Why First Impressions Matter: The Science of Snap Judgments

In 2006, Princeton psychologists Janine Willis and Alexander Todorov showed participants photographs of strangers for exactly 100 milliseconds — one-tenth of a second — and asked them to rate trustworthiness, competence, likeability, attractiveness, and aggressiveness. The ratings participants gave in 100 milliseconds correlated significantly with ratings made by participants who were given unlimited time. More exposure time produced more confidence and slightly more negative ratings, but the underlying judgment was essentially complete in the first blink. The brain had already decided.

The 100-Millisecond Window

The Willis and Todorov finding is not isolated. It is part of a convergent body of research across social neuroscience, cognitive psychology, and evolutionary biology showing that social judgments — particularly judgments about trustworthiness and threat — are among the fastest computations the brain performs.

The brain's speed here is not accidental. In ancestral environments, the cost of misreading a social encounter could be catastrophic: trusting someone who meant harm, or failing to trust someone who was safe. The pressure to resolve social uncertainty quickly was intense, and the result is a system that makes rudimentary judgments about other people at a rate that precedes most conscious processing.

Thin-slicing — the ability to extract meaningful patterns and make accurate predictions from very brief samples of behavior, appearance, or interaction, first systematically studied by psychologist Nalini Ambady at Harvard University.

What makes thin-slicing remarkable is not just its speed but its validity. Brief observations, in many domains, predict outcomes that matter — not always, and not equally across all traits, but substantially more than chance and often comparably to much longer observations.

Source: 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 — the foundational thin-slicing experiment demonstrating that trait judgments made after 100 ms correlate highly with unlimited-exposure ratings.

Source: Ambady, N. & Rosenthal, R. (1992), "Thin Slices of Expressive Behavior as Predictors of Interpersonal Consequences: A Meta-Analysis," Psychological Bulletin, 111(2), 256–274 — the original meta-analysis of thin-slicing research across 38 studies, finding consistent predictive validity from brief behavioral samples.

What Gets Judged Instantly — and What Doesn't

The brain does not evaluate all traits with equal speed or accuracy. Research by Alexander Todorov and colleagues at Princeton has identified trustworthiness as the single fastest and most automatic judgment — the trait that receives a snap evaluation before any other, likely because of its survival relevance. After trustworthiness, dominance and threat are evaluated nearly as quickly.

Other traits require more processing time and more informational richness. Intelligence, for example, is poorly judged from static faces alone (accuracy is near chance level). Personality traits like conscientiousness and openness to experience are better judged from brief behavioral observations — someone's office, their online profile, their music library — than from face photographs.

Trait	Judgment speed	Accuracy from face photo alone	Better predicted by
Trustworthiness	Very fast (< 100 ms)	Moderate	Face, posture, vocal cues
Dominance / threat	Very fast	Moderate-high	Face, body size, posture
Attractiveness	Very fast	High	Face, symmetry, averageness
Competence	Fast	Moderate	Face + contextual cues
Intelligence	Slow	Near chance	Behavioral observation
Conscientiousness	Slow	Low from face	Personal space, belongings
Warmth	Moderate	Low-moderate	Voice, facial dynamics

The practical implication is that snap judgments are not uniformly valid or invalid. They are more valid for some traits than others, and the traits for which they are most valid (threat, dominance, trustworthiness) happen to be those most relevant to basic social safety.

Source: Todorov, A., Pakrashi, M. & Oosterhof, N.N. (2009), "Evaluating Faces on Trustworthiness after Minimal Time Exposure," Social Cognition, 27(6), 813–833 — the detailed study of trustworthiness judgment speed and automatic activation.

Source: Gosling, S.D., Ko, S.J., Mannarelli, T. & Morris, M.E. (2002), "A Room with a Cue: Personality Judgments Based on Offices and Bedrooms," Journal of Personality and Social Psychology, 82(3), 379–398 — the study showing personal spaces predict personality more accurately than face photographs.

