Tag Archives: face recognition

How the brain recognises faces

This post was contributed by Dr Clare Sansom, Senior Associate Lecturer, Department of Biological Sciences 

The first of two evening lectures on the Wednesday of Birkbeck Science Week 2015 was given by Martin Eimer of the college’s Department of Psychological Sciences.

He, like the other Science Week lecturers, was introduced by the Dean of the Faculty of Science, Nicholas Keep; Professor Keep explained that Eimer, a native of Germany and a recently elected Fellow of the German Academy of Sciences, had built up his research lab at Birkbeck over the last fifteen years.

Language

His internationally recognised research concerns the relationship between brain function and behaviour in health and disease. The topic he selected for his lecture was a fascinating one: how our brains recognise human faces and what happens when this automatic process goes wrong.

Eimer began by outlining some reasons why we find faces so interesting to look at. When we look at a face we may be able to recognise that individual, either immediately or with difficulty, but – if our brains are working correctly – we will be able to tell what the person is feeling, or what they are looking at.

It seems that the facial expressions that are associated with basic emotions such as happiness, surprise, fear and disgust are common between most if not all cultures. And we also use faces to lip-read. People with hearing impairments are dependent on this, and learn to do it very well, but we all have some intrinsic lip-reading ability that we use automatically in noisy environments.

Next, he used perceptual demonstrations to illustrate that we process faces rather differently to other objects. If we look at a photo of a familiar or famous person that has been turned upside down we automatically think it looks odd, and we find the face hard to identify. This so-called ‘inversion effect’ is also seen with other objects but is much more pronounced with faces.

A stranger effect occurs if the photo of a face is altered so that only the eyes and mouth are upside down. This looks grotesque, but turning the altered photo upside down so that the eyes and mouth only are the right way up makes it look surprisingly normal. This was named the ‘Thatcher illusion’ by the scientists who discovered it in 1980, perhaps as an imaginative way of taking revenge for an early round of education cuts.

It is likely that we instinctively respond so differently to faces out of the normal upright orientation because our brains have an inbuilt ‘face template’. Even young infants respond to ‘face-like’ stimuli with two eyes, a nose and a mouth in approximately the right proportions and positions.

Face recognition, too, depends on small differences in these parameters between individuals (e.g. the height of the eyes above the nose and the distance between them). Contrast polarity is also important, and we find it much harder to identify face images if their contrast is inverted (as in a photographic negative). Interestingly, however, the task becomes easier if the eye region only is reverted to normal contrast. This suggests that we attach a particular importance to that region. It is also difficult to determine gaze direction if the contrast polarity around the eyes are inverted.

Eimer introduced another optical illusion in which half of each of the faces of George Clooney and Harrison Ford had been combined into a composite. The audience found it almost impossible to distinguish the two actors until the half-faces were separated. We had all instinctively formed a new face from the components and failed, for obvious reasons, to match it to an individual. This trick, which is known as holistic face processing, is also specific to faces.

The second half of the lecture dealt with the neuroscience of face recognition, and what happens when it goes wrong. When we look at a face (or any object) information from the image focused on the retina is initially transferred to a part of the back of the brain known as the primary visual cortex. It is then transferred to other parts of the brain, including the inferior temporal cortex, where objects are recognised.

Several types of experiments have been developed for measuring exactly what goes on in the brain. These include functional magnetic resonance imaging (fMRI), which generates brightly coloured images associated with changes in blood flow to parts of the brain, and electroencephalography (EEG) which records electrical activity on the scalp.

These techniques are complementary; EEG is faster but can only record signals from the surface of the brain. Between them, they have allowed scientists to identify several areas in the brain that are activated when faces, but not other objects, are perceived and a rapid, strong electrical impulse that seems to be a unique response to faces.

It is much easier to recognise the face of a familiar individual – family member, friend or celebrity – than to distinguish between the faces of unknown people. This task, however, is required in many professions: most often and most obviously passport officers and detectives, but also, for example, teachers at the beginning of each new school year. Some people are much better at doing this than others, but even the most skilled make mistakes, and the UK immigration service (and, no doubt, the equivalent bodies in other countries) is looking into ways of doing it automatically.

People at the other end of the spectrum – who find it particularly difficult to recognise faces – are said to have a condition called prosopagnosia, or ‘face blindness’. These people have a severe but very specific defect in recognising faces: their intellect and their vision are normal, and they can recognise individuals easily enough from their voice, gait or other cues.

This condition is divided into two types: acquired prosopagnosia, which arises after brain damage, and developmental prosopagnosia, which can be apparent from early childhood, without any obvious brain damage. The acquired type is typically more severe; the eponymous Man who Mistook his Wife for a Hat described in Oliver Sacks’ fascinating book suffered from this condition. The rapid brain response to faces is missing from an EEG of a person with acquired prosopagnosia, and other tests will show that the brain regions that are specifically associated with face processing have been damaged.

About 2% of the population can be said to have some degree of developmental prosopagnosia. There is no association with intelligence and it affects many successful professionals. Eimer showed part of a TV programme featuring an interview with a woman who is particularly badly affected. She explained the problems she has encountered throughout her life, ranging from following characters in films to telling her own daughter from other little girls with bunches in the school playground. Her father had also suffered from the condition, and she had been very relieved to receive a formal diagnosis.

The EEG patterns of individuals with developmental prosopagnosia are less different from normal than those of people with brain damage, but they are recognisable. Interestingly, differences in brain responses to upright as compared to inverted faces are not seen in people with developmental prosopagnosia.

Face recognition abilities form a continuum and many people who think of themselves as being ‘terrible’ at recognising faces will find that they are in the normal range. Eimer’s group has a website that includes an online test, the Cambridge Face Memory Test. Participants are asked to memorise a face and then pick it out from a group of three; the tests start easy but become more challenging. People with very high and very low scores will be invited to be involved in further research in the Brain and Behaviour Lab at Birkbeck

Interested? Find out more

Share