Speaking of Psychology - Encore - The people who never forget a face, with Josh Davis, PhD, and Kelly Desborough

Episode Date: December 28, 2022

Super-recognizers have an extraordinary ability to recognize faces—they can pick faces they’ve seen only briefly out of a crowd and can recognize childhood acquaintances they haven’t seen in decades. Josh Davis, PhD, a professor of applied psychology at the University of Greenwich, and super-recognizer Kelly Desborough discuss the origins of this ability, why you can’t train yourself to be a super-recognizer, how super-recognizers compare with facial-recognition algorithms, and why police departments and security organizations are interested in working with super-recognizers. Learn more about your ad choices. Visit megaphone.fm/adchoices

Speaking of Psychology is taking a winter break, so we're rerunning one of our favorite episodes from the past. Last January, I talked to psychologist Josh Davis and super-recognizer Kelly Desboro about what it means to be a super-recognizer, someone with an extraordinary ability to recognize faces. We hope you enjoy this episode from the archives. Speaking of Psychology, we'll be back with new episodes in January. Thank you for listening. Last year, we ran an episode about face blindness. People who are faceblind have trouble recognizing the faces of even their closest friends and family members. They're at one end of a spectrum of facial recognition ability, and today we're going to talk about the people at the other end of that bell curve, the super recognizers or people who never forget a face.

Scientists only began to study super recognizers a little over a decade ago, and since then, they've been. learned a great deal about this extraordinary ability. They've also begun to explore, together with police departments and other security organizations and businesses, how super recognizers might contribute to police investigations and other security work. So how well can super recognizers remember faces? Can they really recognize a person whom they saw for only a few minutes many years ago? Are super recognizers as good as computer facial recognition algorithms? And can you train yourself to be a super recognizer or at least improve your own facial recognition ability? Or is this something you have to be born with? Welcome to speaking of psychology, the

flagship podcast of the American Psychological Association that examines the links between psychological science and everyday life. I'm Kim Mills. We have two guests today. Our first is Dr. Josh Davis, a professor of applied psychology at the University of Greenwich in London. Dr. Davis began his research career studying facial recognition ability and eyewitness identification. Since 2011, he has focused mainly on understanding super recognizers exceptional abilities in this area. He's developed screening tests to help better identify super recognizers, and he works with police departments and other organizations around the world to explore the ways in which super recognizers can contribute to security work.

Our second guest is Kelly Despera. Ms. Despera has always had an extraordinary ability to recognize and remember faces, but she didn't know that she was a super recognizer until about four years ago, when she took the online super recognizer screening test and received the highest score that researchers had ever seen. Today she works for a company called Super Recognizers International, which trains and employs Super Recognizers to assist with everything from murder investigations to security at sporting events. Thank you both for joining us today. Hi, Kim. Nice to see you.

Hello. Thank you very much for inviting us. It's very exciting. All right. So we're going to go with first names here just to keep it on a little informal basis. So, Josh, let's start with you. How do you identify and define a super recogniser and how good does someone have to be at recognising faces to qualify? On the whole, we've sort of set a rather arbitrary standard for super recognition that people should be able, to be in the top 2% of the population, achieving scores, should I say, on four primary criteria. Short-term face memory, long-term face memory, simultaneous face matching, which is the art of passport officers checking identity documents that the person in front of them is the person they say they are, and also spotting faces in a crowd. And we expect super recognises to be in the

top 2% on all four criteria. Kelly, did you always know this was something that you were very, very good at? How did you first recognize that you had an extraordinary ability that everybody else didn't have? I had no idea. I had no idea that I was doing anything different to anybody else. So I imagined that people could recognize a person they hadn't seen for 30 or years that they'd only seen for a little while or just noticed someone from a small part of their face. It's just, yeah, I thought everyone was doing the same thing. And I was reading a crime novel, and in that novel, the story tells of the detective who calls on the Super Recogniser unit at Scotland Yard to send them some help. So I googled the term Super Recogniser and found one of

Josh's tests, did the test, and the rest is history. So yes, I had no idea at all that I had this skill, and it turns out that it was something quite rare. And do you have any examples of, from your daily lifetimes that you've recognized somebody from long ago or totally out of context? I suppose there's two ways to explain it. One is somebody you know vaguely and somebody you don't know. And that's the really big one is the don't know person. But the example I would give is I was in a local town at a parade. And I just, in a big crowd, I happened to look over my shoulder and I saw a girl stood a few people behind me.

