By Alex Fradera
Identifying people from video and photographs is a core task for a modern police force, and London – which led the world in implementing and using CCTV – has attempted to meet this need by developing a pool of 140 “police identifiers” made up of Metropolitan Police officers with a strong track record of making IDs from photographs. But who are these individuals? Are they really super-recognisers as the Met has claimed? True super-recognisers are usually identified by formal tests and their dramatic ability to recognise human faces outstrips typical performance to the same extent that many prosopagnosics (people with face-blindness) lag behind. Or, in identifying so many suspects, did the police identifiers just catch a string of lucky breaks? Addressing this through a battery of neuropsychological tests, Josh Davis and his UK-university collaborators scrutinise the scrutinisers in a new paper in Applied Cognitive Psychology.
Davis’s team had a benchmark group of verified super-recognisers – 10 members of the public who had shown outstanding performance at standardised face recognition tests in a mass testing exercise completed by hundreds of people at London’s Science Museum. Thirty-six members of the police “identifiers” pool had to prove their mettle against these super-recognisers and against a control group of 143 members of the public with normal face-recognition abilities.
And prove it they did. Police identifiers were as good as the super-recognisers (and better than the controls) at picking a briefly-viewed face from a subsequent line-up, or spotting its absence, even when the target face was first studied at a different angle. They were as good at distinguishing very similar faces from one another. They were as good at picking out subjects seen once previously in a normal setting, when they were only provided with limited information – the internal features of eyes, nose and mouth. And they were as good as the super-recognisers at a “famous faces” test, naming or providing factual details about famous people and accurately rejecting faces that were not famous.
One possibility is that police identifiers appeared to perform so well simply because of a response bias – for example, having a greater tendency towards saying a face is “familiar” (liberal response) or “unfamiliar/don’t know” (conservative). But the researchers checked this and the police identifiers were no more zealous than the control participants in pursuing identifications, true or false, they were simply more accurate.
Taken together, the results suggest that the Met’s police identifiers group is indeed composed of superior ability individuals who perform at a true super-recogniser level. We shouldn’t treat these individuals as if they were magical; none achieved a perfect score across every test, and they are still likely to be susceptible to the same confirmatory biases that plague all of us, such as being swayed by their pre-existing prejudices. But almost all scored above the control mean on every test, and as a group they show extremely impressive ability.
These officers were selected into the Met’s identifier unit based on their outstanding results on the job. A problem with this approach is that know that past performance, containing as it does instances of fortune as well as merit, is not the best indicator of the future. But ability on standardised lab tests is, and this group has it (or at least the members tested here did). This research therefore supports the claim that the police identifier population is “an important investigative tool in the armoury of law enforcement” – one which forces and services in other countries would do well to consider.