Facing off: the secret lives of super-recognisers

Like many super-recognisers, Senior Sergeant Chris Tritton has learned when to keep quiet about his extraordinary ability to place a face.

“It gets awkward – it’s easier to not say something,” he explains.

Before he realised his facial identity processing ability was anything special, he’d been people-watching in the airport terminal while waiting for a flight.

“A couple of weeks later, a guy came to my sister’s house while I was over there,” Tritton says.

“And I said, ‘Oh, you were on my flight to Sydney. You were sitting on this seat and working on your laptop.’ I thought it was an interesting icebreaker, but he was very uncomfortable.”

Yet Tritton’s uncanny ability has proven an asset in policing.

From the sunny Surfers Paradise Station, where I catch up with him, he now heads up the ‘Super Recogniser Network’ within the Queensland Police Service (QPS).

The 20-strong group, identified through an optional online test, is made up of police officers who, like him, are super-recognisers.

Spread out around the state, they come from different areas, including water police, patrol services and criminal investigation. They remain in their substantive positions but take time out of each shift to conduct identifications.

Queensland is the first Australian state to use super-recognisers in this way, although there are established units overseas, such as in the UK.

It’s anticipated that super recognisers’ freakish face identification skills will prove vital during the upcoming Brisbane 2032 Olympic and Paralympic Games where, using intelligence tips, they will be deployed to identify potential threats among the crowds.

Super-recognition: An accidental discovery

Super-recognisers were first identified by Harvard University and University College London researchers who were conducting research into prosopagnosia.

Prosopagnosia, also known as face blindness, occurs when people have difficulty recognising familiar faces and, sometimes, even their own.

Following media coverage of their work, the researchers were contacted by people who believed they were the opposite of prosopagnosic.

The researchers noted that, just like Tritton, these individuals commonly altered their social behaviour to accommodate the fact that they frequently recognised other people who did not recognise them.

“All describe situations in which they correctly recognised near strangers whom they had not seen for years who had since undergone major changes of appearance (such as) growing from a child into an adult or adopting a radically different hairstyle,” the researchers wrote in the Psychonomic Bulletin & Review in 2009.

(Tritton once recognised a 25-year-old woman whom he hadn’t seen for almost two decades. She had been a six-year-old child living in the same street as him.)

Tritton once recognised a 25-year-old woman whom he hadn’t seen for almost two decades.

The first academic paper on super-recognisers recorded that when four of these people were put to the test, they performed “far above average” on two different tests of face recognition – the Before They Were Famous (BTWF) Test and the Cambridge Face Memory Test (CFMT).

“[They] are about as good at face recognition and perception as developmental prosopagnosics are bad,” the researchers noted.

It suggested that prosopagnosia, at least in some cases, might not be a neurological disorder, but rather, represent the lower limit of normal variation.

It’s now generally understood that facial recognition abilities are normally distributed, so while most people are average, one paper suggests that the top 2-3 per cent are exceptional (super-recognisers).

A similar number at the other end of the bell curve are poor (prosopagnosics).

Practical applications

In a world of artificial intelligence and automation, it’s strangely reassuring to know that super-recognisers still outperform computers when it comes to catching crooks.

Tritton plays CCTV footage which shows an older man with white hair and a white beard rummaging through a chest of drawers in a private room in an aged care home at Brisbane’s Chapel Hill.

The security camera had earlier been installed by the family of the woman who occupied the room.

“This guy was going through all the rooms and rifling the contents … the victims were up to 104 years old,” Tritton says.

Frames grabbed from the footage had been fed into a new QPS-developed specialised facial-recognition software called ‘QFACE’, which cross-checks the image of an offender with more than 10 million faces stored in the QPS database.

The software recommends up to 300 potential matches, with the super-recognisers then going through them manually to make the final identification.

The actual offender was 50 faces down the computer-generated list of other similar-looking men, but Tritton picked him out straight away, matching the CCTV footage to a photo taken eight years earlier, when the man was clean-shaven.

His technique involves quickly flipping through the list of possibilities until he alights on one demanding further scrutiny.

“Initially you just get a gut feeling,” Tritton explains. “It’s then, ‘OK, so I’ve got that, now I need to support that with some evidence.’”

Super-recognisers undertake further training in facial comparison, so they can support their hunches with these additional points of confirmation.

OK, so I’ve got that, now I need to support that with some evidence.

Senior Sergeant Chris Tritton

He points out the features which helped to make the final identification of the nursing home thief – including even helixes on the outer rim of the ears, a tight lip structure, and deep creases above the centre of his eyes.

“Generally, it’s around the eyes and ears which tend to stay the same throughout our lives,” Tritton says.

After ransacking 17 aged care residents’ rooms, the man was sentenced to a year’s imprisonment.

Since its inception in 2019, the group of super-recognisers has helped to identify more than 1000 offenders for crimes ranging from robberies, burglaries, sexual offences, and domestic violence.

They have also helped to identify vulnerable people – including an unidentified young woman who lay unconscious in a hospital intensive care unit after a car accident.

The officers have even used their skills to crack cold cases and, during the COVID pandemic, to identify NSW-QLD border breachers – who were wearing masks – from surveillance footage.

Underpinning abilities

It’s believed that super-recogniser skills are innate, and highly heritable. But despite a growing body of research, the neural underpinnings remain poorly understood.

One thing is clear though: their superior face processing abilities don’t generalise to similar tasks.

For example, a University of New South Wales (UNSW) study published in PLOS One study in May tested the face identification ability of more than 1600 Australian police officers and found that their skills transferred only partially to body identity decisions (where the face was not visible).

“They were no better than controls at deciding which visual scene that faces had initially been encountered in,” the researchers wrote.

Curious about my own facial recognition abilities, I visited the UNSW Face Test, established as a screening tool for super-recognisers.

Lead investigator at the Face Research Lab at UNSW, Associate Professor David White, has published extensively on individual differences in people’s ability to perform face processing tasks.

Almost 10 years ago, one of his studies revealed that people who match faces for a living – passport officers – wrongly accepted more than one in ten (14%) fraudulent photos.

Anyone can complete the online screener, which is made up of a recognition memory task and a match-to-sample sorting task, with some practice tests using characters from the animated television series The Simpsons.

As I tackle the tasks, I don’t get any of the gut feeling that Tritton talks about.

So, it’s no surprise that my overall score is 65%.

According to the online feedback I receive, this puts me in the top 25% of the population – but a long way from the rarefied ranks of super-recognisers.