Tessa Ndjonkou sits down with Professor John P. Davis from Greenwich University to tackle the phenomenon of super-recognisers.
A new algorithm to improve recognition, processing, and tracing is being developed daily. But before being programmed and coded into artificial intelligence and your smartphone, facial recognition was an inherently human skill, and like all human abilities, it sits on a spectrum. At the lowest end of this spectrum, you will find prosopagnosics or “face blind” people who struggle or cannot recognise faces, including those of close friends and family members. Most studies suggest they could make up to 2.5% of the global population. On the opposite end of this spectrum are “super-recognisers”, who could make up to 2% of the world population. The term was first coined in 2009, and the ever-growing research reveals that they have excellent facial recognition abilities that are above the norm. For example, it is common for super-recognisers to correctly identify people they haven’t seen for decades, even though they’ve only interacted briefly.
Since 2015, a one-of-a-kind super-recognizer police unit has been based in the London Metropolitan police in Scotland Yard. Led by chartered psychologist and professor at the University of Greenwich Josh P. Davis and ex-Detective Chief Inspector Mike Neville, the unit is made up of approximately two-hundred people. It is the most significant functioning super-recogniser or facial analyst task force to date. Although Ireland does not yet have its own super-recogniser brigade, it has already made a noticeable mark in the field. Detective Patrick O’Riordan, an Irish Metropolitan Police super-recogniser based in London, has made hundreds of correct identifications throughout his career. In a 2018 article for the Irish Independent, he details how his unique skill was pivotal in finding the main suspect behind the murder of fourteen-year-old Alice Gross in a matter of days, using CCTV images and plainclothes work.
Assistant professor and Swiss National Science Foundation Fellow Meike Ramon studies the neurological processes underlying this ubiquitous gift. In one of her protocols, she shows subjects a series of visual stimuli while they wear caps with sixty-four integrated electrodes. Using electroencephalography (EEG) and magnetic resonance imaging (MRI), she demonstrated how the fusiform gyrus, located in both the temporal and occipital lobes, registers higher electrophysiological activity than a control group when super-recognisers perform long and short-term super-recognizing and face-matching tasks.
After extensive testing of millions of candidates, the Association of Super-Recognisers realised that some super-recognisers might be extremely good at short-term face-matching tasks but would need to improve at long-term facial recognition tasks, which could be an asset. In an interview with Professor Josh P. Davis from the University of Greenwich, he explained that face-matching is more of a perceptual task than a memorisation one. In over a decade of working with super-recognisers, he realised that some had the innate ability to correctly match CCTV footage of suspects, something forensic experts need rigorous training. Not only could they make quicker associations, but their accuracy was higher.
The progress and efficiency of Artificial Intelligence (AI) questions the very relevance of the existence of a super-recogniser task force. If a machine can scan your passport and your face at customs, have we evolved past the need for in-person checks?
For Professor Josh P. Davis, artificial intelligence is not a deterrent to super-recognisers but their greatest ally in preventing miscarriages of justice. AI and super-recognisers are both available for several reasons, so it is in their best interests to work together and cover the other’s vulnerabilities. Notably, current AI struggles to calculate facial definition and cannot account for additional criteria and other factors super-recognisers use to identify people in the field: their physique, gait or even their head shape. On the other hand, human error is frequently caused by environmental variables such as light, picture resolution and tiredness.
Regarding global security, perhaps the most pressing bias that both AI and humans possess is a racial bias. The "cross race" effect in humans refers to the propensity to be better at identifying faces from our own ethnic group and ethnic groupings with whom we've interacted for the most extended amount of time. Professor Josh P. Davis believes super-recognisers might be a helpful weapon in the fight against police brutality. When systemic racism and inadequate levels of qualifications coalesce, there are miscarriages of justice that can result in disproportionate incarceration and death of Black and Brown people. A super-recogniser task force could perhaps counteract the bias held by facial recognition programs or individual members of law enforcement. However, the United States has shown little enthusiasm to implement such an initiative.
Professor Josh P. Davis and his fellow researchers, students and vetted super-recognisers continue to expand the breadth of knowledge on super-recognisers and the science behind facial recognition as a whole. He hopes for broader collaboration between super-recognisers working in law enforcement and security across different countries. Though the subject is no longer an unexplored and unexplained quirk, it is under scrutiny by industries that do not share the same confidence in human abilities. Programming professionals prefer to place their trust in algorithms and raise concerns about the ethics of having real people carry out identifications. Media exposure of super-recognisers, however, has been predominantly positive and has allowed for significant strides in findings. However, it has also created a mythology around the skill that does not always live up to the reality of possessing this ability. The life of a super-recogniser can be fascinating, but it can also be mundane.