The Mind's Eye

by Oliver Sacks

  My mother had hoped I would follow in her footsteps and become a surgeon, but when she realized how lacking in visual powers I was (and how clumsy, lacking in mechanical skill, too) she resigned herself to the idea that I would have to specialize in something else.

  A few years ago, at a medical conference in Boston, I spoke about Torey’s and Hull’s experiences of blindness, how “enabled” Torey seemed to be by the powers of visualization he had developed, and how “disabled” Hull was—in some ways, at least—by the loss of his powers of visual imagery and memory. After my talk, a man in the audience came up to me and asked how well, in my estimation, sighted people could function if they had no visual imagery. He went on to say that he had no visual imagery whatever, at least none that he could deliberately evoke, and that no one in his family had any, either. Indeed, he had assumed this was the case with everyone until, as a student at Harvard, he had come to participate in some psychological tests and had realized that he apparently lacked a mental power that all the other students, in varying degrees, had.

  “And what do you do?” I asked him, wondering what this poor man could do.

  “I am a surgeon,” he replied. “A vascular surgeon. An anatomist, too. And I design solar panels.” But how, I asked him, did he recognize what he was seeing?

  “It’s not a problem,” he answered. “I guess there must be representations or models in the brain that get matched up with what I am seeing and doing. But they are not conscious. I cannot evoke them.”

  This seemed to be at odds with my mother’s experience—she, clearly, did have extremely vivid and readily manipulable visual imagery, though (it now seemed) this may have been a bonus, a luxury, and not a prerequisite for her career as a surgeon.

  Is this also the case with Torey? Is his greatly developed visual imagery, though clearly a source of much pleasure, not as indispensable as he takes it to be? Might he, in fact, have been able to do everything he did, from carpentry to roof repair to making a model of the mind, without any conscious imagery at all? He himself raises this question.

  The role of mental imagery in thinking was explored by Francis Galton in his 1883 book Inquiries into Human Faculty and Its Development. (Galton, a cousin of Darwin’s, was irrepressible and wide-ranging, and his book includes chapters on subjects as various as fingerprints, eugenics, dog whistles, criminality, twins, synesthesia, psychometric measures, and hereditary genius.) His inquiry into voluntary visual imagery took the form of a questionnaire, with such questions as “Can you recall with distinctness the features of all near relations and many other persons? Can you at will cause your mental image … to sit, stand, or turn slowly around? Can you … see it with enough distinctness to enable you to sketch it leisurely (supposing yourself able to draw)?” The vascular surgeon would have been hopeless on such tests—indeed, it was questions such as these that had floored him when he was a student at Harvard. And yet, finally, how much had it mattered?

  As to the significance of such imagery, Galton is ambiguous and guarded. He suggests, in one breath, that “scientific men, as a class, have feeble powers of visual representation” and, in another, that “a vivid visualizing faculty is of much importance in connection with the higher processes of generalized thoughts.” He feels that “it is undoubtedly the fact that mechanicians, engineers and architects usually possess the faculty of seeing mental images with remarkable clearness and precision” but adds, “I am, however, bound to say, that the missing faculty seems to be replaced so serviceably by other modes of conception … that men who declare themselves entirely deficient in the power of seeing mental pictures can nevertheless give lifelike descriptions of what they have seen, and can otherwise express themselves as if they were gifted with a vivid visual imagination. They can also become painters of the rank of Royal Academicians.”

  A mental image, for Galton, was picturing a familiar person or place in the mind’s eye; it was a reproduction or reconstruction of an experience. But there are also mental images of a much more abstract and visionary kind, images of something which has never been seen by the physical eye but which can be conjured up by the creative imagination and serve as models for investigating reality.6

  In his book Image and Reality: Kekulé, Kopp, and the Scientific Imagination, Alan Rocke brings out the crucial role of such images or models in the creative lives of scientists, especially nineteenth-century chemists. He focuses especially on August Kekulé and the famous reverie, while he was riding a London bus, that led him to visualize the structure of a benzene molecule, a concept that would revolutionize chemistry. Although chemical bonds are invisible, they were as real to Kekulé, as visually imaginable, as the lines of force around a magnet were for Faraday. Kekulé said of himself that he had “an irresistible need for visualization.”

