Some people have an extraordinary ability to remember faces. They can, for instance, easily recognise or recall unfamiliar people they stood behind in a queue or saw on social media.
Scientists have now devised a more sensitive facial recognition test to help tease out the best of these “super-recognisers”, who represent just 2% to 3% of the population.
Nikki became aware of her talent when she told her family she’d seen the same woman three times around Rome and they thought she was crazy. The test confirmed she wasn’t.
“It’s a strange ability that, at first, seems absolutely useless,” she says. “But it’s quite delightful being able to see things other people can’t see.
“I love being able to list the films in which I’ve seen actors and actresses play very small or obscure parts, and going to parties and remembering people I’ve only seen for a few minutes many years ago. It’s a minor super-power, but one that contributes to who I am.”
Others can find it a bit awkward. “I often have to lie that I’ve never met or seen people before,” says Sallie. “It freaks them out. ‘Oh I saw you last week at Woolworths’ makes you sound like a stalker.”
It’s a unique skill, and although it seems to have a hereditary basis, it doesn’t otherwise link with other consistent traits such as intelligence. And it’s different to having a photographic memory.
“We hear many anecdotes of super-recognisers recognising faces of people they haven’t seen for years or recognising them when their appearance is vastly different, like in costumes in movies or in baby photos,” explains James Dunn from Australia’s University of New South Wales, lead author of a paper in the journal PLOS ONE.
“Similarly, to do well on our test and be a super-recogniser you need to recognise faces despite dramatic changes in appearance. This means that what they are doing is more than simply remembering the faces like a photo.”
Unlike existing tests, the new test is designed to discriminate between the best performers and find the upper limits of ability. Indeed, no-one has achieved a perfect score from more than 24,000 participants to date.
“We found that on other tests it was not uncommon for super-recognisers to score perfectly,” says Dunn. “This led us to wonder about the limits of their ability – just how good could the best super-recognisers be?”
“To separate the best from the rest, we needed to devise an extremely challenging test. Our test is difficult because it requires people to recognise faces despite substantial changes in appearance from one encounter to the next.”
Simulating real life conditions also improves the test’s ecological validity.
To create it, 40 images were carefully selected from a database of more than 200 undergraduate students to represent a range of different cultural backgrounds and including “target-foil” pairs – those most similar in facial appearance to each other.
For each person, five images were chosen – one a studio-quality pic with optimal lighting and a neutral facial expression, as used in existing tests, and the others taken from social media accounts with uncontrolled factors including different poses, expressions, lighting conditions and image quality.
As expected, given strategies to recruit super-recognisers with media headlines such as “So, you think you’re good at recognising faces”, a high proportion of participants performed in the top 5% across the samples – above two standard deviations from the mean.
Notably, 35 achieved scores in the top 0.3%, or three standard deviations from the mean, breaking through the ceiling of existing tests.
“What we have learned from analysing this test is that while all super-recognisers are exceptional at face recognition, some are better than others,” says Dunn. “We hope that with more people completing the test we can find the Einstein of face recognition.”
Other advantages of this test, which the researchers say is designed to supplement rather than replace existing tests, include its capacity to be delivered online to large numbers of people, and thus net a larger pool of super-recognisers.
Unlike other tests, it has a separate initial screening that prevents people from accessing the actual test, thus preventing participants from practising it repeatedly.
It also tests general facial recognition that includes both memory and matching ability, each targeted separately by existing tests yet highly correlated. These include, for instance, detecting a memorised face from CCTV surveillance or matching photo ID to the real person.
The researchers hope this will help shed more light on the neurology of visual recognition abilities, of which face recognition is the gold standard.
“People vary a great deal in face recognition ability – from high scoring super-recognisers to people with ‘face blindness’ that struggle to recognise friends and family,” says Dunn.
“We don’t know the cognitive and perceptual differences that give rise to these stark individual differences in ability. Unlocking this mystery can help us understand how visual expertise is developed in humans and can help to improve face identification in important applied settings.
The researchers invite people to go online and put their face recognition powers to the test.
Originally published by Cosmos as How well can you identify faces?
Natalie Parletta is a freelance science writer based in Adelaide and an adjunct senior research fellow with the University of South Australia.
Read science facts, not fiction...
There’s never been a more important time to explain the facts, cherish evidence-based knowledge and to showcase the latest scientific, technological and engineering breakthroughs. Cosmos is published by The Royal Institution of Australia, a charity dedicated to connecting people with the world of science. Financial contributions, however big or small, help us provide access to trusted science information at a time when the world needs it most. Please support us by making a donation or purchasing a subscription today.