UCL News


These cells recognise faces and only faces

13 February 2006

There is a specific mechanism in our brains that deals only with recognising peoples' faces and it is separate from the mechanism that allows us to recognise objects like houses, cars, horses or even people's bodies, according to a study led by Dr Brad Duchaine at UCL (University College London).

He shows how we recognise faces by analysing one man, who can't tell one face from another, in a paper that will appear online in Cognitive Neuropsychology journal on 13th February 2006.

Dr Duchaine [UCL Institute of Cognitive Neuroscience] said: "There have been many theories about whether there is a part of the brain that deals specifically with faces or whether faces and other objects are handled by the same brain areas. We've found that there is a different, very separate, bit of the brain that lets you recognise faces. If those cells aren't working, someone may not be able to tell two faces apart but they will recognise two horses apart. This indicates that we go about looking at, analysing and recognising faces in a different way from how we recognise objects.

"Whether there is a separate social bit of the brain is an interesting scientific question that we want to answer. So that we could draw firm conclusions to prove our facial recognition theory, we addressed all the alternatives in a single case study - Edward, a 53 year-old married man with a PhD in theology and physics, who happens to be unable to recognise faces."

For more than 35 years, researchers have debated whether face recognition is carried out by face-specific mechanisms or whether it involves more generic mechanisms that are also used for objects. Prosopagnosic patients (people who have difficulty recognising faces) have been some of the most powerful sources of evidence for there being face-specific mechanisms.

Scientists have put forward a number of different theories about why some people can't recognise faces. One theory states that people with face recognition just have problems with objects that have a lot of curved surfaces (Curvature explanation); another theory states that it's caused by problems with perceiving distances between parts such as judging the distance between eyes (Configural-processing explanation); another puts it down to a problem in recognising any individual item in a class - objects and faces (individuation explanation); another points to an inability to develop expertise with regularly encountered objects (Expertise explanation).

But, for each case of prosopagnosia that has been scientifically tested there have always been several untested alternative explanations that could account for the inability to recognise faces. Each of these individuals has not been sufficiently tested to provide conclusive evidence for face-specific processes.

This study addresses all the existing theories that make a case against there being a specific mechanism in the brain that deals just with faces. Dr Duchaine said: "We reject each in turn and eliminate all alternative accounts. The results show that face recognition uses mechanisms in the brain that are different from those used in recognising anything else."

The tests at UCL were done using a variety of types of objects - horses, guns, cars, greebles (novel objects). Edward performed just as well on as the test group people at differentiating between different objects but he just couldn't tell faces apart.

Notes for Editors:

1. For further information and a copy of the paper that appears online on Cognitive Neuropsychology titled "Prosopagnosia as an impairment to face-specific mechanisms: Elimination of the alternative hypotheses in a developmental case" please contact Alex Brew in the UCL press office on 020 7679 9726 or a.brew@ucl.ac.uk. Out-of-hours contact 07747 565 056.

2. You may be able to interview the Prosopagnosic patient, Edward, who was used in this study. Please contact Alex Brew as above to interview him.

3. To interview Dr Duchaine contact him on +44 20 7679 1005 or contact the press office as above.