Where people actually look at your face: what eye-tracking data shows

You have been staring at your own photo for several minutes — eyes, brow, nose, mouth, back to the eyes. Every feature has become a separate problem.

Nobody else looks at that face this way. The first landing point is near the eyes, recognition peaks after two fixations, and the viewer has moved on before a feature-by-feature audit could start. Context in our first-impression psychology glossary.

Key numbers

• 1.06° below the midpoint of the eyes — the average first-saccade landing point in face identification. Peterson & Eckstein (2012)
• two fixations — where face-recognition accuracy peaked in one controlled study. Hsiao & Cottrell (2008)
• 300 ms — about how long a single fixation lasted. Hsiao & Cottrell (2008)
• 40% — the share of looking time spent on the eye region in an early face-photograph study. Janik et al. (1978)
• 12% measured vs 72% estimated — mutual eye contact in a getting-acquainted conversation, and what people guessed it was. Rogers et al. (2018)
• 3.65 fixations — all people made even when given a full 1,500 ms to look at a face. Or, Peterson & Eckstein (2015)
• 100 ms — enough exposure for several face-trait judgements to correlate highly with unlimited-time judgements. Willis & Todorov (2006)

Table of contents

• Where does the first look land?
• How many looks does it take?
• How much of the time goes to the eyes, and what gets the rest?
• And in an actual conversation?
• Does everyone scan a face the same way?
• How fast is the judgement made?
• What does this mean for the face you are worried about?
• What eye-tracking cannot tell you
• What we could not verify
• How to cite this page
• The bottom line
• Sources

Where does the first look land?

Just below the eyes, in the blank space between the eyes and the nose. In a face-identification task the average first saccade landed 1.06° below the midpoint of the eyes; judging emotion pushed it to 1.94°, judging gender kept it at 1.09°. Our face perception glossary is useful here, because “looking at the eyes” is usually shorthand for sampling a wider region.

The point held up: stretching a presentation from 350 ms to 1,500 ms did not move it, and 50 people identifying famous faces landed at 1.12° below the eyes. Forcing them to fixate the forehead instead cost 0.143 in proportion correct; forcing them to the mouth cost 0.148. Peterson & Eckstein 2012 Not a dot on every face — a population tendency with a price for missing it.

How many looks does it take?

Two. Recognition accuracy peaked at two fixations and did not improve with more; one already beat chance. Given a two-fixation budget people used 1.81; given no restriction, 3.28.

Each fixation lasted about 300 ms. Hsiao & Cottrell 2008 Given a full 1,500 ms, people still made only 3.65 fixations. Or et al. 2015 One qualification: this was recognition. Deciding whether a face is attractive or safe is not the same task as naming it.

A fair reading: a few samples support recognition. Not: every social judgement ends after two looks.

How much of the time goes to the eyes, and what gets the rest?

A disproportionate share, not all of it. In the early face-photograph study observers spent about 40% of looking time on the eye region; across faces of four species — human, monkey, dog and cat — the eyes drew the most fixations, then the nose, then the mouth. Guo et al. 2010

The rest follows the task. When speech in a videotaped face got harder to hear, mouth fixations rose from roughly 35 to 55%. Hessels 2020 “Where people look” is partly a question about what the face is carrying.

And in an actual conversation?

Far lower than people believe. Measured mutual eye contact averaged 12% of a conversation; asked to estimate it, 171 students said 72%. Mutual face gaze averaged 63% — looking at someone's face is not the same as meeting their eyes.

During conversation, gaze on the partner's face split roughly 48% eyes and 29% mouth while speaking and 43% eyes and 34% mouth while listening; people looked away about 29% of the time while speaking against 10% while listening. Rogers et al. 2018

Real dates show the same gap between belief and measurement. In 240 speed-dates, participants put their own eye contact at 60%; the eye-trackers measured 8.25%. Each additional minute of genuine mutual eye contact roughly tripled the odds of a yes (OR = 2.80). Hoffmann et al. 2024 Live interaction is a moving exchange, not a portrait held under a tracker.

Does everyone scan a face the same way?

No, and the differences are larger than the textbook triangle suggests. Western Caucasian observers fixated the eye region while East Asian observers fixated the centre of the face, across all three tasks tested. Blais et al. 2008 Individual patterns are stable enough to identify a person, and were still recognisably the same 18 months later. Mehoudar et al. 2014

Then the complication. Measure only the very first eye movement and both cultures land in the same featureless spot between the eyes and nose, 8% of the interocular distance apart. Or et al. 2015 And the familiar “left eye first” result reverses when the trial starts from the left of the screen — which undercuts a lot of published heat maps. Arizpe et al. 2012

How fast is the judgement made?

Very fast. Trait judgements after 100 ms correlated highly with judgements made with no time limit — attractiveness, likeability, trustworthiness, competence, aggressiveness. Going to 500 ms made them more negative and more confident without improving their agreement with unlimited-time judgements.

Willis & Todorov 2006 Consistent threat judgements of neutral faces can form from whatever arrives in the first 39 ms. Bar, Neta & Linz 2006 None of that means a stranger has understood your biography. It means the first-impression system commits to a direction before deliberate inspection catches up.

What does this mean for the face you are worried about?

Your mirror audit has no counterpart in ordinary looking. The eye region genuinely is where attention concentrates — that part is real — but nothing here shows strangers scoring pores, angles and asymmetries in sequence.

