The origins of autism are thought to lie in early brain development. During the first three years of life, the brain grows at a tremendous rate. In autistic children neurons seem to connect haphazardly, causing widespread abnormalities, especially in the cerebellum, which integrates thinking and movement, and in the limbic region, which integrates experience with specific emotions. Abnormalities in these regions seem to stunt interest in one’s environment and in social interaction. Autistic children have narrowed fields of attention and a poor ability to recognize faces. They are more likely to view a face, for example, as individual components rather than as a whole. Imaging studies have shown that when autistic children see a familiar face, their pattern of brain activation is different from that of normal children.

That narrowed focus may explain the autistic child’s ability to concentrate endlessly on a single repetitive activity, such as rocking in a chair or watching clothes tumble in a dryer. Only one out of 10 autistic children shows special skills.

In a 1999 paper, Snyder and his colleague John Mitchell challenged the compulsive-learning explanation for savant abilities, arguing that the same skills are biologically latent in all of us. “Every­one in the world was skeptical,” says V. S. Ramachandran, director of the Center for Brain and Cognition at the University of California at San Diego. “Snyder deserves credit for making it clear that savant abilities might be extremely important for understanding aspects of human nature and creativity.”

Snyder’s office at the University of Sydney is in a Gothic building complete with pointed towers and notched battlements. Inside, Nadia’s drawings of horses adorn the walls; artwork by other savants hangs in nearby rooms.

Snyder’s interest in autism evolved from his studies of light and vision. Trained as a physicist, he spent several years studying fiber optics. At one time he was interested in studying the natural fiber optics in insects’ eyes. The question that carried him from vision research to autism had to do with what happens after light hits the human retina: How are the incoming signals transformed into data that are ultimately processed as images in the brain? Snyder was fascinated by the processing power required to accomplish such a feat.

During a sabbatical at the University of Cambridge in 1987, Snyder devoured Ramachandran’s careful studies of perception and optical illusions. One showed how the brain derives an object’s three-dimensional shape: Falling light creates a shadow pattern on the object, and by interpreting the shading, the brain grasps the object’s shape. “You’re not aware how your mind comes to those conclusions,” Snyder says. “When you look at a ball, you don’t know why you see it as a ball and not a circle. The reason is your brain is extracting the shape from the subtle shading around the ball’s surface.” Every brain possesses that innate ability, yet only artists can do it backward, using shading to portray volume.

“Then,” Snyder says, speaking slowly for emphasis, “I asked the question that put me on a 10-year quest.” How can we bypass the mind’s conceptual thinking and gain conscious access to the raw, uninterpreted information of our basic perceptions? Can we shed the assumptions built into our visual processing system?

A few years later, he read about Nadia and other savant artists in Oliver Sacks’s The Man Who Mistook His Wife for a Hat and Other Clinical Tales. As he sat in his Sydney apartment one afternoon with the book in hand, an idea surfaced. Perhaps someone like Nadia who lacked the ability to organize sensory input into concepts might provide a window into the fundamental features of perception.

Snyder’s theory began with art, but he came to believe that all savant skills, whether in music, calculation, or spatial relationships, derive from a lightning-fast processor in the brain that divides things—time, space, or an object—into equal parts. Dividing time might allow a savant child to know the exact time when he’s awakened, and it might help Eric find the sweet spot in a room by allowing him to sense milli­second differences in the sounds hitting his ears. Dividing space might allow Nadia to place a disembodied hoof and mane on a page precisely where they belong. It might also allow savant twins to instantaneously count matches spilled on the floor (one said “111”; the other said “37, 37, 37”). Meanwhile, splitting numbers might allow math savants to factor 10-digit numbers or easily identify large prime numbers, which are impossible to split.

Compulsive practice might enhance these skills over time, but Snyder contends that practice alone cannot explain the phenomenon. As evidence, he cites rare cases of sudden-onset savantism. Orlando Serrell, for example, was hit on the head by a baseball at the age of 10. A few months later he began recalling an endless barrage of license-plate numbers, song lyrics, and weather reports.

If someone can become an instant savant, Snyder thought, doesn’t that suggest we all have the potential locked away in our brains? “Snyder’s ideas sound very New Age. This is why people are skeptical,” Ramachandran says. “But I have a more open mind than many of my colleagues simply because I’ve seen [sudden-onset cases] happen.”