And I just glimpsed her for a split second. And in that split second, I knew that that was Amber, who is the sister of a girl I went to kindergarten with 36 years prior. And she wasn't even my friend. She was my friend's sister, who I think I saw a couple of times when I went around to play at my friend's house. But it was just that split second. And I worked out that it was 36 years since I would have last saw her.

But again, at the time, I didn't know I was doing anything weird. I thought that was normal. And I would bound up to people and say, oh, my goodness. I can't believe it. It's hit, hi, how are you doing? And they would just look at me blankly as I was a complete loon. And I would say, it's me, it's Kelly, and thinking, you know, they've clearly forgotten who I am and I'm a forgettable type of person. But I guess, I guess an example of the way a super recogniser works that illustrates it best is I was given the image of a known predator that we were looking for at a sporting event.

I looked his face for maybe two or three seconds and that imprinted it in my mind and I went into the crowd. And on that day in that crowd, there were 42,000 people there and I was able to find him and alert security to his presence there. So that's what a super recogniser can do. That just sounds miraculous. So Josh, what do we know about how super recognizers are able to do this? Do we have clues from brain imaging research or other studies? We think, first of all, that's probably inherited. At least face recognition ability seems to have a genetic component because if you run twin studies, as some researchers have done, they find that identical twins tend to score far more similarly on face recognition tests than non-identical twins. And as of course identical twins share 100% of their genes and non-identical twins share less of their genes. gene, therefore that seems to illustrate that we do have a genetic component. And in fact,

we have done some parents and child testing, and we do find a correlation between a parent's ability and their child's ability and super recognisers score slightly higher than, sorry, the children of super recognises tend to score higher than the children of controls. So there's that genetic link. there does seem to be from face recognition something to do with your experiences as well. And a very obvious example here is that most people are better at recognising faces of their own ethnicity than of other faces. And it seems to happen with ages versus other ages as well. And we've found that we had one super recogniser in a study in particular.

we asked super recognisers to recognise baby faces. It's sort of the opposite to what Kelly is describing. And we found that adult super recognisers were better than controls at recognising baby faces. But in our sample, we coincidentally had a doctor who worked with babies all the time. And her scores were the only ones who were actually higher on the baby face test than the adult face test because she was used to faces. So there is something driving it with experience as well. And that seems to be the key factors. Kelly, you talked about recognizing somebody you hadn't seen since kindergarten, who clearly was now an adult.

Is that common for you that you know somebody as a child and you see them 35 years later? you know it's the same person, even though they've radically changed? Yeah, it appears that time doesn't seem to really play a part in this. The facial features must say similar enough for me to be able to recognize it, but there can be 20, 30 more than that years apart, or they were a child, now they're an adult, and I can still tell them apart. So, yeah, I can also recognize people who've drastically changed their appearance, so they might have facial hair, completely different hair style, different hair colour. So it seems like

the facial features will always remain imprinted and I can pick them up. Can I add to that? I mean, the very first research on super recognisers used before they were famous test. This was by researchers in Harvard. And effectively, they showed famous people's photographs when they were children. And The super recognisers are always far better at doing that type of test. We quite often use that as a demonstration when I'm giving talks to super recognises or other people. And you'll show, I don't know, a child image of, say, Britney Spears. And the super recognises, they've almost got their hands up almost immediately identifying Brittany far quicker than everybody else.

And, you know, people with average ability quite often just sit there going, no one can do this because it's impossible. So something is happening. And we've used an unfamiliar type of project as well. Well, you'll show an adult face with three babies. One of them is the adult. And super recognisers tend to score higher on that one. And yet they've never met the babies or the adults. So, Kelly, are you aware of what you.

you're looking at when you look at a face? Do you focus on anything in specific or you just get like a boom image of the whole face? Yeah, the latter. I'm not looking at details. I just know. It's a split second. It's a decision that just pops into my head. So yeah, I don't know how the science works and Josh might be able to elaborate or I think they're still looking into it, right, Josh? But I feel like I see the whole. I see the shape. I see a shape. So I can get people from the backs of their heads too. So that tells me that I'm seeing a shape of someone rather than the facial features. And also I can tell them from just a small part of their face or if they're wearing a mask, I can tell them from their eyes. So the masks haven't caused a problem in identifying people.