  Indeed, a conversation about chemistry can hardly be maintained without such images and models, and in Mindsight, the philosopher Colin McGinn writes, “Images are not just minor variations on perception and thought, of negligible theoretical interest; they are a robust mental category in need of independent investigation.… Mental images … should be added as a third great category … to the twin pillars of perception and cognition.”

  Some people, like Kekulé, are clearly very powerful visualizers in this abstract sense, but most of us use some combination of experiential visualization (imaging one’s house, for example) and abstract visualization (imagining the structure of an atom). Temple Grandin, though, feels she is a different sort of visualizer.7 She thinks entirely in terms of literal images she has seen before, as if she is looking at a familiar photograph or a film running in her head. When she imagines the concept of “heaven,” for instance, her instant association is to the film Stairway to Heaven, and the image in her mind is that of a staircase ascending into the clouds. If someone remarks that it is a rainy day, she sees, in her mind’s eye, the same “photograph” of rain, her own literal and iconic representation of rain. Like Torrey, she is a powerful visualizer; her extremely accurate visual memory allows her to walk through, in her mind, a factory she is designing, noting structural details even before it is built. Growing up, she assumed this was how everyone thought, and she is puzzled, now, by the idea that some people cannot summon visual images at will. When I told her I could not do so, she asked, “How do you think, then?”

  When I talk to people, blind or sighted, or when I try to think of my own internal representations, I find myself uncertain whether words, symbols, and images of various types are the primary tools of thought or whether there are forms of thought antecedent to all of these, forms of thought essentially amodal. Psychologists have sometimes spoken of “interlingua” or “mentalese,” which they conceive to be the brain’s own language, and Lev Vygotsky, the great Russian psychologist, used to speak of “thinking in pure meanings.” I cannot decide whether this is nonsense or profound truth—it is the sort of reef I end up on when I think about thinking.

  Galton himself was puzzled about visual imagery: it had an enormous range, and although it sometimes seemed an essential part of thinking, at other times it seemed irrelevant. This uncertainty has characterized the debate over mental imagery ever since. A contemporary of Galton’s, the early experimental psychologist Wilhelm Wundt, guided by introspection, believed imagery to be an essential part of thought. Others maintained that thinking was imageless and consisted entirely of analytical or descriptive propositions, and behaviorists did not believe in thinking at all—there was only “behavior.” Was introspection alone a reliable method of scientific observation? Could it yield data that were consistent, repeatable, measurable? It was only in the early 1970s that this challenge was faced by a new generation of psychologists. Roger Shepard and Jacqueline Metzler asked subjects to perform mental tasks that required rotating an image of a geometrical figure in their minds—the sort of imaginary rotation my mother performed when she drew the lizard’s skeleton from memory. They were able to determine in these first quantitative experiments t
hat rotating an image took a specific amount of time—an amount proportional to the degree of rotation. Rotating an image through sixty degrees, for instance, took twice as long as rotating it through thirty degrees, and rotating it through ninety degrees, three times as long. Mental rotation had a rate, it was continuous and steady, and it took effort, like any voluntary act.

  Stephen Kosslyn entered the subject of visual imagery from another angle, and in 1973 published a seminal paper contrasting the performance of “imagers” and “verbalizers” who were asked to remember a set of drawings they had been shown. Kosslyn hypothesized that if internal images were spatial and organized like pictures, the “imagers” ought to be able to focus selectively on a part of the image, and that time would be required for them to shift their attention from one part of the image to another. The time required, he thought, would be proportional to the distance the mind’s eye had to travel.

