Why Some People Never Forget a Face — New Study Reveals the Visual Strategy of 'Super-Recognizers'

Some people never forget a face — now we know why

A new study from researchers at the University of New South Wales (UNSW) in Sydney finds that so-called super-recognizers don’t remember faces by trying harder; they simply look at faces differently. Rather than scanning a face broadly or fixating on the center, super-recognizers instinctively zero in on the most distinctive, diagnostic facial features.

"Their skill isn't something you can learn like a trick. It’s an automatic, dynamic way of picking up what makes each face unique," said lead author James Dunn, a psychology researcher at UNSW.

How the study worked

The team tested 37 super-recognizers and 68 people with typical face-recognition ability. While participants viewed images of unfamiliar faces on a screen, the researchers recorded eye movements with eye-tracking equipment to map where and for how long people looked at different facial regions.

Those eye-movement patterns were then provided to deep neural networks trained for face matching. When the algorithms used gaze patterns from super-recognizers, their face-matching accuracy improved compared with when they used data from typical observers.

What the findings suggest

The authors argue these results point to individual differences arising very early in visual processing—possibly even at the level of retinal encoding. The study builds on earlier work from the same group showing that super-recognizers tend to decompose a face into parts (a "jigsaw" approach) and then integrate those parts, rather than treating the face as a single, undifferentiated whole.

Rather than simply gathering more information, super-recognizers prioritize features that carry stronger identity "clues." As Dunn puts it, it’s like a visual caricature: exaggerating distinctive features makes a face easier to recognise, and super-recognizers appear to do that instinctively by tuning to the most diagnostic facial elements.

Implications and context

The findings could inform improvements to artificial facial-recognition systems by showing which visual patterns are most informative. However, the researchers note that humans still often outperform AI in social settings because we integrate additional cues such as movement, context, and social signals.

There is also evidence for a strong genetic basis to exceptional face memory, and because facial identity processing plays a key role in primate social behaviour, the biological roots of this skill may not be uniquely human.

Publication: The study appears in the Proceedings of the Royal Society B: Biological Sciences.