The Science of First Impressions

The old saying — that you never get a second chance to make a first impression — turns out to be almost literally true. Landmark work by Janine Willis and Alexander Todorov demonstrated that after just one hundred milliseconds of exposure to a face, most people have already formed stable judgments about that person’s trustworthiness, competence, likeability, and aggressiveness. Additional viewing time barely shifts the initial impression; it mostly increases the observer’s confidence in it.

First impressions are not curiosities of social psychology. They drive hiring, voting, dating, medical decisions, and countless everyday judgments about strangers. When the machinery is invisible to us, we mistake our first read for objective knowledge. Understanding how the machinery works is the first step toward not being run by it.

Face processing is one of the most specialised systems in the human brain. Regions in the ventral occipito-temporal cortex, particularly the fusiform face area, respond within roughly 170 milliseconds of face exposure. Overlapping social-evaluation circuitry — in the amygdala, orbitofrontal cortex, and medial prefrontal cortex — turns those perceptual signals into inferences about intent, character, and category almost immediately. A large body of research by Todorov and colleagues has mapped the two primary dimensions along which faces are evaluated: *trustworthiness* (roughly, valence) and *dominance* (roughly, arousal). These dimensions are stable across cultures and are computed from subtle configural cues — the tilt of the brow, the width of the smile, the shape of the eyes. Once assigned, these attributions influence downstream decisions well beyond their accuracy. Faces perceived as more trustworthy receive more lenient sentences from juries; faces perceived as more competent win more elections; and once an impression is formed, subsequent behaviour is often reinterpreted to fit it — a well-documented confirmation bias. The underlying architecture is exactly the predictive one described in *Why Your Brain Is Not an Objective Recorder*. The brain does not gather evidence and then judge; it judges rapidly on partial evidence and then gathers confirmation.

Recent studies have refined the picture in three ways. First, first impressions are surprisingly resistant to disconfirming evidence — a single trust-violating action outweighs many trust-consistent ones, but the reverse is not symmetrical. Second, cultural priors calibrate the specific features that map to trust or dominance, so cross-cultural impressions can misfire. Third, exposure to a broader range of faces reduces the strength of some spurious attributions, suggesting that these snap judgments are not immutable — the priors can be updated with sustained new experience.

For everyday life, the practical work is not to try to suppress first impressions; it is to *notice* them as first impressions rather than facts. In consequential contexts — hiring, evaluations, professional referrals — introducing structure and time between first read and final decision demonstrably improves accuracy. In personal life, the practice is quieter: naming a first impression as one’s own prior, and holding the door open to being wrong.

**“I don’t judge people by appearance.”** The evidence says otherwise, for everyone. The relevant question is whether we know we are doing it. **“First impressions are usually accurate.”** They are surprisingly reliable *between observers* — most people converge on similar impressions from the same face — but this is not the same as being *accurate*. High inter-observer agreement can coexist with systematically biased perception.