A Face Is More Than the Sum of Its Parts

You stop at a shop window and wonder why someone inside is blatantly staring at you — until you realize that person is you. Scenarios like this are impossible for most of us to imagine, but quite common for sufferers of acquired prosopagnosia (AP), a condition hindering the ability to recognize faces which can occur after brain damage. In a new study, Belgian researchers have found that the condition is linked to an inability to process faces holistically, that is, as a whole. Meike Ramon and Bruno Rossion from the Université de Louvain in Belgium have been investigating the case of PS, a 59 year-old kindergarten teacher and one of the few cases of pure acquired prosopagnosia in the world. She has been suffering from AP since a closed head injury in 1992. Past accounts of the condition have focused on AP sufferers’ difficulty in processing the eye-region of a face, or in perceiving relative distances between facial features. Ramon and Rossion found that both impairments are linked to a common cause: the inability to process faces as the sum of its parts.

Participants in the study were asked to match images of faces, which had been manipulated to differ either in a single feature or the distance between two features. As expected, the patient PS had difficulty in the matching task when changes to the faces occurred randomly. Strikingly, however, when told which feature had been changed (e.g. distance between the eyes), her performance profile paralleled that of healthy subjects.

The findings suggest that AP patients are unable to process different elements of the face in parallel and instead “apply a locally restricted, serial processing style, which is particularly inefficient for certain types of information.” Knowing which information to look for makes this strategy relatively more efficient. Although this may not help AP patients in real-life situations, “it does however shed light on what makes normal face recognition so overwhelmingly efficient: Our capacity to simultaneously integrate the multiple facial elements into a unique representation,” says Ramon. ♦