So throughout the pandemic, it hasn't impaired your ability to basically do your job, right? No. No, it's not as easy. Don't get me wrong. But there's still that instantaneous recognition if you do see the person you're looking for. I could add to that that we have done research. We found super recognisers were better than controls at recognising faces with face masks on, whether they were familiar with them or unfamiliar. And something else that Kelly said that strikes me was that we have done research using EEG as well, measuring brain activity. And we also found that super recognisers seem to have a greater spike of activity very early in face processing

when they saw a face that was familiar to them in comparison to controls. And that might link to what Kelly said about sort of instantaneously being able to recognize as someone familiar to them as well. But that does need some replication. It was published in Cortex, which is quite a nice journal to get published in. So you know which part of the brain or parts of the brain are involved in recognition at this point, Josh? That particular research was EEG, which measures activity. And if you like, it's not very good at geographically mapping the areas in the brain that is active at the time.

But other people have done fMRI studies, which is far more accurate. at measuring the brain. And there's an area that seems to be associated with face recognition called the fusiform gyrus. And when that area gets damaged, for instance, people often develop prosopagnosia as you introduce the topic and become face blind. So there is a lot of evidence from fMRI as well. It's more active when people are viewing faces.

So, Kelly, I have to ask you a question on behalf of our producer. Louer Lee Wynerman, who is the mother of identical twins. Do identical twins look different to you all the time? Completely. Yeah, I have identical twins in my family. And I spent a lot of time with them when they were younger. So when they were toddlers. And I saw them as two completely separate faces. And again, I wish I'd have known why I was seeing that at the time. But no, they just, again, in a glimpse or a side view, they would just, I would know. which one was Kirsty and which one was Lacey. And yeah, although I fully appreciate how similar they look, obviously, but I can completely tell them apart. They're just, the differences are glaringly obvious to me. Well, let me ask you both, this one,

do super recognisers have better than average memory in any other areas of life? Well, I can answer that. We have used object memory tests and some other sort of visual processing tests, and they tend to be slightly better than controls at these other tests, suggesting that they do have superior memory for visual things in general, but actually the differences between controls and super recognises far greater when it comes to faces. And perhaps we found voices as well. We've done some recent research, finding that they tend to be slightly better at voices, which might suggest that it's something about human identification, human recognition, rather than purely face recognition.

Kelly's always described her ability to recognize people by other things from the back of the head sometimes. So it's not just about the face. And you just mentioned voice recognition as well. Is that another burgeoning area of research right now that people can just tell who someone is based on hearing their voice? Yeah. So I have a PhD student, Ryan Jenkins, who's been researching voice recognition. We find that people with super recognition ability for faces tend to be better at voices as well, but it's not 100% by any means. It's a small proportion. So that's research is ongoing. One of the things is developing tests to actually measure this ability is always a bit harder than it perhaps you imagine when you start down this PhD progress. So a few minutes ago,

Josh mentioned that there's a genetic component to this. So, Kelly, are there other people in your family who have this ability? I think that my mom, my mother might have it. So she's pulled some stunts that I think that would suggest that she found a phone, an iPhone on the floor in a public garden. And she picked it up and it was somebody was calling it and it showed the face of the person calling and she knew she'd just walk past that person. So she took it back to them. So I said, hmm, mom, I think maybe something's going on here. So yeah, I would say she's a high,

she's got a high potential there. So she hasn't taken the test? She hasn't. She's probably too scared to fail. I'm going to encourage her. So is facial recognition ability something you can train yourself to do or even to just get better at? Some researchers have tried to teach people to be better at face recognition, in particular in perhaps the sort of identity verification role such as passport officers, because that's so important. So perhaps you can improve slightly the ability to perhaps do a face matching task by asking people to focus on features or other aspects of the face, it doesn't seem to improve

face recognition in general, this type of training scheme. So I'm afraid your ability seems to peak at about 30 years of age, keeps improving beyond the teenage years, but then there's a long, unfortunate, slow decline from that point on. Well, then perhaps it's lucky, Kelly, that you also are a professional photographer, which is also a visual field. Do you feel that it's linked in any way to your super recognising ability? And do you only take portraits? No, I do all sorts of commercial photography. So I don't think it's linked, to be honest with you.

There's all sorts of different people from different walks of life, different skills that are super recognises. And I can't see a pattern there. I put a lot of attention to detail in my work. and I am meticulous, and that helps with my job, but I don't see a link. I don't see a link there. Well, speaking of jobs, your other job, as I mentioned in the intro, was that you work for Super Recognizers International doing investigative and security identification work. What kind of organizations do you work with, and what do you do for them on an average day?