Bruce Miller, a neurologist at the University of California at San Francisco, has seen similar transformations in patients with frontotemporal dementia, a degenerative brain disease that strikes people in their fifties and sixties. Some of these patients, he says, spontaneously develop both interest and skill in art and music. Brain-imaging studies have shown that most patients with frontotemporal dementia who develop skills have abnormally low blood flow or low metabolic activity in their left temporal lobe. Because language ability is concentrated in the left side of the brain, these people gradually lose the ability to speak, read, and write. They also lose face recognition. Meanwhile, the right side of the brain, which supports visual and spatial processing, is better preserved.

“They really do lose the linguistic meaning of things,” says Miller, who believes Snyder’s ideas about latent abilities complement his own observations about frontotemporal dementia. “There’s a loss of higher-order processing that goes on in the anterior temporal lobe.” In particular, frontotemporal dementia damages the ventral stream, a brain region that is associated with naming objects. Patients with damage in this area can’t name what they’re looking at, but they can often paint it beautifully.

Miller has also seen physiological similarities in the brains of autistic savants and patients with frontotemporal dementia. When he performed brain-imaging studies on an autistic savant artist who started drawing horses at 18 months, he saw abnormalities similar to those of artists with frontotemporal dementia: decreased blood flow and slowed neuronal firing in the left temporal lobe.

One blustery, rainy morning I drove to Mansfield, Australia, a small farm town 180 miles northeast of Melbourne. I was heading to a day clinic for autistic adults, where I hoped to meet a savant. The three-hour drive pitched and rolled across a hilly terrain, occasionally cutting through dense eucalyptus forests punctuated with yellow koala-crossing signs. From time to time I saw large, white-crested parrots; in one spot, a flock of a thousand or more in flight wheeled about like a galaxy.

I finally spotted my destination: Acorn Outdoor Ornaments. Within this one-story house, autistic adults learn how to live independently. They also create inexpensive lawn decorations, like the cement dwarf on the roof.

Joan Curtis, a physician who runs Acorn and a related follow-up program, explained that while true savants are rare, many people with autism have significant talents. Nurturing their gifts, she said, helps draw them into social interaction. One of the participants I met at Acorn was a man named Guy. Although he was uncomfortable shaking my hand, all things electronic fascinated him, and he questioned me intently about my tape recorder.

Every horizontal surface in Guy’s room was covered with his creations. One was an electric fan with a metal alligator mouth on the front that opened and closed as it rotated from side to side. On another fan a metal fisherman raised and lowered his pole with each revolution. And then I saw the sheep. Viewed from the left, it was covered in wool. Viewed from the right, it was a skeleton, which I learned Guy had assembled without any help. Guy didn’t say much about himself. He cannot read nor do arithmetic, but he has built an electric dog that barks, pants, wags its tail, and urinates.

During my visit, another Acorn participant, Tim, blew into the room like a surprise guest on The Tonight Show. He was in a hurry to leave again but asked me my birthday, which is July 15, 1970.

“Born on a Wednesday, eh?” he responded nonchalantly— and correctly.

“How did you do that?” I asked.

Savants can access a startling capacity for recalling endless detail or for performing lightning-quick calculations.

“I did it well,” he replied.

“But how?” I asked.

“Very well,” he replied, with obvious pleasure. Then he was out the door and gone.

How do calendar savants do it? In 1995 Timothy Rickard, a cognitive psychologist at the University of California at San Diego, evaluated a 40-year-old man with a mental age of 5 who could assign a day of the week to a date with 70 percent accuracy. Because the man had been blind from birth, he couldn’t study calendars or even imagine calendars. He couldn’t do most arithmetic either, so he wasn’t using a mathematical algorithm. And he could only do dates falling within his lifetime, which suggests that he used memory.

He could, however, do a bit of rudimentary arithmetic, such as answer this question: If today is Wednesday, what day is two days from now? Rickard suspects that memorizing 2,000 dates and using such arithmetic would allow 70 percent accuracy. “That doesn’t reduce it to a trivial skill, but it’s not inconceivable that someone could acquire this performance with a lot of effort,” he says. It’s especially plausible given the single-minded drive with which autistics pursue interests.

Yet Tim, the calendar savant at Acorn, can calculate dates as far back as 1900, as well as into the future. And there are reports of twins who could calculate dates 40,000 years in the past or the future. Still, practice may be part of it. Robyn Young, an autism researcher at Flinders University in Adelaide, Australia, says some calendar savants study perpetual calendars (there are only 14 different calendar configurations; perpetual calendars cross-reference them to years).

