borkenau, P., & Liebler, A. (1992), Trait inferences: sources of validity at zero acquaintance, Journal of Personality and Social Psychology 62(4), 645-657, https://doi.org/10.1037/0022-3514.62.4.645; Kenny, D. A., Albright, L., Malloy, T. E., & Kashy, D. A. (1994), Consensus in interpersonal perception: acquaintance and the Big Five, Psychological Bulletin 116(2), 245-258, https://doi.org/10.1037/0033-2909.116.2.245; Carney, D. R., Colvin, C. R., & Hall, J. A. (2007), A thin slice perspective on the accuracy of first impressions, Journal of Research in Personality 41(5), 1054-1072, https://doi.org/10.1016/j.jrp.2007.01.004; Funder, D. C. (1995), On the accuracy of personality judgment: a realistic approach, Psychological Review 102(4), 652-670, https://doi.org/10.1037/0033-295X.102.4.652; Connelly, B. S., & Ones, D. S. (2010), An other perspective on personality: meta-analytic integration of observers’ accuracy and predictive validity, Psychological Bulletin 136(6), 1092-1122, https://doi.org/10.1037/a0021212; Back, M. D., Schmukle, S. C., & Egloff, B. (2010), Why are narcissists so charming at first sight? Decoding the narcissism-popularity link at zero acquaintance, Journal of Personality and Social Psychology 98(1), 132-145, https://doi.org/10.1037/a0016338; Paulhus, D. L. (1998), Interpersonal and intrapsychic adaptiveness of trait self-enhancement: a mixed blessing?, Journal of Personality and Social Psychology 74(5), 1197-1208, https://doi.org/10.1037/0022-3514.74.5.1197; Todorov, A., Olivola, C. Y., Dotsch, R., & Mende-Siedlecki, P. (2015), Social attributions from faces: determinants, consequences, accuracy, and functional significance, Annual Review of Psychology 66, 519-545, https://doi.org/10.1146/annurev-psych-113011-143831; Todorov, A., Mandisodza, A. N., Goren, A., & Hall, C. C. (2005), Inferences of competence from faces predict election outcomes, Science 308(5728), 1623-1626, https://doi.org/10.1126/science.1110589; Todorov, A., & Porter, J. M. (2014), Misleading first impressions: different for different facial images of the same person, Psychological Science 25(7), 1404-1417, https://doi.org/10.1177/0956797614532474; Asch, S. E. (1946), Forming impressions of personality, Journal of Abnormal and Social Psychology 41(3), 258-290, https://doi.org/10.1037/h0055756; Sullivan, J. (2019), The primacy effect in impression formation: some replications and extensions, Social Psychological and Personality Science 10(4), 432-439, https://doi.org/10.1177/1948550618771003; Nauts, S., Langner, O., Huijsmans, I., Vonk, R., & Wigboldus, D. H. J. (2014), Forming impressions of personality: a replication and review of Asch’s (1946) evidence for a primacy-of-warmth effect in impression formation, Social Psychology 45(3), 153-163, https://doi.org/10.1027/1864-9335/a000179; Günaydın, G., Selçuk, E., & Zayas, V. (2017), Impressions based on a portrait predict, 1-month later, impressions following a live interaction, Social Psychological and Personality Science 8(1), 36-44, https://doi.org/10.1177/1948550616662123; Mann, T. C., & Ferguson, M. J. (2015), Can we undo our first impressions? The role of reinterpretation in reversing implicit evaluations, Journal of Personality and Social Psychology 108(6), 823-849, https://doi.org/10.1037/pspa0000021; Cone, J., & Ferguson, M. J. (2015), He did what? The role of diagnosticity in revising implicit evaluations, Journal of Personality and Social Psychology 108(1), 37-57, https://doi.org/10.1037/pspa0000014 [examples from Cornell University’s news release, Re-examining the ‘first impressions’ adage, 12 June 2015: https://news.cornell.edu/stories/2015/06/re-examining-first-impressions-adage]; Mehrabian, A., & Wiener, M. (1967), Decoding of inconsistent communications, Journal of Personality and Social Psychology 6(1), 109-114, https://doi.org/10.1037/h0024532; Mehrabian, A., & Ferris, S. R. (1967), Inference of attitudes from nonverbal communication in two channels, Journal of Consulting Psychology 31(3), 248-252, https://doi.org/10.1037/h0024648; Mehrabian, A. (1981), Silent messages: implicit communication of emotions and attitudes, Wadsworth [formula and scope note quoted from the author’s page, archived at https://web.archive.org/web/20261012194735/http://www.kaaj.com/psych/smorder.html]; Lapakko, D. (2007), Communication is 93% nonverbal: an urban legend proliferates, Communication and Theater Association of Minnesota Journal 34, 7-19, https://doi.org/10.56816/2471-0032.1000; Boothby, E. J., Cooney, G., Sandstrom, G. M., & Clark, M. S. (2018), The liking gap in conversations: do people like us more than we think?, Psychological Science 29(11), 1742-1756, https://doi.org/10.1177/0956797618783714; Human, L. J., Biesanz, J. C., Parisotto, K. L., & Dunn, E. W. (2012), Your best self helps reveal your true self: positive self-presentation leads to more accurate personality impressions, Social Psychological and Personality Science 3(1), 23-30, https://doi.org/10.1177/1948550611407689; Human, L. J., Sandstrom, G. M., Biesanz, J. C., & Dunn, E. W. (2013), Accurate first impressions leave a lasting impression: the long-term effects of distinctive self-other agreement on relationship development, Social Psychological and Personality Science 4(4), 395-402, https://doi.org/10.1177/1948550612463735