  Kosslyn was able to show that all of these were indeed the case, indicating that visual images were essentially spatial and organized in space like pictures. His work has proved immensely fertile, but the ongoing debate about the role of visual imagery continues, as Zenon Pylyshyn and others have maintained that the mental rotation of images and “scanning” them could be interpreted as the result of purely abstract, nonvisual operations in the mind/brain.8

  By the 1990s, Kosslyn and others were able to combine imagery experiments with PET and fMRI scanning, which allowed them to map the areas of the brain involved as people engaged in tasks requiring mental imagery. Mental imagery, they found, activated many of the same areas of the visual cortex as perception itself, showing that visual imagery was a physiological reality as well as a psychological one, and used at least some of the same neural pathways as visual perception.9

  That perception and imagery share a common neural basis in the visual parts of the brain is suggested by clinical studies, too. In 1978 Eduardo Bisiach and Claudio Luzzatti in Italy related the cases of two patients who both developed a hemianopia following a stroke and could not see to the left side. When they were asked to imagine themselves walking down a familiar street and describe what they saw, they mentioned only the shops on the right side of the street; but when they were then asked to imagine turning around and walking back, they described the shops they had not “seen” before, the shops that were now on the right side. These beautifully examined cases showed that a hemianopia might cause not only a bisection of the visual field but a bisection of visual imagery as well.

  Such clinical observations on the parallels between visual perception and visual imagery go back at least a century. In 1911, the English neurologists Henry Head and Gordon Holmes examined a number of patients with subtle damage to the occipital lobes—damage that led not to total blindness but to blind spots within the visual field. They found, by questioning their patients carefully, that blind spots in exactly the same locations occurred in the patients’ mental imagery as well. And in 1992, Martha Farah et al. reported that in a patient who lost partial vision on one side due to an occipital lobectomy, the visual angle of his mind’s eye was also reduced, in a way that perfectly matched his perceptual loss.

  For me, the most convincing demonstration that at least some aspects of visual imagery and visual perception might be inseparable occurred when I was consulted in 1986 by Mr. I., an artist who became completely colorblind following a head injury.10 Mr. I. was distressed by his sudden inability to perceive colors, but even more by his total inability to evoke them in memory or imagery. Even his occasional visual migraines were now drained of color. Patients like Mr. I. suggest that the coupling of perception and imagery is very close in the higher parts of the visual cortex.11

  · · ·

  Sharing characteristics and even sharing neural areas or mechanisms is one thing, but Kosslyn and others go further than this, suggesting that visual perception depends on visual imagery, matching what the eye sees, the retina’s output, with memory images in the brain. Visual recognition, they feel, could not occur without such matching. Kosslyn proposes, furthermore, that mental imagery may be crucial in thought itself—problem solving, planning, designing, theorizing. Support for this comes from studies asking subjects to answer questions that would seem to require visual imagery—for example, “Which is darker green, a frozen pea or a pine tree?” or “What shape are Mickey Mouse’s ears?” or “In which hand does the Statue of Liberty hold her torch?”—or asking them to solve problems that can be worked out either by means of imagery or by means of more abstract, nonvisual thinking. Kosslyn speaks here of a doubleness in the way people think, contrasting the use of “depictive” representations, which are direct and unmediated, with “descriptive” ones, which are analytic and mediated by verbal or other symbols. Sometimes, he suggests, one mode will be favored over another, depending on the individual and on the problem to be solved. Sometimes both modes will proceed in tandem (although depiction is likely to outpace description), and at other times one may start with depiction—images—and proceed to a purely verbal or mathematical representation.12

  What, then, of people like me, or the vascular surgeon in Boston who cannot evoke any visual images voluntarily? One must infer, as my colleague in Boston does, that we, too, have visual images, models, and representations in the brain, images that allow visual perception and recognition but are below the threshold of consciousness.13

