The Neuroscience of First Impressions: Why 100 Milliseconds Shape Trust

The 100-millisecond finding

Willis and Todorov's 2006 study in Psychological Science presented participants with photographs of faces for durations of 100ms, 500ms, or 1000ms, then asked them to rate the faces on dimensions including trustworthiness, competence, likeability, aggressiveness, and attractiveness. The key finding: judgments made at 100ms correlated highly with judgments made by other participants with no time limit, and longer exposure did not significantly raise that agreement. Going from 100 to 500ms made judgments somewhat more negative and raised confidence; beyond that, extra time mostly added confidence and produced more differentiated impressions.

Note what this measures: agreement between quick and unhurried judgments of a photograph, not whether the judgments are accurate about the person. Still, it means an initial impression forms almost immediately, and later information is interpreted against it.

In a related study, Todorov and colleagues (2005) showed people the faces of US congressional candidates for as little as one second. The candidate judged more competent-looking won 68.8% of the 2004 Senate races, better than chance, and the competence judgments were linearly related to the margin of victory.

What the amygdala is assessing

Bar, Neta and Linz (2006) pushed exposure times even lower. Judgments of how threatening a neutral face looked were already consistent at 39 milliseconds, while judgments of intelligence were less consistent at that speed. The authors proposed that coarse, low-spatial-frequency visual information is enough for threat judgments.

Brain imaging points to the amygdala, a structure involved in detecting threat and social relevance. In an fMRI study by Winston and colleagues (2002), the amygdala responded more to faces that participants judged untrustworthy, whether or not they were asked to judge trustworthiness at the time. The response appears to run automatically, alongside any deliberate evaluation.

Thin-slice judgments: Nalini Ambady's research

Nalini Ambady's work on thin-slice judgments extended the first impression research from static images to brief dynamic samples. In her most-cited paper (Ambady & Rosenthal, 1993), strangers watched silent video clips of college teachers, under 30 seconds in total, and rated their nonverbal behaviour. Their combined ratings significantly predicted the end-of-semester evaluations given by the teachers' actual students, who had spent a whole term with them. A second study found the same for high school teachers rated by their principal.

In a third study in the same paper, even thinner slices of 6 and 15 seconds were still strongly related to the evaluations. Ratings of specific small behaviours, and of the teachers' physical attractiveness, predicted less well than the overall impression. The thin slice was capturing something stable about how each teacher came across.

What the thin slice is capturing is not a performance. It is the accumulated expression of habitual patterns, the automatic, unconsidered nonverbal behaviour that a person produces when they are not specifically attending to it. This is exactly what makes it diagnostic.

The halo effect and its practical consequences

The halo effect, the tendency for a positive impression on one dimension to produce positive impressions on other dimensions, is directly relevant to first impression research. When the initial assessment of a person is positive, observers subsequently rate them higher on dimensions that are not causally related to the original positive cue. In a classic study, Dion, Berscheid and Walster (1972) found that people judged photographs of attractive individuals as having more socially desirable personalities and expected them to lead more successful lives: "what is beautiful is good".

The halo effect works in both directions. Several small signals of tension, such as guarded posture, averted gaze and a tight voice, can add up to one overall negative impression, just as several positive signals add up to a positive one.

What is actually controllable

Much of a first impression comes from features that are not consciously controlled in the moment: habitual posture, default facial expression, movement quality, vocal tone. Trying to manage all of them consciously in real time is hard, and the effort of self-monitoring takes attention away from the person you are talking to.

What is controllable is the preparation that precedes the interaction. Specifically:

• Your state before you walk in. Anxiety shows in posture, gaze and voice. A short calming routine before a high-stakes interaction, such as two minutes of slow breathing, is easier to apply than trying to manage every signal while the conversation is under way. It is a practical rule of thumb rather than a tested intervention for first impressions.
• Habitual postural patterns. The posture you enter with is your default postural pattern, the one that is running automatically, below conscious attention. Changing this pattern requires consistent training over time, not a conscious adjustment made in the moment.
• How you frame the meeting. Treating an interaction as an opportunity rather than a threat may reduce the anxiety that shows in your behaviour. Evidence that this framing changes how others see you is limited, so treat it as a useful habit, not a proven technique.