Face recognition is crucially important for humans, and the vast majority of us are able to identify thousands of faces individually, or to easily pick out familiar faces in a crowd. A special expertise is needed to make such distinctions, and this expertise is nearly universal not only in humans but in other primates. How, then, do people with prosopagnosia manage?
In the last few decades, we have become very conscious of the brain’s plasticity, how one part or system of the brain may take over the functions of a defective or damaged one. But this does not seem to occur with prosopagnosia or topographical agnosia; they are usually lifelong conditions that do not lessen as one grows older. People with prosopagnosia, therefore, need to be resourceful and inventive, need to find strategies, ways of circumventing their deficits: recognizing people by an unusual nose or beard, spectacles, or a certain sort of clothing.3 Many prosopagnosics recognize people by voice, posture, or gait; and, of course, context and expectation are paramount—one expects to see one’s students at school, one’s colleagues at the office, and so on. Such strategies, both conscious and unconscious, become so automatic that people with moderate prosopagnosia can remain unaware of how poor their facial recognition actually is, and are startled if it is revealed to them by testing (for example, with photographs that omit ancillary clues such as hair or eyeglasses).4
Thus, though I may be unable to recognize a particular face at a glance, I can recognize various things about a face: that there is a large nose, a pointed chin, tufted eyebrows, or protruding ears. Such features become identifying markers by which I recognize people. (I think, for similar reasons, I find it easier to recognize a caricature than a straightforward portrait or photograph.) I am reasonably good at judging age and gender, though I have made a few embarrassing blunders here. I am far better at recognizing people by the way they move, their “motor style.” And even if I cannot recognize particular faces, I am sensitive to the beauty of faces, and to their expressions.5
I avoid conferences, parties, and large gatherings as much as I can, knowing that they will lead to anxiety and embarrassing situations—not only failing to recognize people I know well, but greeting strangers as old friends. (Like many prosopagnosics, I avoid greeting people by name, lest I use the wrong one, and I depend on others to save me from egregious social blunders.)
I am much better at recognizing my neighbors’ dogs (they have characteristic shapes and colors) than my neighbors themselves. Thus when I see a youngish woman with a Rhodesian ridgeback hound, I realize that she lives in the apartment next to mine. If I see an older lady with a friendly golden retriever, I know this is someone from down the block. But if I should pass either woman on the street without her dog, she might as well be a complete stranger.
The idea that “the mind”—an immaterial, airy thing—could be embodied in a lump of flesh—the brain—was intolerable to seventeenth-century religious thinking; hence the dualism of Descartes and others. But physicians, observing the effects of strokes and other brain injuries, had long had reason to suspect that the functions of the mind and brain were linked. Toward the end of the eighteenth century, the anatomist Franz Joseph Gall proposed that all mental functions must arise from the brain—not from the “soul,” as many people imagined, or from the heart or the liver. Instead, he envisioned within the brain a collection of twenty-seven “organs,” each responsible for a different moral or mental faculty. Such faculties, for Gall, included what we would now call perceptual functions, such as the sensation of color or sound; cognitive faculties, like memory, mechanical aptitude, or speech and language; and even “moral” traits such as friendship, benevolence, or pride. For these heretical ideas, he was exiled from Vienna and wound up eventually in revolutionary France, where he hoped a more scientific approach might be embraced.6
The physiologist Jean-Pierre Flourens decided to investigate Gall’s theory by removing slices of the brain in living animals, chiefly pigeons. But he could not find any evidence to correlate specific areas of the cortex with specific faculties (perhaps because one needs very delicate and discrete ablations to do so, especially in the tiny pigeon cortex). So Flourens believed that the cognitive impairments his pigeons exhibited as he removed more pieces of cortex reflected only the amount of cortex removed, not its location, and what applied to birds, he felt, probably also applied to human beings. The cortex, he concluded, was equipotential, as homogeneous and undifferentiated as the liver. “The brain,” Flourens said, only half jesting, “secretes thought as the liver secretes bile.”
