The science of first impressions: why the first few minutes weigh so much

Published Sep 19, 2026 · 9 min read · By Julia, 4Keeps

"Don't judge a book by its cover" is good advice that almost nobody follows, and there's a reason. Your brain isn't built to wait for all the facts. It's built to guess, fast, and then check its guess against whatever comes next. On a first date that means two people are forming a working theory of each other from a handful of seconds, a photo, a handshake, and whatever came out of their mouths first. Here's what's actually happening under the hood, why those early moments carry so much weight, and what to do about it on both sides of the table.

Early on, the other person has almost no data about you. So every piece they do get counts for more than it ever will again.

Your brain is a prediction machine

For most of the last century we pictured perception as a camera: information comes in, the brain processes it, and a judgment comes out. The current model in neuroscience runs closer to the opposite direction. Under what's called predictive processing (associated with researchers like Karl Friston and philosopher Andy Clark), the brain is constantly generating predictions about what it's about to see, hear, and feel, and the senses mostly report back where those predictions were wrong. Neuroscientist Lisa Feldman Barrett puts it bluntly: your brain's most important job isn't thinking or feeling, it's anticipating, so it can keep you alive and spend energy wisely.

The currency of that system is prediction error: the gap between what you expected and what you got. Small errors get quietly absorbed. Big ones grab your attention and force an update. This is why a stranger who does something unexpected (an odd joke, a flash of rudeness, surprising warmth) sticks in your memory while a hundred ordinary moments evaporate.

Apply that to meeting someone. The moment you see a new person, your brain doesn't wait for a résumé. It reaches for everything it already knows (every face you've ever seen, every person who reminded you of this one, every story about "guys like him" or "women like her") and produces a best guess about who this is and how it'll go. That guess is your first impression. It's not a conclusion. It's a prior.

Why the beginning is so heavy: limited data

Statisticians have a precise way to describe what happens next, and it explains almost everything about first impressions. In Bayesian reasoning you start with a prior belief and update it with each new piece of evidence. How much any single piece of evidence moves you depends on how much you already have.

Think of it like a batting average. After one at-bat, a player is either hitting 1.000 or .000. One swing decides everything. After five hundred at-bats, one more swing barely moves the number. People work the same way. On date one, your date has maybe three data points about you: your photos, your first message, and the first thirty seconds in person. Each one is a third of everything they know. By month six, the fact that you were ten minutes late once is one data point among thousands, and it barely registers.

That's the whole case for caring about first impressions in one sentence: they're formed when the evidence is thinnest, so each piece of evidence is at its most powerful. It's not that the brain is shallow. It's that it's doing honest math on a tiny sample.

What the research says about how fast it happens

• A tenth of a second for a face. Princeton psychologists Janine Willis and Alexander Todorov (2006) showed people photos of faces for just 100 milliseconds. Their snap ratings of trustworthiness, likability, attractiveness, and competence lined up closely with ratings made by people who had unlimited time to look. More time mostly made people more confident, not more different.
• Thirty silent seconds for a person. In Nalini Ambady and Robert Rosenthal's classic "thin slices" study (1993), strangers watched soundless 30-second clips of college teachers. Their ratings predicted the end-of-semester evaluations from students who had spent months in those classrooms. Even clips cut down to six seconds still predicted the ratings.
• A few minutes for a relationship. Communication researchers Michael Sunnafrank and Artemio Ramirez (2004) paired up strangers on the first day of a class for a short conversation. How much each person expected to get out of knowing the other after just three to ten minutes predicted how close they actually became nine weeks later.
• Warmth first, competence second. Susan Fiske, Amy Cuddy, and Peter Glick (2007) showed that we size up strangers on two main dimensions, warmth (are you friend or foe?) and competence (can you act on it?), and that warmth is judged first and weighted most. On a date, "is this person kind and safe?" gets answered before "is this person impressive?"

Why first impressions stick: the brain protects its prior

If first impressions were just quick, they wouldn't matter much. The problem is that they're sticky, for a few well-documented reasons.

• The primacy effect. In a famous 1946 experiment, Solomon Asch described a person as "intelligent, industrious, impulsive, critical, stubborn, envious." Another group heard the exact same words in reverse order. The first group rated the person far more favorably. Same facts, different opening, different verdict. Early information sets the frame that later information gets read through.
• The halo effect. One strong early trait colors everything else. In a 1972 study by Karen Dion, Ellen Berscheid, and Elaine Walster, people assumed that more attractive strangers were also kinder, more interesting, and more likely to have happy marriages. A great first impression on one dimension quietly lends credit to the rest of you.
• Confirmation bias. Once you have a hypothesis, you notice evidence that fits it and explain away evidence that doesn't. If she decided you were arrogant in minute one, your confident story in minute twenty reads as more arrogance.
• Negativity bias. Roy Baumeister and colleagues reviewed the research in a paper titled simply "Bad Is Stronger Than Good" (2001). Across domains, negative information gets more attention and more weight than positive information of the same size, likely because missing a threat was historically costlier than missing an opportunity. A bad first moment takes several good ones to undo.
• Self-fulfilling prophecy. This is the one that should get your attention. In a 1977 study by Mark Snyder, Elizabeth Tanke, and Ellen Berscheid, men talked on the phone with women they'd never seen, after being shown a photo they believed was her. Men who thought she was attractive were warmer on the call, and the women, who knew nothing about the photo, responded by becoming warmer, more confident, and more likable to outside listeners. A first impression doesn't just read a person. It changes how you treat them, which changes how they act, which "confirms" the impression.

Put those together and you get a system that forms a guess on very little data, then quietly defends it. That's efficient when you're deciding whether a stranger on a dark street is a threat. It's much less useful when you're deciding whether someone could be your person.