The Halo Effect: One Trait Colors Everything

The halo effect is the tendency for a single positive (or negative) impression of a person in one domain to influence evaluations of that person in unrelated domains. It was first documented systematically by psychologist Edward Thorndike in 1920, who found that military officers' ratings of their subordinates on traits like intelligence, physique, leadership, and character were far more positively correlated than the traits themselves could justify — the officers were projecting an overall impression onto all individual ratings.

Halo effect — a cognitive bias in which the perception of one positive quality in a person or thing creates a disproportionate positive impression of unrelated qualities, and conversely, a single negative quality creates a broader negative impression.

Physical attractiveness is the most studied and best-documented halo. Dion, Berscheid, and Walster's 1972 study "What Is Beautiful Is Good" showed that participants rated physically attractive individuals as more intelligent, more morally virtuous, more likely to be professionally successful, and more likely to have happy marriages — solely on the basis of photographs. No behavioral information was provided. The attractive face simply radiated positive attributes in all directions.

The halo effect has measurable real-world consequences. In legal settings, studies find that attractive defendants receive more lenient sentences for the same crimes as less attractive defendants, with an effect size that varies by crime type. In academic settings, teachers give higher grades to students whose physical appearance is more attractive, holding performance constant. In organizational settings, attractive individuals are more likely to be hired, receive higher starting salaries, and are promoted more quickly — a career-long advantage produced by an automatic perceptual bias operating largely outside conscious awareness.

Source: Thorndike, E.L. (1920), "A Constant Error in Psychological Ratings," Journal of Applied Psychology, 4(1), 25–29 — the original halo effect study with military officers, demonstrating the consistency of impression spillover across trait ratings.

Source: Dion, K., Berscheid, E. & Walster, E. (1972), "What Is Beautiful Is Good," Journal of Personality and Social Psychology, 24(3), 285–290 — the foundational attractiveness halo study.

Source: Langlois, J.H. et al. (2000), "Maxims or Myths of Beauty? A Meta-Analytic and Theoretical Review," Psychological Bulletin, 126(3), 390–423 — the large-scale meta-analysis confirming the attractiveness halo across social, professional, and legal domains.

The Primacy Effect: First In, First Weighted

The primacy effect in person perception describes the systematic tendency to weight information about a person more heavily when it is encountered early in a sequence rather than late, even when all the information is eventually available and equivalent.

Primacy effect — in person perception, the tendency for early information about a person to disproportionately shape the overall impression, with subsequent information interpreted through the lens of the initial impression rather than evaluated independently.

Solomon Asch demonstrated this in a series of classic studies in 1946. Participants were given two lists of six traits describing a person: one group received "intelligent, industrious, impulsive, critical, stubborn, envious"; another received the same traits in reverse order: "envious, stubborn, critical, impulsive, industrious, intelligent." The trait content was identical. The order produced markedly different overall impressions — the group that received positive traits first evaluated the person substantially more favorably.

Asch's interpretation was that early information establishes an overall impression — a gestalt — that then serves as an interpretive frame for subsequent information. "Critical" means something different as a description of someone already established as intelligent and industrious than it does as a description of someone already established as envious and stubborn. Early information does not merely register first; it actively shapes what subsequent information means.

This mechanism has a corollary: first impressions are not just formed quickly; they are resistant to revision. Once a positive or negative overall impression is established, confirming information is accepted and weighted normally, while disconfirming information is explained away, reinterpreted, or devalued. The impression, once formed, filters its own updates. This connects directly to confirmation bias — see the detailed analysis in cognitive biases explained.

Source: Asch, S.E. (1946), "Forming Impressions of Personality," Journal of Abnormal and Social Psychology, 41(3), 258–290 — the foundational person perception study establishing the primacy effect through trait list ordering.

The Neuroscience: What the Brain Is Actually Doing

The speed of first impressions is a computational achievement. Understanding what the brain is doing in those first 100 milliseconds illuminates why the process is both fast and resistant to overriding.

The primary visual cortex processes basic visual features — edges, contrast, motion — within 50–70 milliseconds of exposure. By 100 milliseconds, higher-level face processing in the fusiform face area (FFA) has begun — the brain has already identified that it is looking at a face. By 150–200 milliseconds, neural signals have reached the amygdala, which begins computing emotional relevance. By 200–250 milliseconds, the orbitofrontal cortex is integrating the threat/reward evaluation with contextual knowledge.

