University of Greenwich

October 15, 2013

Download

State of Physical Access Trend Report 2024

Super-Recognisers in the Metropolitan Police

The August 2011 riots in many English cities cast a heavy pall across the country. It was extremely distressing for most members of the public who stayed in their homes to avoid nearly a week of crime-riddled evenings.

There was a strong sense of outrage and demand for retribution. However, unless a suspect was apprehended at the scene, for the police, the thousands of hours of CCTV footage was in many cases the only evidence linking offenders to their lawlessness.

In London, a substantial proportion of all suspect arrests were a direct consequence of identifications from CCTV made by a small group of front-line police officers — so-called super-recognisers. One identified 180 sometimes heavily disguised rioters. As the majority of suspects confess when confronted by such evidence, these officers were clearly somewhat special.
[mk_mini_callout]

Download the 2016 Video Surveillance Report

Get the latest report on CCTV and video surveillance in 2016 by downloading this free report[/mk_mini_callout]

Some super-recognisers had already come to the attention of the Metropolitan Police Service due to a steady stream of previous CCTV identification successes. As a consequence, psychologists led by a team from the University of Greenwich had tested the facial-recognition and other cognitive abilities of these officers to determine whether they possessed exceptional skills.

The super-recognisers’ scores on these tests were compared to a control group of over 100 members of the general population. This research revealed that some officers possessed extraordinary facial-recognition abilities shared by 1 to 2 per cent of the population. Their performances also met the scientific definition of super-recognisers established by researchers at Harvard University in the USA.

What makes a super-recogniser?

What makes super-recognisers so special? How can their abilities be best put to use? Science currently knows very little about super-recognition, although there is some evidence that it may have a genetic origin and is therefore innate. You cannot teach someone to become a super-recogniser. Those tested by the University of Greenwich were in particular superior at spontaneously recognising very poor-quality 12-year-old images of celebrities no longer in the public eye. This test has parallels to police officers identifying suspects from unclear CCTV footage, even if they had not seen those suspects for many years.

The super-recognisers were also superior at quickly learning the identity of previously unfamiliar faces, and subsequently recognising that face depicted in a different photograph within an array of highly similar faces. For most people, this task can be error prone.

Interestingly, the super-recognisers were most accurate at spotting that arrays did not contain target faces. This demonstrates that as well as being superior at recognising that they have seen a face before, super-recognisers are also excellent at recognising when they have not seen a face before. This is extremely useful to the police because it suggests these super-recognisers are far less likely to provide a false lead. Thereby they save vital time and resources.

As a consequence of their operational successes and the psychological tests, super-recognisers in the Metropolitan Police are now the first to receive images of the most serious crimes, as well as those matching their operational expertise. For instance, robbery squad super-recognisers view images of robbery offences. These changes to CCTV image distribution procedures have enhanced identification rates.

Other forces could probably utilise their own super-recognisers in a similar manner. Indeed, there may be 2,000 to 4,000 so far unidentified super-recogniser police officers across the UK. There may also be many super-recogniser CCTV operators or other security workers whose skills have so far not been fully utilised.

Of course, super-recognisers can only identify suspects they have previously encountered. Other officers or members of the public who possess average facial-recognition ability should always be encouraged to view CCTV images. Regardless of facial-recognition ability, most humans can reliably identify those familiar to us even in poor quality images — we may all be able to contribute to a police investigation.

Research at the University of Greenwich into super-recognisers from the police and other members of the public is ongoing

Free Download: The Video Surveillance Report 2023

Discover the latest developments in the rapidly-evolving video surveillance sector by downloading the 2023 Video Surveillance Report. Over 500 responses to our survey, which come from integrators to consultants and heads of security, inform our analysis of the latest trends including AI, the state of the video surveillance market, uptake of the cloud, and the wider economic and geopolitical events impacting the sector!

Download for FREE to discover top industry insight around the latest innovations in video surveillance systems.

VideoSurveillanceReport-FrontCover-23
Subscribe
Notify of
guest
19 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
holmesd
holmesd
October 15, 2013 5:02 am

It’s good to know that all advancements aren’t necessarily technological. Making the most of a person’s skills and finding out if more people possess these skills is the best way to get the most out of security systems and personnel. After a shakey start on the trial, I got them all correct, so feeling quite smug!

Rob Ratcliff
Rob Ratcliff
October 15, 2013 5:44 am
Reply to  holmesd

A super-recogniser in our midst holmesd? You did much better than I did. It’s really interesting stuff, and I would encourage everyone to try the online test, which is just for fun.