But are first impressions accurate?

Sometimes, and less often than they feel. Thin slices are genuinely good at picking up some traits, especially how outgoing and energetic someone is. They're much worse at the traits that matter most in a partner: trustworthiness, loyalty, kindness under stress, how someone handles conflict. In his book Face Value (2017), Todorov himself makes the case that the faces we judge as trustworthy aren't reliably more trustworthy. Your prediction machine is fast and confident. It is not a lie detector.

The brain is also especially likely to misread the thing everyone brings to a first date: nerves. Two findings help here:

• The illusion of transparency. Kenneth Savitsky and Thomas Gilovich (2003) found that people consistently overestimate how visible their anxiety is to others. You feel like your nerves are on a billboard. Your date mostly can't see them.
• The liking gap. Erica Boothby and colleagues (2018) found that after a first conversation, people routinely underestimate how much the other person liked them. You're probably doing better than you think.

How to make a good first impression (the inputs you control)

You can't control someone else's priors, their ex, or the face they happen to associate with bad news. You can control the first data points you hand them. Since those are weighted most, spend your effort there.

• 1. Remember the first impression starts before you meet. On an app, your first photo is the 100-millisecond face test. Choose one that's clear, well lit, recent, and warm, with a real smile and eye contact with the camera. Our guide to optimizing your dating profile goes photo by photo.
• 2. Arrive on time and a little early. Late is a data point, and on date one it's an expensive one. Early also gives you a minute to settle, which helps with everything below.
• 3. Own the first ten seconds. Stand up, make eye contact, smile with your eyes, say their name. These are the signals the warmth detector is scanning for, and they're almost entirely under your control.
• 4. Lead with warmth, save the impressive stuff. Because warmth gets judged first and weighted most, curiosity beats credentials. Ask about them before you tell them about you. Competence will come through on its own.
• 5. Dress for the person you are, one notch up. Clothing is one of the few cues visible from across the room. It should look like you on a good day, not a costume. Fit and grooming matter far more than price.
• 6. Keep the opening light and positive. Negativity bias means an early complaint (the traffic, your job, your last date) lands harder than you mean it to. Save the heavy topics for when there's more data in the bank. We cover the timing in when to mention sensitive things.
• 7. Put the phone away, face down or out of sight. A phone on the table is a constant low-level signal that someone else might be more important.
• 8. End on a high. First impressions have a twin: last impressions. Daniel Kahneman's peak-end research shows people remember an experience largely by its best moment and how it ended. Leave while the conversation is still good rather than letting it fade out. For the rest of the date itself, see the first date done right.

If you got off on the wrong foot

A shaky start isn't fatal. It just means you need more new data, and the right kind. Prediction errors are what force an update, so the fix is to be clearly, consistently different from the impression you gave. If you were nervous and stiff, get warmer and more curious as the night goes on. If you said something that landed wrong, a light, direct acknowledgment ("that came out wrong, what I meant was...") is a strong signal of self-awareness, which people rate highly. What doesn't work is over-apologizing, which keeps the bad moment at the center of the conversation and gives it more weight, not less.

The other half: go easy on your own prediction machine

Everything above applies to you too. You're forming a first impression of your date on the same thin data, with the same biases, and you're probably more confident in it than the evidence supports. That's worth taking seriously, because the costs aren't symmetric: writing off a good person on a nervous first date is a mistake you'll never find out you made.

• Separate "red flag" from "first-date nerves." Rudeness to the server, lying, pressure, and contempt are real data even on day one. Awkwardness, a bad joke, and a slow start usually aren't.
• Notice when your brain is matching a pattern, not a person. If someone "just reminds you of" your ex, that's your prior talking. Ask what they've actually done.
• Give "neutral to pleasant" a second date. Attraction often grows with familiarity, which is one of the most replicated findings in psychology. We break down that mechanism in can you manufacture chemistry?
• Watch for the self-fulfilling prophecy. If you walk in expecting a dud, you'll be a little colder, they'll be a little flatter, and you'll get the dud you predicted. Show up warm and let them earn the verdict.

A first impression is your brain's opening bid, not its final answer. Make yours count, and let theirs be revised.

The science comes down to this. People are prediction machines working with almost no data at the start, so the first few moments carry outsized weight. That's a reason to prepare, not to panic. Hand people strong early data, trust that you're doing better than it feels, and give others the same benefit of more data before you decide.

Sources

• Willis & Todorov (2006), “First impressions: making up your mind after a 100-ms exposure to a face” Psychological Science, 17(7), 592–598.
• Ambady & Rosenthal (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.
• Sunnafrank & Ramirez (2004), “At first sight: persistent relational effects of get-acquainted conversations” Journal of Social and Personal Relationships, 21(3), 361–379.
• Dion, Berscheid & Walster (1972), “What is beautiful is good” Journal of Personality and Social Psychology, 24(3), 285–290 — the attractiveness halo.
• Baumeister, Bratslavsky, Finkenauer & Vohs (2001), “Bad is stronger than good” Review of General Psychology, 5(4), 323–370.
• Snyder, Tanke & Berscheid (1977), “Social perception and interpersonal behavior: on the self-fulfilling nature of social stereotypes” Journal of Personality and Social Psychology, 35(9), 656–666.
• Kahneman, Fredrickson, Schreiber & Redelmeier (1993), “When more pain is preferred to less: adding a better end” Psychological Science, 4(6), 401–405 — the peak-end rule.
• Clark (2013), “Whatever next? Predictive brains, situated agents, and the future of cognitive science” Behavioral and Brain Sciences, 36(3), 181–204 — predictive processing.