The whole sequence, from first visual contact to a rudimentary trustworthiness evaluation, takes less time than the average eye blink.

Fusiform face area (FFA) — a region in the fusiform gyrus of the temporal lobe that is selectively activated by face perception, involved in the holistic processing of faces as distinct from other objects.

The critical structural feature is that this cascade largely bypasses conscious awareness. By the time the prefrontal cortex — the seat of deliberate reasoning — receives the signal and could potentially override the automatic evaluation, the first impression has already been formed and is influencing subsequent processing. The conscious mind does not generate the first impression; it receives it.

Todorov and colleagues have mapped the facial features that drive trustworthiness and dominance judgments using computational models. Trustworthy-looking faces share features with happy expressions (upturned mouths, visible whites around the irises in an upward gaze); threatening-looking faces share features with angry expressions. The brain is, in part, reading emotional expression history in stable facial structure — a shortcut that works better than chance but produces systematic errors for individuals whose neutral face happens to resemble anger or happiness.

Source: Todorov, A., Said, C.P., Engell, A.D. & Oosterhof, N.N. (2008), "Understanding Evaluation of Faces on Social Dimensions," Trends in Cognitive Sciences, 12(12), 455–460 — the two-dimensional model of face evaluation (trustworthiness vs. dominance) and its neural correlates.

Source: Engell, A.D., Haxby, J.V. & Todorov, A. (2007), "Implicit Trustworthiness Decisions: Automatic Coding of Face Properties in the Human Amygdala," Journal of Cognitive Neuroscience, 19(9), 1508–1519 — the fMRI study showing amygdala activation to trustworthy vs. untrustworthy faces without explicit instruction.

How Accurate Are Snap Judgments? The Evidence

The crucial question about first impressions is not just how fast they form but whether they predict anything real. The answer is more complex than either "snap judgments are meaningless noise" or "your gut is always right."

In Nalini Ambady's foundational thin-slicing research, 30-second video clips of teachers were shown to raters unfamiliar with the teachers. The raters' impressions of warmth, enthusiasm, and effectiveness correlated significantly with end-of-semester student evaluations — from students who had spent an entire semester with those teachers. Six-second clips produced the same result. Two-second clips produced weaker but still above-chance correlations.

In political science, Todorov and colleagues found that face-based competence ratings of US Senate and House candidates — made in one-second exposures by participants with no knowledge of the candidates — predicted actual election outcomes with about 68% accuracy. Snap judgments of political candidates' faces appear to capture something about their electability, though whether this is a feature (voters are sensing something real) or a bug (voters are selecting on irrelevant perceptual features) is a live debate.

In medicine, brief observations of surgeons' vocal characteristics — warmth, hostility, dominance, anxiousness — extracted from short audio clips of patient interactions predicted malpractice claim history with significant accuracy. Surgeons with dominant, unconcerned vocal tones had higher malpractice histories; those with warm, concerned tones had lower ones.

Q: Do snap judgments reflect accuracy or just consistent bias?
A: Both. For some traits — trustworthiness signals, emotional states, sexual orientation in some study designs — brief exposures carry genuine information that produces above-chance accuracy. For others — intelligence, honesty, moral character — brief exposures carry little genuine signal but produce consistent impressions driven by appearance-based heuristics. The consistency should not be confused with validity.

Source: 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 — the study showing that 30-second and 6-second video clips predicted full-semester student evaluations of teachers.

Source: Todorov, A., Mandisodza, A.N., Goren, A. & Hall, C.C. (2005), "Inferences of Competence from Faces Predict Election Outcomes," Science, 308(5728), 1623–1626 — the election prediction study using one-second face exposure ratings.

Source: Ambady, N. et al. (2002), "Surgeons' Tone of Voice: A Clue to Malpractice History," Surgery, 132(1), 5–9 — the malpractice prediction study from brief voice recordings.

When First Impressions Go Wrong: The Systematic Errors

The same mechanisms that make thin-slicing useful produce systematic errors when they are applied to cues that have no genuine diagnostic value.