Yeah, I work for law enforcement agencies, police departments, security, private security companies, and we do all sorts of things from, as you mentioned, murder investigations, assault investigations, through to live events where we're on foot doing, you know, covert operations, looking for people in crowds. So there's a variety of ways in which we can apply the skill. We can produce facial comparison reports for people who are trying to prove the identity of somebody. So it's very varied

And I'm also working on quite quick moving investigations such as missing people, trafficked and exploited children. I work for the NCPTF, which is the National Child Protection Task Force over in America there, who do an amazing job investigating missing children and rescuing them. They're amazing as a global force that are volunteering together to help. and I'm one of their what they'd like to call mutants. They have a team of all sorts of different skills and I'm proud to say I'm one of their mutants. So yeah, that's the sort of thing we're doing.

And have you found missing children yourself? Yeah. That must be very, very satisfying. It is. I'm able to be given some faces of children that we know are being exploited. We have the predator as well. I had a case where we had a predator and two children linked together. I was able to find links to them online and we picked them up around six hours after I identified them.

So my leads will be given to the police department who are investigating that. So obviously I don't just say it's them, go get them. They will have to do a thorough investigation. So I'm just the lead. and then it saves them days, weeks, months of searching and trawling, I can get them results really quickly, then they can follow those leads. And in that case, it's six hours before they were, the kids were rescued and the predator was arrested.

So, yeah, we're getting some brilliant results. We've had some great results in murder investigations too, where my evidence has helped to narrow down timelines for the police and improve premeditation. So I can create a nice package of evidence for the prosecution. In recent years, we've heard a lot about artificial intelligence, facial identification algorithms, and how they're being used in everything from social media to policing. Well, there are obviously some ethical and legal questions around this, but just as a purely practical question, are super recognizers as good as AI algorithms, or are they better?

So the last big piece of research really on this was done in the United States. and they recruited Super Recognizers, controls, also forensic facial analysts, and the best algorithms at the time, as identified by the National Institute of Science and Technology in the USA. And they found that the Super Recognizer was operating at the same level as the best algorithms and far better than controls. But what they also found, which I think is really key here, is that actually if you combined the decision-making of a super-recognizer with an algorithm, accuracy was even higher. And one of the reasons is because algorithms make mistakes that a human would never make,

and humans make mistakes that an algorithm probably wouldn't make. So therefore, you're removing those errors that are probably unique to both and getting better outcomes. Does AI have the ability that Kelly described, which is she can, can tell people from seeing the backs of their heads? Almost certainly not. I don't think because AI has to be trained with images for it to be able to perform these tasks. And I don't think there are any databases of faces from the back of the head to train AI.

So that's how it works. You have to sort of throw into the mix hundreds, if not thousands of faces for AI to be most effective, a bit like humans. And if you haven't got those images, then humans are going to beat AI every single time. And we can also deal with really low quality images, you know, really pixelated images aren't going to deter us. And also, you know, like I said, masks, facial hair, bits and pieces that might throw an algorithm off.

We kind of override those. So I think I am a great advocate of facial recognition software and I enjoy working with it. I get some of my best results when I team up with guys that use. facial recognition software. So I think putting the two of us together makes a superpower. So is it your view that people who check passports, say, in customs, should we be testing people for their facial recognition ability before they get jobs like this? Absolutely. And they do in some countries. You would never want someone with face blindness, prosopagnosia being a passport officer because they just could not do the job. I'm not necessarily

sure that you need super recognisers for that job, but what you want is people who have above average abilities in those sorts of roles, because they're far more likely to make a correct identification, but they're also far more likely not to make a wrong identification than people with average or lower ability. That's key as well. Kelly, is a company that you work for, advocate for this? I mean, do you consult to, you know, operations that maybe could use some sort of a test for their employees? Absolutely. We team up with lots of different companies to assist.

We work for a company that do assess passport photographs and ID photos. So, yeah, it's definitely something we can apply and help with. And we do identify super recognisers through the testing and then we put them through our own training courses, which we'll teach them all about the law surrounding image use. It will enhance their skills with behaviour analysis as well if we're actually out and about. So we can enhance skills around super recognition too to make a really rounded super recognises that's able to go out into the world and use their skill for good.

So some of our listeners might not be aware that in your country and you're both in England, that there are closed circuit TVs pretty much everywhere, right? You folks are on camera all the time. It's a little different from the U.S. where, you know, we're very sort of hinky about that concept. Are there ethical issues surrounding what it is that you do, for example, trying to find criminals? I mean, do people in England feel like their privacy has been invaded? So they do worry about privacy in terms of computerized face recognition systems, definitely. But I don't think anybody worries about super-recognized.