But even if savants practice, they may still tap into that universal ability Snyder has proposed. Here it helps to consider art savants. That Nadia began her drawings with minor features rather than overall outlines suggests that she tended to perceive individual details more prominently than she did the whole—or the concept—of what she was drawing. Other savant artists show the same tendencies. Stephen Wiltshire, for instance, is now famous for his intricately detailed illustrations of buildings and cityscapes, a talent that first emerged when he was only 5.

Autistic children differ from nonautistic children in another way. Normal kids find it frustrating to copy a picture containing a visual illusion, such as M. C. Escher’s drawing in which water flows uphill. Autistic children don’t. That fits with Snyder’s idea that they are recording what they see without interpretation and reproducing it with ease in their own drawings.

Even accomplished artists sometimes employ strategies to shake up their preconceptions about what they are seeing. Guy Diehl is not a savant, but he is known for his crystal-clear still lifes of stacked books, drafting implements, and fruit. When Diehl finds that he’s hit a sticking point on a painting, he may view it in a mirror or upside down. “It reveals things you otherwise wouldn’t see, because you’re seeing it differently,” he says. “You’re almost seeing it for the first time again.”

Snyder is experimenting with grander ideas, such as enhancing creativity by turning down our conceptual abilities.

Diehl showed me how art students use this technique to learn to draw. He put a pair of scissors on a table and told me to draw the negative space around the scissors, not the scissors themselves. The result: I felt I was drawing individual lines, not an object, and my drawing wasn’t half bad, either.

Drawing exercises are one way of coaxing conceptual machinery to take five, but Snyder is pursuing a more direct method. He has suggested that a technique called trans­cranial magnetic stimulation (TMS), which uses magnetic fields to disrupt neuronal firing, can knock out a normal person’s conceptual brain machinery, temporarily rendering him savantlike.

Young and her colleague Michael Ridding of the University of Adelaide tried it. Using transcranial magnetic stimulation on 17 volunteers, they inhibited neural activity in the frontotemporal area. This language and concept-supporting brain region is affected in patients with frontotemporal dementia and in the art savant whom Miller studied. In this altered state, the volunteers were asked to perform the kinds of tasks savants gravitate to—horse drawing, calendar calculating, and multiplying.

Five of the 17 volunteers improved—not to savant levels, but no one expected that, because savants practice. Further, trans­cranial magnetic stimulation isn’t a precise tool for targeting brain regions. But the five volunteers who improved were those whose frontotemporal areas had been successfully targeted, according to separate neurological assessments. “Obviously I don’t think the idea is so outlandish anymore,” Young says. “I think it is a plausible hypothesis. It always was, but I didn’t expect we’d actually find the things we did.”

Snyder has even stronger findings from a more recent TMS study. Twelve normal subjects were allowed briefly to glimpse between 50 and 150 ellipses on a monitor. They were given only 1.5 seconds, long enough to see the ellipses but too short to count them, and then were asked to guess how many objects were there. A brief session of TMS—just 15 minutes—improved the performance of 10 of the 12 subjects. (They were asked to do the counting right after the TMS was switched off; doing the trial while the stimulation was under way, Snyder says, would be like “trying to focus on something while being hit over the head with a baseball bat.”) Eight of the 10 whose performance had improved saw it decline an hour later, as the effects slowly wore off. “The numerosity study is very clear,” Snyder says. “By using TMS, we become more literal.”

Snyder is experimenting with grander ideas as well, such as enhancing creativity by turning down our conceptual abilities.“We see constellations instead of the individual stars,” Snyder says. “To be creative, you have to join the dots up in a unique way. But how can you do that if you keep superimposing the ways you already know?” He imagines a combination of training and hardware that might, for example, help an engineer get past a sticking point on a design project by offering a fresh angle on the problem. One method would involve learning to monitor one’s own brain waves. By watching one’s own brain waves during drawing exercises, Snyder imagines it may be possible to learn to control them in a way that shuts down the brain’s concept-making machinery—perhaps the left temporal lobe itself.

Even if further research never fully reveals why savants have extraordinary skills, we may at least learn from their potential. Snyder is optimistic. “I envisage the day,” he says, “when the way to get out of a [mental rut] is you pick up this thing—those of us with jobs that demand a certain type of creativity—and you turn off part of your brain. I’m very serious about this.”