The strongest documented case is race. Research by Jennifer Eberhardt at Stanford University and others has found that facial features associated with stereotypical Black appearance predict longer criminal sentences, even controlling for crime severity, criminal history, and other legally relevant factors. The mechanism operates through the halo effect in reverse: the automatic racial categorization triggers stereotypic associations that color subsequent evaluations of unrelated attributes like dangerousness and culpability.

The "babyfaceness" effect is another documented source of error. People with baby-faced features (large eyes relative to face size, small nose, round face) are perceived as warmer, more naive, less dominant, and more honest — a cross-culturally robust finding by Leslie Zebrowitz at Brandeis University. Baby-faced defendants are more likely to be found not guilty of intentional crimes but more guilty of negligence — because the impression of naivety fits one crime type but not the other. This is a first impression producing different legal outcomes based on face geometry, not character.

Source: Eberhardt, J.L., Davies, P.G., Purdie-Vaughns, V.J. & Johnson, S.L. (2006), "Looking Deathworthy: Perceived Stereotypicality of Black Defendants Predicts Capital-Sentencing Outcomes," Psychological Science, 17(5), 383–386 — the study showing that racial stereotypicality of appearance predicted death sentence rates in capital cases.

Source: Zebrowitz, L.A. & McDonald, S.M. (1991), "The Impact of Litigants' Babyfaceness and Attractiveness on Adjudications in Small Claims Courts," Law and Human Behavior, 15(6), 603–623 — the babyfacedness study in legal outcomes.

The Primacy Asymmetry: Positive vs. Negative First Impressions

Not all first impressions are equally sticky. Research by Tali Sharot and others on updating and belief revision suggests that positive and negative initial impressions are not symmetrically resistant to revision.

In general, negative first impressions are more resistant to revision than positive ones. This negativity bias in impression updating reflects the broader finding that negative information is processed more thoroughly, weighted more heavily, and remembered longer than positive information of equivalent intensity. A negative first impression may require substantially more positive information to update than a positive first impression requires negative information to damage.

Negativity bias — the general psychological tendency for negative events, information, and emotions to have greater impact on attitudes, memory, and judgment than positive events of equivalent magnitude.

The evolutionary rationale is familiar: the cost of incorrectly trusting someone who means harm is typically higher than the cost of incorrectly distrusting someone who is safe. A negatively biased impression-revision system errs in the direction of caution. The problem is that in modern contexts — hiring decisions, medical evaluations, police encounters — the negative first impression generated by an irrelevant cue (appearance, accent, race) produces durable downstream effects that resist correction.

Source: Ito, T.A., Larsen, J.T., Smith, N.K. & Cacioppo, J.T. (1998), "Negative Information Weighs More Heavily on the Brain: The Negativity Bias in Evaluative Categorizations," Journal of Personality and Social Psychology, 75(4), 887–900 — the ERP and behavioral study demonstrating negativity bias in evaluation.

Can First Impressions Be Revised?

They can. The question is under what conditions and at what cost.

The research consistently finds that revision is possible but requires three things: motivation, opportunity, and informational quantity. A single disconfirming piece of information is typically insufficient to revise a strong first impression — the impression system applies something like a burden-of-proof standard that disconfirming evidence must meet before updating occurs. This is not principally stubbornness; it is Bayesian updating, which appropriately weights prior beliefs when new evidence arrives. The problem is that the prior beliefs themselves may have been formed on thin, unreliable evidence.

Revision is more likely when the initial impression was consciously flagged as uncertain. When people recognize that they have only seen someone briefly, or that their impression was formed under distorting conditions, they hold the impression more loosely and update more readily. When the initial impression feels confident and was formed in circumstances that felt informative, revision is harder.

The cognitive dissonance literature, explored in depth at the Festinger cognitive dissonance experiment, is relevant here: holding a strong first impression and encountering clear contradictory evidence creates dissonance, which can motivate reinterpretation of the new evidence rather than revision of the impression.

Source: Rydell, R.J. & McConnell, A.R. (2006), "Understanding Implicit and Explicit Attitude Change: A Systems of Reasoning Analysis," Journal of Personality and Social Psychology, 91(6), 995–1008 — the study of conditions under which explicit attitude change (first impression revision) occurs and what distinguishes them.