It's because, if you like, they epitomise what an old-fashioned police officer should be able to do. They know all the criminals. You see it on cop shows. There's nothing strange about super recognises within policing identifying criminals and people like that. Everyone expects it. It's the algorithms they are more suspicious of, definitely. And it's a way of life for us being on camera. I would far rather that my whole country was covered in cameras.

And if something happens to me or one of my loved ones, that it's. going to get picked up and we've going to have some evidence to work from the number of cases that are solved through surveillance footage and, you know, come to a conclusion much quicker and more accurately because of that. I'm a strong believer that we should stick cameras everywhere and I certainly have them all over my house. I don't. For me personally, I'm protective of my privacy, but I do agree that in some areas you do want cameras. You want them in town. centers and airports and places like that, train stations, travel centers, and they should be scrutinized properly.

So my understanding is that the cameras were installed because there was the thinking among law enforcement that they would be a deterrent, but it seems that they're not, that a lot of people will just blithely go about committing crimes because they figure there are tens and thousands and millions of these images. No one's ever going to find me. Do you feel that the public, at least in the UK, is aware of the fact that you have been able to crack a lot of cases? Yeah, I think it's been very important. The super-recognizer successes that have been in the media.

In fact, my PhD was on the way the faces are used in court for identification purposes. And at that time, they weren't at all. There was a lot of cameras around, and nobody was ever really looking at them. It's only later that they started properly scrutinising the images, making sure that they actually try and identify people committing crimes on camera. And that breeds success. More success, you get more criminals identified. And so much so that, you know, my work has transferred to other countries around the world,

especially at the moment we're identifying a lot of super recognises within German police forces, is super recognises in India working in identity verification, creating lots of jobs there for people working in sort of roles, checking identity documents for the government and other organisations. We're not really working in America, the USA, as of yet. But who knows? I'm on my way. I'm coming to get you. So, Kelly, I mean, it sounds like you really appreciate this, this, ability that you have, but I'm wondering if you ever feel that there are drawbacks. I mean, do people try to fool you? I don't see any drawbacks. I think it's a relief now I know I'm not

being stalked constantly because, of course, if you're walking around a shopping mall and you see the same person over and over again, you start to think, I'm being followed by this person, but essentially what's happening is I'm just recognizing the same person in the same shopping mall with me. There's nothing worse than that. So there aren't really any drawbacks. to be honest with you. It's very useful. It's a bit annoying for people who are around me when I keep telling them that that's the guy in the film 20 years ago that was an extra in Indiana Jones and the last crusade.

But no, it's, I embrace the skill. I'm so delighted that I can use it for good, that I'm seeing results that are meaning things to, it means something to families who have lost somebody or who are fighting for justice. So it's just an honour that I'm able to translate this skill into something. think useful for society. So Josh, last question. Where is the research going next? Well, that's a big question. We would like to do some more brain scanning research to try and understand exactly what's going on in super recognises brains. And then perhaps from that, we can

understand how to better apply their skills in different workforces for whatever reason. because we don't really know where we might end up going in the future for the best use of super recognises. I'm probably more of an applied psychologist than a theoretical psychologist, more interested in seeing, you know, the practical aspects work like Kelly does. How else can we use Kelly's skills to, you know, help society in general? Are there places that we're missing the use of super recognition, Kelly? What are you looking for in your company? Are you expanding? Yeah, we certainly are.

Our passion is to help law enforcement. That's what we would like to do is to be able to go to law enforcement agencies and help them to identify super recognizers within their ranks because there are going to be hundreds of them. And it would be fantastic to think that we could identify them, helped train them to know their skill better and for them to be used within those police departments for the better. So that's our passion. Super recognises international would love to get out there and get amongst the law enforcement agencies to help them identify their own. That's where we

think we'd get the best results for society for. Well, I want to thank you both for joining us today. I think you're both doing extraordinarily important work and I thank you for it. Thank you very much. Thanks so much, Kim. You can find previous episodes of Speaking of Psychology on our website at www. www. speakingof psychology.org or on Apple, Stitcher, or wherever you get your podcasts. And if you can, please leave us a review. If you have comments or ideas for future podcasts, you can email us at speaking of psychology at APA.org.

That's Speaking of Psychology, all one word, at APA.org. Speaking of psychology is produced by Lee Winerman. Our sound editor is Chris Condyenne. Thank you for listening. For the American Psychological Association, I'm Kim Mills.