Call it the Two-Sample Rule: a couple of samples near the eyes, combined with the rest of the face and the context, then gone. It is why our first impression statistics treat perception as the unit, not a ladder of facial parts.

There is a measured sense of scale for this. Covering half a face with a mask cost 3.5% matching accuracy for familiar faces and 7.5% for unfamiliar ones — while a disguise built to make someone look unlike themselves cost 35%. Noyes et al. 2021 Half your face is worth a few percent. The thing you have been auditing is smaller than that.

So the move is not more auditing. It is the signals that survive a quick read — expression, grooming, framing, whether the presentation hangs together. If asymmetry is the worry, do people notice facial asymmetry beats “is this side perfect?”

The objection is real: one feature can change a whole impression. Eye-tracking does not make that go away. It shows the change is read through a compressed pattern, not an itemised checklist.

What eye-tracking cannot tell you

It tells you where the eyes landed, not what the person concluded. A fixation is an eye position, not a measure of attention, feeling or attraction.

The usual laboratory limits apply: small samples, static faces on a screen, fixed starting points, sometimes a chin rest instead of a conversation. The “two fixations suffice” result rested on 8 male and 8 female students at one university. Hsiao & Cottrell 2008

The lab can also invert its own finding. When a real person sat in the waiting room instead of appearing on video, participants fixated that person less often than a blank baseline object. Laidlaw et al. 2011 Everything above describes how people look at photographs. That is not how they look at you.

Treat heat maps as sampling strategy, not a map of a private mind. Our attractiveness research methods glossary covers the gap between what a measurement supports and what a headline claims.

What we could not verify

There is no universal first-look point, no universal scan duration and no single feature that decides attraction. What exists is task-, culture- and context-specific measurement.

Three claims we could not trace to any primary source: that 8.2 seconds of gaze means someone is in love (PubMed returns one unrelated paper); that a first impression takes 7 seconds (0 PubMed hits — the measured numbers are 100 ms and 39 ms); and that the brain processes a face in 13 ms (a real study, but it tested named scenes in rapid serial presentation, not faces). The much-quoted “43% of looking time on the eyes” traces to a 1978 study of 16 men that reported about 40%.

How to cite this page

Real World Appeal, "Where people actually look at your face: what eye-tracking data shows," realworldappeal.com, September 2026, https://realworldappeal.com/blog/eye-tracking-face-statistics. Research links sit in the text; data checked 21 September 2026.

The bottom line

People sample near the eyes, use very few fixations, and reach a social impression fast. The live numbers are not the lab numbers, and neither set justifies treating your face as a checklist.

The Two-Sample Rule: people do not scan your face; they sample it near the eyes and decide from a small slice.

If you want an outside read on your own first impression, the free Real World Appeal test gives structured perspective — not a clinical instrument, and not a score.

Sources

• Peterson MF & Eckstein MP (2012). Looking just below the eyes is optimal across face recognition tasks. Proc Natl Acad Sci U S A 109(48):E3314-23 Linked source
• Hsiao JH & Cottrell G (2008). Two fixations suffice in face recognition. Psychol Sci 19(10):998-1006 Linked source
• Janik SW, Wellens AR, Goldberg ML, et al. (1978). Eyes as the center of focus in the visual examination of human faces. Percept Mot Skills 47(3 Pt 1):857-8 Linked source
• Rogers SL, Speelman CP, Guidetti O, et al. (2018). Using dual eye tracking to uncover personal gaze patterns during social interaction. Sci Rep 8(1):4271 Linked source
• Willis J & Todorov A (2006). First impressions: making up your mind after a 100-ms exposure to a face. Psychol Sci 17(7):592-8 Linked source
• Hessels RS (2020). How does gaze to faces support face-to-face interaction? A review and perspective. Psychon Bull Rev 27(5):856-881 Linked source
• Or CC, Peterson MF & Eckstein MP (2015). Initial eye movements during face identification are optimal and similar across cultures. J Vis 15(13):12 Linked source
• Blais C, Jack RE, Scheepers C, et al. (2008). Culture shapes how we look at faces. PLoS One 3(8):e3022 Linked source
• Arizpe J, Kravitz DJ, Yovel G, et al. (2012). Start position strongly influences fixation patterns during face processing: difficulties with eye movements as a measure of information use. PLoS One 7(2):e31106 Linked source
• Hoffmann A, Schiestl S, Sinske P, et al. (2024). Sharing and Receiving Eye-Contact Predicts Mate Choice After a 5-Minute Conversation: Evidence from a Speed-Dating Study. Arch Sex Behav 53(3):959-968 Linked source
• Noyes E, Davis JP, Petrov N, et al. (2021). The effect of face masks and sunglasses on identity and expression recognition with super-recognizers and typical observers. R Soc Open Sci 8(3):201169 Linked source
• Laidlaw KE, Foulsham T, Kuhn G, et al. (2011). Potential social interactions are important to social attention. Proc Natl Acad Sci U S A 108(14):5548-53 Linked source
• Mehoudar E, Arizpe J, Baker CI, et al. (2014). Faces in the eye of the beholder: unique and stable eye scanning patterns of individual observers. J Vis 14(7):6 Linked source
• Bar M, Neta M & Linz H (2006). Very first impressions. Emotion 6(2):269-78 Linked source
• Guo K, Tunnicliffe D & Roebuck H (2010). Human spontaneous gaze patterns in viewing of faces of different species. Perception 39(4):533-42 Linked source