Flourens’s notion of an equipotential cortex dominated thought until the studies of Paul Broca in the 1860s. Broca performed autopsies on many patients with expressive aphasia, all of whom, he showed, had damage limited to the frontal lobes on the left side. In 1865, he was able to say, famously, “We speak with our left hemisphere,” and the notion of a homogeneous and undifferentiated brain, it seemed, was laid to rest.
Broca felt that he had located a “motor center for words” in a particular part of the left frontal lobe, an area we now call Broca’s area.7 This seemed to promise a new sort of localization, a genuine correlation of neurological and cognitive functions with specific centers in the brain. Neurology moved confidently ahead, identifying “centers” of every sort: Broca’s motor center for words was followed by Wernicke’s auditory center for words, and Déjerine’s visual center for words, all in the left hemisphere, the language hemisphere, and a center for visual recognition in the right hemisphere.
But while visual agnosia of a general sort was recognized in the 1890s, there was little idea that there could be agnosia for particular visual categories like faces or places—even though major figures like Hughlings Jackson and Charcot had already described specific agnosias for faces and places following damage to the posterior areas of the right hemisphere. In 1872, Jackson described a man who, following a stroke in this area, lost his ability “to recognize places and persons. At one time he did not know his wife … and having wandered from home was unable to find his way back.” Charcot, in 1883, provided an account of a patient who had enjoyed exceptional powers of visual imagery and memory, but lost these suddenly. Charcot describes how this man “cannot even recall his own face. Recently in a public gallery his path seemed to be stopped by a person to whom he was about to offer his excuses, but it was merely his own image reflected in a glass.”
Still, even by the middle of the twentieth century, many neurologists doubted whether the brain had category-specific recognition areas. This may have played a part in delaying the recognition of face-blindness, despite the evidence from clinical cases.
In 1947, Joachim Bodamer, a German neurologist, described three patients who were unable to recognize faces but had no other difficulties with recognition. It seemed to Bodamer that this highly selective form of agnosia needed a special name—it was he who coined the term “prosopagnosia”—and that such a specific loss must imply that there was a discrete area in the brain specialized for face recognition. This has been a matter of dispute ever since: is there a special system dedicated only to face recognition, or is face recognition simply one function of a more general visual recognition system? Macdonald Critchley, writing in 1953, was highly critical of Bodamer’s article and of the very idea of face-blindness. “It seems scarcely credible,” he wrote, “that human faces should occupy a perceptual category which is different from all other objects in space, animate and inanimate. Can there be any attribute of size, shape, colouring or motility which distinguishes a human face from other objects in such a way as to preclude identification?”
But in 1955, the English neurologist Christopher Pallis published a beautifully detailed and documented study of his patient A.H., a mining engineer at a Welsh colliery who had kept a journal and was able to give Pallis an articulate and insightful description of his experiences. One night in June of 1953, A.H. apparently suffered a stroke. He “suddenly felt unwell after a couple of drinks at his club.” He appeared to be confused
and was taken home to bed, where he slept poorly. Getting up the following morning, he found his visual world completely transformed, as he reported to Pallis:
I got out of my bed. My mind was clear but I could not recognize the bedroom. I went to the toilet. I had difficulty finding my way and recognizing the place. Turning round to go back to bed I found I couldn’t recognize the room, which was a strange place to me.
I could not see colour, only being able to distinguish light objects from dark ones. Then I found out all faces were alike. I couldn’t tell the difference between my wife and my daughters. Later I had to wait for my wife or mother to speak before recognizing them. My mother is 80 years old.
I can see the eyes, nose, and mouth quite clearly but they just don’t add up. They all seem chalked in, like on a blackboard.
His difficulty was not limited to recognizing people in real life:
I cannot recognize people in photographs, not even myself. At the club I saw someone strange staring at me and asked the steward who it was. You’ll laugh at me. I’d been looking at myself in a mirror.… I later went to London and visited several cinemas and theatres. I couldn’t make head or tail of the plots. I never knew who was who.… I bought some copies of Men Only and London Opinion. I couldn’t enjoy the usual pictures. I could work out what was what by accessory details, but it’s no fun that way. You’ve got to take it in at a glance.