What People Get Wrong About First Impressions

The most common misconception is that first impressions are superficial, unreliable, or easily overridden by subsequent information. The research shows all three of these beliefs are wrong to varying degrees.

First impressions are not superficial in the sense of being unimportant. They predict real outcomes including hiring decisions, legal verdicts, election results, and medical communication quality. They may be based on shallow cues, but their consequences are deep.

They are not uniformly unreliable. For some traits (emotional states, broad trustworthiness and threat signals), brief observations carry genuine diagnostic signal. For others (intelligence, moral character), they carry almost none. The challenge is that people's confidence in their snap judgments does not track this distinction — high confidence accompanies both accurate and inaccurate impressions equally.

And they are not easily overridden. The primacy effect and the negativity bias in revision mean that the default is for first impressions to persist and to filter subsequent information rather than update in response to it. Deliberate correction is possible but requires sustained effort and motivation.

See how the halo effect intersects with the broader psychology of persuasion in the psychology of persuasion, where the "liking" principle shows that attractiveness-based halos directly amplify compliance.

Q: How long does a first impression last?
A: There is no precise answer because durability depends on trait type, impression strength, and subsequent experience. For clear, strong impressions (someone who acted very aggressively in a first meeting), the impression can persist for years and influence interpretation of benign behavior. For weak, uncertain impressions, revision occurs more readily with new information. In organizational contexts, negative first impressions from a job interview can influence performance evaluations for months.

Q: Are people aware that they are forming first impressions?
A: Partially. People are generally aware that they are evaluating new acquaintances. But the specific content of snap judgments — the fact that trustworthiness was assessed in under 200 milliseconds by the amygdala before conscious evaluation — is not available to introspection. People experience the result of the impression (a feeling of trust or unease) without access to the process that generated it.

Q: Do people who know about the halo effect resist it?
A: Not reliably. Like most cognitive biases, the halo effect operates primarily through automatic processing that does not consult conscious knowledge before acting. Knowing that attractive people get an unearned credibility boost does not prevent the automatic credibility boost from occurring. Correction requires noticing the bias in the moment and applying a deliberate counterweight — which is effortful and imperfect.

Q: Is the primacy effect stronger than the recency effect?
A: In person perception, yes — the primacy effect typically dominates. In memory tasks for lists of items, there is a strong recency effect (most recently presented items are most remembered) alongside the primacy effect. In social impressions, recency effects are smaller and primacy effects are more robust, particularly when the initial encounter was significant.

Q: Can first impressions be strategically managed?
A: Yes, within limits. Research on impression management shows that dress, posture, speech rate, and warmth cues can systematically shift others' impressions in predictable directions. But the most powerful driver of trustworthiness impressions — structural facial features — is not controllable. And sophisticated impression managers can sometimes trigger a "trying too hard" counter-reaction that undermines the effect they sought.

Q: Do online first impressions follow the same rules as in-person ones?
A: Largely yes, with some differences. Profile photographs trigger the same rapid trustworthiness and attractiveness evaluations. The primacy effect operates in online contexts — early messages in a chat thread disproportionately shape the conversation's tone. What differs is the reduced availability of vocal, postural, and movement cues, and the increased prominence of language-based cues (grammar, vocabulary, response speed).

The Takeaway

The brain's first-impression machinery is a high-speed social evaluation system shaped by evolutionary pressures that made quick social assessment essential. It forms judgments in 100 milliseconds, weights them heavily against subsequent information, applies halo effects that allow a single trait to inflate or deflate all others, and produces predictions about real-world outcomes that are better than chance in some domains and dramatically worse than chance in others. The gap between the system's subjective feeling of accuracy and its actual accuracy is one of the largest and most consequential in social psychology — people feel equally confident in snap judgments that are valid (threat assessment from body language) and in ones that are not (intelligence from facial bone structure). Understanding this gap does not eliminate first impressions; it makes the question of when to trust them a more tractable one.

---

Read next:

• The Psychology of Decision Making
• The Bystander Effect Explained
• The Skinner Box and Operant Conditioning