Why humans still have the upper hand when it comes to facial recognition

Super-recognisers have an uncanny ability to recognise faces, and are able to identify people they have not seen for many years.

For instance, one individual recognised a person who she'd last seen at six years of age. Given this chance encounter occurred some 40 years later, it was unsurprising that the identification was not reciprocated. In fact, some super-recognisers pretend that they do not recognise others to avoid social embarrassment or even accusations of stalking.

Research into super recognition is in its infancy, and it is unclear why some people have excellent face-recognition skills. One likely possibility is that the ability is inherited. Although no genetic investigations have been reported to date, we do know that these skills are more similar in identical compared to non-identical twins, and there have been several reports of face blindness running in families.

It is also possible that super-recognisers use more effective strategies to look at faces than typical people. Much work suggests we look at faces in a 'holistic' manner, taking account not only of the size and shape of facial features, but, even more importantly, the spacing between them. Many people with face blindness are unable to gather this information, whereas super-recognisers are particularly efficient at holistic processing.

In another study, we monitored the eye movements of super-recognisers and typical people while they looked at faces. Unsurprisingly, typical participants focused on the eyes - an effect that is reduced in people with face blindness. In contrast, super-recognisers spent most time looking at the nose. It is possible that fixation on the centre of the face allows the viewer to see a wider area, enabling holistic processing. By looking at the nose, super-recognisers may also be less distracted by the social and emotional information that is found in the eye region.

Current work is exploring whether super-recognisers can process all faces this efficiently, or if they are only exceptional at certain face-recognition tasks. For instance, some individuals are able to very accurately decide whether two pieces of CCTV footage contain the same or different faces, but they cannot memorise faces for long periods of time. Superior face-recognition skills may also be limited by expertise with different types of face. All work to date has examined Caucasian super-recognisers using adult Caucasian faces - it may be that these individuals perform less well with faces that they seldom encounter, such as children's faces or those from other ethnicities.

Such findings would have important implications for the use of super-recognisers in policing and security settings, suggesting that an 'army' of officers may be required for specific face recognition tasks.

One might question whether a simpler option is to instead use automated face-recognition technology. A recent study found that a combination of computer and human resources resulted in the most accurate face recognition decisions, presenting a potential avenue for the most effective use of super-recognisers.

However, this study only used images that are optimal for computer recognition - all faces were looking towards the camera and were unobscured by clothing or accessories. It is well known that automated software fares much less well with differing viewpoints and lighting conditions, leaving us with the somewhat comforting conclusion that humans may be the optimal resource for complex face-recognition tasks for some time to come.

Dr Sarah Bate is an associate professor of psychology at Bournemouth University, and author of the book Face Recognition and its Disorders. More information about super recognition and face blindness can be found on her website www.prosopagnosiaresearch.org, with the opportunity to register for research participation