  · · ·

  If the central role of visual imagery is to permit visual perception and recognition, what need is there for it if a person becomes blind? And what happens to its neural substrates, the visual areas which occupy nearly half of the entire cerebral cortex? We know that in adults who lose their eyesight, there may be some atrophy of the pathways and relay centers leading from the retina to the cerebral cortex—but there is little degeneration in the visual cortex itself. Functional MRIs of the visual cortex show no diminution of activity in such a situation; indeed, we see the reverse: they reveal a heightened activity and sensitivity. The visual cortex, deprived of visual input, is still good neural real estate, available and clamoring for a new function. In someone like Torey, this may free up more cortical space for visual imagery; in someone like Hull, relatively more may be employed by other senses—auditory perception and attention, perhaps, or tactile perception and attention.14

  This sort of cross-modal activation may underlie the fact that some blind people, like Dennis Shulman, “see” Braille as they read it with their finger. This may be more than just an illusion or a fanciful metaphor; it may be a reflection of what is actually happening in his brain, for there is good evidence that reading Braille can cause strong activation of the visual parts of the cortex, as Sadato, Pascual-Leone, et al. have reported. Such activation, even in the absence of any input from the retina, may constitute a crucial part of the neural basis for the mind’s eye.

  Dennis also spoke of how the heightening of his other senses had increased his sensitivity to the most delicate nuances in other people’s speech and self-presentation. He could recognize many of his patients by smell, he said, and he could often pick up states of tension or anxiety they might not even be aware of. He felt that he had become far more sensitive to others’ emotional states since losing his sight, for he was no longer taken in by visual appearances, which most people learn to camouflage. Voices and smells, by contrast, he felt, could reveal people’s depths.

  The heightening of other senses with blindness allows a number of very remarkable adaptations, including “facial vision,” the ability to use sound or tactile clues to sense the shape or size of a space and the people or objects in it.

  Martin Milligan, the philosopher, who had both eyes removed at the age of two (because of malignant tumors), has written of his own experience:

  Born-blind people with normal hearing don’t just hear sounds: they can hear objects (that is, have an awareness of them, chiefly through their ears) when they are fairly close at hand, provided these objects are not too low; and they can also in
the same way “hear” something of the shape of their immediate surroundings.… Silent objects such as lamp-posts and parked cars with their engines off can be heard by me as I approach them and pass them as atmosphere-thickening occupants of space, almost certainly because of the way they absorb and/or echo back the sounds of my footsteps and other small sounds.… It isn’t usually necessary to make sounds oneself to have this awareness, though it helps. Objects of head height probably slightly affect the air currents reaching my face, which helps towards my awareness of them—which is why some blind people refer to this kind of sense-awareness as their “facial” sense.

  Facial vision tends to be most highly developed in those who are born blind or lose their sight at an early age; for the writer Ved Mehta, who has been blind since the age of four, it is so well developed that he walks confidently and rapidly without a cane, and it is sometimes difficult for others to realize that he is blind.

  While the sound of one’s footsteps or one’s cane may suffice, other forms of echolocation have been reported. Ben Underwood developed an astonishing, dolphin-like strategy of emitting regular clicks with his mouth and accurately reading the resulting echoes from nearby objects. He was so adept at moving about the world in this way that he was able to play field sports and even chess.15

  Blind people often say that using a cane enables them to “see” their surroundings, as touch, action, and sound are immediately transformed into a “visual” picture. The cane acts as a sensory substitution or extension. But is it possible to give a blind person a more detailed picture of the world, using more modern technology? Paul Bach-y-Rita was a pioneer in this realm and spent decades testing all sorts of sensory substitutes, though his special interest lay in developing devices that could help the blind by using tactile images. (In 1972, he published a prescient book surveying all the possible brain mechanisms by which sensory substitution might be realized. Such substitution, he emphasized, would depend on the brain’s plasticity—and that the brain had any plasticity at all was a revolutionary concept at the time.)