A.H. had other visual problems: a small defect in one corner of his visual fields, transient difficulty with reading, a total inability to perceive color, and difficulty identifying places. (He had initially had some odd sensations on the left side, too—a “heaviness” of the left hand and a “stinging” feeling in his left index finger and the left corner of his mouth.) But he had no object agnosia: he was able to sort out geometrical figures, to draw complex objects, to assemble jigsaw puzzles and play chess.
Since Pallis’s time, a number of patients with prosopagnosia have come to autopsy. Here the data are clear: virtually all patients who acquire prosopagnosia, irrespective of the cause, have lesions in the right visual association cortex, in particular on the underside of the occipitotemporal cortex; there is nearly always damage in a structure called the fusiform gyrus. These autopsy results gained additional support in the 1980s, when it became possible to visualize the brains of living patients by using CT scans and MRIs—here, too, prosopagnosic patients showed lesions in what came to be called the “fusiform face area.” (Abnormal activity in the fusiform face area has also been correlated with hallucination of faces, as Dominic ffytche and his colleagues have shown.)
In the 1990s, such lesion studies were complemented by functional imaging—visualizing the brains of people with fMRIs as they looked at pictures of faces, places, and objects. These functional studies demonstrated that looking at faces activated the fusiform face area much more strongly than looking at other test images.
That individual neurons in this area could show preferences was first demonstrated in 1969 by Charles Gross and his colleagues, using electrodes in the inferotemporal cortex of macaques. Gross found cells that responded dramatically to the sight of a monkey’s paw—but also, less strongly, to a variety of other stimuli, including a human hand. Subsequently, he found cells with a relative preference for faces.8
At this purely visual level, faces are distinguished as configurations, in part by detecting the geometrical relationships between eyes, nose, mouth, and other features (as Freiwald, Tsao, and Livingstone have established).9 But there is no preference at this level for individual faces; indeed, generic or cartoon faces can elicit the same responses as real ones.
Recognition of particular faces or objects is only achieved at a higher cortical level, in the multimodal area of the medial temporal lobe, which has rich reciprocal connections not only to the fusiform face area but to other areas subserving sensory association, emotion, and memory. Christof Koch, Itzhak Fried, and their colleagues have shown that cells in the multimodal medial temporal lobe area show remarkable specificity, responding only, for example, to images of Bill Clinton, or spiders, or the Empire State Building, or cartoons from The Simpsons. Specific neural units may also respond to hearing or reading the name of the person or object; thus in one patient, a set of neurons responded strongly to pictures of the Sydney Opera House and also to the letter string “Sydney Opera,” though not to the names of other landmarks, such as “Eiffel Tower.”10
Neurons in the medial temporal lobe are capable of encoding representations of individual faces, landmarks, or objects so that they can be easily recognized in a changing environment. Such representations can be constructed rapidly, within less than a day or two after exposure to an unfamiliar individual.
Although such studies involve electrode recordings from single neurons, each of these cells is connected to thousands of other neurons, each of which in turn is connected to thousands more. (Some single cells, moreover, may respond to more than one individual or object.) So a single cell’s response really represents the apex of an immense computational pyramid, perhaps drawing on direct or indirect inputs from the visual, auditory, or tactile cortex, text-recognition areas, memory and emotional areas, and so on.
In humans, some ability to recognize faces is present at birth or soon after. By six months, as Olivier Pascalis and his colleagues have shown in one study, babies are able to recognize a broad variety of individual faces, including those of another species (in this study, pictures of monkeys were used). By nine months, though, the babies became less adept at recognizing monkey faces unless they had received continuing exposure to them. As early as three months, infants are learning to narrow their model of “faces” to those they are frequently exposed to. The implications of this work for humans are profound. To a Chinese baby brought up in his own ethnic environment, Caucasian faces may all, relatively, “look the same,” and vice versa.11 One prosopagnosic acquaintance, born and raised in China, went to Oxford as a student and has lived for decades in the United States. Nonetheless, he tells me, “European faces are the most difficult—they all look the same to me.” It seems that there is an innate and presumably genetically determined ability to recognize faces, and this capacity gets focused in the first year or two, so that we become especially good at recognizing the sorts of faces we are likely to encounter. Our “face cells,” already present at birth, need experience to develop fully.
