Toward a Neurofunctional Definition of "Face-Blindness"
Alison M. Harris & Geoffrey K. Aguirre

University of Pennsylvania Philadelphia, PA
In 1947 Dr. Joachim Bodamer, a German neurologist, published a description of a 24-year-old patient who had come under his care in 1944 following a bullet wound to the head. Patient S, as Bodamer referred to him, had suffered damage to the areas of his brain associated with visual processing; as a result, he was completely blind for several weeks after his injury, and even as his sight gradually returned, he continued to have difficulties perceiving color and form.

Yet in addition to these impairments, which fit with other contemporary accounts of perceptual recovery, S displayed another striking deficit. As Bodamer wrote: "S recognised a face as such, i.e. as different from other things, but could not assign the face to its owner. He could identify all the features of a face, but all faces appeared equally "sober" and "tasteless" to him. Faces had no expression, no "meaning" for him. . . . He could distinguish men and women only by their hair or head covering and even then not always with certainty. Even S's own face, viewed in the mirror, evoked no spark of recognition: 'It could be that of another person, even that of a woman.'" (from Ellis & Florence, 1990, p. 86).

This remarkable dysfunction, which Bodamer called "prosopagnosia" (from the Greek prosop for "face" and agnosia for "ignorance"), has had profound implications for our understanding of how our brains organize visual information. Also known as face-blindness, prosopagnosia is characterized by an inability to recognize familiar faces despite relatively unimpaired perception for other objects. Although problems with face recognition could stem from disruption of any number of mental processes, from basic vision to memory, those who have face-blindness often learn to rely on other visual cues (the way Patient S focused on hair) as well as other senses (hearing a person's voice, for example). Face-blindness is thus commonly thought to reflect damage to a system within the brain specific to the visual processing of facial identity.

A better view of face-blindness
With the advent of high-resolution neuroimaging techniques in the last decade, scientists have been able to observe this face-processing system in the intact brain. In one of the major achievements of cognitive neuroscience, research using functional magnetic resonance imaging (fMRI), which measures changes in blood flow associated with neural activity, has shown that a number of brain regions respond preferentially to faces compared to other visual categories. In many cases these "face-selective" regions overlap with sites of brain damage reported in people with face-blindness. This makes it tempting to conclude that prosopagnosia is a direct result of injury to one or more of the face-selective regions. Yet there is a key complication: face-selective areas identified in fMRI can be defined only functionally, by distinguishing the brain areas that react to faces from the areas that react to objects within a given individual. The exact locations of these areas within the brain, however, vary from person to person. Brain damage in prosopagnosics, meanwhile, is usually defined anatomically, in terms of structural landmarks within the brain.

These landmarks are relatively consistent across people. But their functional relevance -- that is, how they are related to behavior -- is often unknown. Compounding this problem, the brain damage associated with closed-head injury and stroke, two of the main causes of face-blindness, is usually quite extensive and can cover multiple anatomical landmarks. So it's difficult to cleanly tie the face-sensitive areas from scan studies to the areas damaged in patients with prosopagnosia.

It therefore remains an open question whether face-blindness is a straightforward result of damage to functionally defined face-selective regions like those seen in healthy individuals. In a recent paper in the journal NeuroImage, "Understanding the Functional Neuroanatomy of Acquired Prosopagnosia," Bettina Sorger, of Maastrich University's Brain Imaging Center, and colleagues address this question by using fMRI to map the functionally defined brain responses to faces and objects in a prosopagnosic (patient PS) with extensive anatomically defined damage to regions associated with face perception.

A mash-up of function and structure 
The advantages of this "neurofunctional" approach are twofold. First, it can test whether PS's spared visual capabilities (that is, those that remain) are associated with normal activity in relevant functionally defined regions. In addition to face-selective brain regions, for instance, previous work with fMRI has demarcated other areas of visual cortex with special patterns of response, including areas that respond selectively to objects (such as toasters, hammers, and butterflies).

Although PS's object recognition seems relatively unimpaired, her brain damage does include a lesion in the vicinity of this functionally measured "object-selective" cortex. In fact, Sorger and colleagues find object-selective activation in the appropriate location, in spite of the adjacent lesion. This result reveals the usefulness of functional neuroimaging in studying what areas have been spared in damaged brain systems. It also validates the importance of functionally defined object-selective brain regions for normal object-recognition performance. Of greater theoretical interest, however, is the converse application of neurofunctional imaging -- that of studying the neural correlates of impaired function.

In the case of PS, can we find normal face-selective-area activations in the absence of normal face recognition? If so, what does this tell us about how face recognition is accomplished by the healthy human brain? To address these questions, Sorger and colleagues mapped both the anatomical extent of brain damage in patient PS and the functional responses of her brain to the presentation of faces. Remarkably, they found that despite dramatic damage to brain regions associated with face perception, some face-selective processing apparently still occurs in PS's brain; that is, there are areas that still respond only to faces -- but PS nevertheless lacks normal face-recognition skills.

This is a surprising finding, to say the least. While this processing is evidently insufficient for normal face recognition, we would not have suspected its existence without the use of functional neuroimaging techniques. Clearly, more work will be needed to determine the origins of face-selective neuroimaging responses in patients with prosopagnosia, as well as why these representations don't give PS normal face recognition. It would be helpful to combine these results with other measures of brain activity, such as electroencephalography (EEG), which could provide further information on other aspects of face processing, such as its time course. But as a demonstration of the potential of neurofunctional approaches, Sorger and colleagues have provided an exciting starting point for better characterizing how faces are processed in the human brain.

Alison M. Harris is a post-doctoral fellow and Geoffrey Karl Aguirre the director at the G.K. Aguire Lab at University of Pennsylvania's Department of Neurology, where they study the neural basis of high-level visual function.

Elsewhere: Look into discussion groups about prosopagnosia on Yahoo and mySpace; find resources for living with prosopagnosia; check out a community blog describing lay experiences by people with faceblindness; and find leads to other research at the Wikiversity entry on prosopagnosia. -- Edited by David Dobbs at 11/19/2007 6:55 PM