Philclark
Philclark
October 15, 2013 5:46 am

Great stuff Josh.
i have a few questions:

– I presume the super recogniser police officers are given a gallery of suspects ahead of viewing the CCTV footage. is this how it works?
– Is there a margin for error for super recognisers? can they get it wrong at any point?
– what are the police doing to communicate this across forces to ensure super recognisers are, er, recognised? In addition it would seem sensible for the police to include tests when recruiting new entrants to the force

Rob Ratcliff
Rob Ratcliff
October 15, 2013 7:11 am
Reply to  Philclark

For the police, they’ll look through CCTV images and they’ll look through images in a gallery and simply remember the faces. Then, the next time they see that face they’ll know, almost magically. For instance, there might be a picture taken from CCTV images of a person vandalising a property with a mask on their face. Then later, that same person might be watching unrelated images and notice the same person’s face, even though they’ve only previously seen, say, their eyes. That was my understanding from watching Josh’s recent presentation, anyway.

holmesd
holmesd
October 15, 2013 7:16 am
Reply to  Philclark

It would be good to see this kind of testing for police and maybe private security as Phil has mentioned

jfeeney4
jfeeney4
October 15, 2013 12:08 pm

It is always refreshing to see how with “all” the technology we possess and use, it’s still a human that can interact and bring concepts to a new level.
Super-recognition – a hidden talent which by all accounts, one can not exactly be trained to do.  Just find it discouraging, when humans enter the equation – the level of expected “error” takes a sharp turn North.

Dr Josh P Davis
Dr Josh P Davis
October 15, 2013 12:49 pm
Reply to  Philclark

  In their jobs the super-recognisers in the police will mainly view the CCTV images of recent crimes that have been published (internally or externally) and then if they have encountered that individual before (perhaps from an arrest or sometimes occasionally from having seen them in previous CCTV images) they will file a report as to the identity of that person. There are procedural rules to ensure that there is no element of collusion. These are therefore spontaneous identifications – we could all do this if for instance our next door neighbour was depicted. The vast majority of suspects (but… Read more »

Rob Ratcliff
Rob Ratcliff
October 16, 2013 7:46 am
Reply to  jfeeney4

Well, indeed human error will always be a factor, but what a human (and a super recogniser specifically) brings to the table is the ability to link two seemingly separate pieces of information (in both cases a face though). Incredible.

holmesd
holmesd
October 16, 2013 8:55 am
Reply to  Rob Ratcliff

If we look at this and the article about how biometrics only resulted in one arrest during the riots, even taking into account the potential for human error, it may be that these super recognisers would have performed much better than the biometric kit did.

safeNsane
safeNsane
October 17, 2013 7:47 am
Reply to  Rob Ratcliff

This is one reason I don’t see computers taking over every thing.  Sometimes that human factor makes all the difference.  Maybe it’s a behavior that catches a human’s eye or the way they walk, stand or a mannerism that a computer just can’t see.  It does make me wonder if they will do more testing to get more of these super recognizers working on the CCTV systems or just walking the streets.

holmesd
holmesd
October 17, 2013 8:39 am
Reply to  safeNsane

I think the idea to to use them in conjunction with CCTV, using them to try and identify people caught on CCTV, as opposed to them just happening upon someone?? There’s a statistic about how many police on the beat actually encounter crimes- it’s not that many

safeNsane
safeNsane
October 18, 2013 7:10 am
Reply to  holmesd

This is kind of along my line of thinking.  I don’t suspect that the super recognizers will spot crimes in process but maybe if they have been show images of people in the area that they should be looking out for that they would be more apt to recognize them and bring them in for questioning.  

Sheh
Sheh
October 18, 2013 10:11 am
Reply to  jfeeney4

I agree with you there that its basically humans which actually changes the equation. But I am still having my doubts on the true utilization of super recognisers. I support any action taken in its true perspective to provide security but giving full weigtage to those recognisers and believing what they are saying is 100% authanticated is a bit out of proportion. I feel taking CCTV footage and analyzing it is a good option but only where picture can be easily identified by many and not on assumptions/authentication of few.

Sheh
Sheh
October 18, 2013 10:14 am
Reply to  Rob Ratcliff

I have just one question. Who will authanticate the super recognizers claims. Will they be the final verdiact givers?

Rob Ratcliff
Rob Ratcliff
October 22, 2013 12:01 pm
Reply to  holmesd

From the presentations I saw at the Met last month, I can definitively say super-recognisers were responsible for something in the order of hundreds of recognitions, comparted to (how many was it) a few?

Rob Ratcliff
Rob Ratcliff
October 22, 2013 12:04 pm
Reply to  safeNsane

The better the tech gets the more that we’re freed up to think on higher things. eg. ANPR systems: the people that used to provide intelligence on where certain vehicles (such as one owned by a suspect) would be at certain times don’t do that anymore, because an intelligent ANPR system has identified the trends to say their most likely location. The intelligence officer is miffed, thinks he’s being replaced, but he’s not. Now he can spend more time actually interpreting and finding additional intelligence and patterns a computer can’t see.

Rob Ratcliff
Rob Ratcliff
October 22, 2013 12:30 pm
Reply to  holmesd

Well, that’s it eh? Bobby on the beat is a deterrant. In the week after the riots stopped crime fell dramatically across London because there were over 16,000 officers on the street. Incredible that after such epic disturbances the streets were quieter than ever.

Rob Ratcliff
Rob Ratcliff
October 22, 2013 12:31 pm
Reply to  Sheh

There are tests that people like Josh can do to identify them, but usually they will be self-identified. The police may compare their identifying stats to the peers and come to a conclusion as well, I suppose.

batye
batye
November 1, 2013 12:59 pm
Reply to  Rob Ratcliff

interesting point, but technology is grow and it changing… I did see presentatition, of new technology where people could be identified by computer the way the walk, posture, gestures and e.t.