It is similar with many other capacities, from stereo vision to linguistic power: some predisposition or potential is built in genetically but requires stimulation, practice, environmental richness, and nourishment if it is to develop fully. Natural selection may bring about the initial predisposition, but experience and experiential selection are needed to bring our cognitive and perceptual capacities to their full realization.
The fact that many (though not all) people with prosopagnosia also have difficulty with recognizing places has suggested to some researchers that face and place recognition are mediated by distinct yet adjacent areas. Others believe that both are mediated by a single zone which is perhaps more oriented to faces at one end and to places at the other.
The neuropsychologist Elkhonon Goldberg, however, questions the whole notion of discrete, hardwired centers or modules with fixed functions in the cerebral cortex. He feels that at higher cortical levels there may be much more in the way of gradients, where areas whose function is developed by experience and training overlap or grade into one another. In his book The New Executive Brain, he speculates that a gradiential principle constitutes an evolutionary alternative to a modular one, permitting a degree of flexibility and plasticity impossible for a brain organized in a purely modular fashion.
While modularity, he argues, may be characteristic of the thalamus—an assemblage of nuclei with fixed functions, fixed inputs and outputs—a gradiential organization is more characteristic of the cerebral cortex, and becomes more and more prominent as one ascends from primary sensory cortex to association cortex, to the highest level of all, the
frontal cortex. Modularity and gradients may thus coexist and complement one another.
People with prosopagnosia, even if their chief complaint is of face-blindness, often have difficulty recognizing other specific things. Orrin Devinsky and Martha Farah have remarked that some prosopagnosics are unable to distinguish an apple from a pear, say, or a pigeon from a raven, although they can correctly recognize the general category of “fruit” or “bird.” Joan C. described a similar problem: “I don’t recognize handwriting in the same way that I don’t recognize faces. That is, I might be able to identify a sample of handwriting by recognizing some salient feature or by seeing it in context, but otherwise, forget it. I’ve even failed to recognize my own handwriting.”
Some researchers have proposed that prosopagnosia is not purely a problem with face-blindness, but one aspect of a more general difficulty in distinguishing the individuals in any class, whether the class is of faces, cars, birds, or anything else.
Isabel Gauthier and her colleagues at Vanderbilt tested a group of car experts and a group of expert birders, comparing them to a group of normal subjects. The fusiform face area, they found, was activated when all of the groups looked at pictures of faces. But it was also activated in the car experts when they were asked to identify particular cars, and in the birders when they were asked to identify particular birds. The fusiform face area is primarily tuned for facial recognition, but some of it, it seems, can be trained to distinguish individual items of other sorts. (If, then, an expert bird spotter or car buff is unlucky enough to acquire prosopagnosia, he will also, we might suspect, lose his facility for identifying birds or cars.)
The brain is more than an assemblage of autonomous modules, each crucial for a specific mental function. Every one of these functionally specialized areas must interact with dozens or hundreds of others, their total integration creating something like a vastly complicated orchestra with thousands of instruments, an orchestra that conducts itself, with an ever-changing score and repertoire. The fusiform face area does not work in isolation; it is a vital node in a cognitive network that stretches from the occipital cortex to the prefrontal area. Face-blindness may occur even with an intact fusiform face area, if the lower occipital face areas are damaged. And people with moderate prosopagnosia, like Jane Goodall or myself, can, after repeated exposure, learn to identify those we know best. Perhaps this is because we are using slightly different pathways to do so, or perhaps, with training, we can make better use of our relatively weak fusiform face areas.