http://www.nytimes.com/2006/07/13/science/13brain.html

A paralyzed man with a small sensor implanted in his brain was able to control a computer, a television set and a robot using only his thoughts, scientists reported yesterday.
Rick Friedman
Matthew Nagle, left paralyzed when he was stabbed five years ago, and the circle he drew on a computer screen by using only his thoughts.
Those results offer hope that in the future, people with spinal cord injuries, Lou Gehrig’s disease or other conditions that impair movement may be able to communicate or better control their world.
“If your brain can do it, we can tap into it,” said John P. Donoghue, a professor of neuroscience at who has led development of the system and was the senior author of a report on it being published in today’s issue of the journal Nature.
In a variety of experiments, the first person to receive the implant, Matthew Nagle, moved a cursor, opened e-mail, played a simple video game called Pong and drew a crude circle on the screen. He could change the channel or volume on a television set, move a robot arm somewhat, and open and close a prosthetic hand.
Although his cursor control was sometimes wobbly, the basic movements were not hard to learn.
“I pretty much had that mastered in four days,” Mr. Nagle, 26, said in a telephone interview from the New England Sinai Hospital and Rehabilitation Center in Stoughton, Mass. He said the implant did not cause any pain.
Mr. Nagle, a former high school football star in Weymouth, Mass., was paralyzed below the shoulders after being stabbed in the neck during a melee at a beach in July 2001. He said he had not been involved in starting the brawl and did not even know what had sparked it. The man who stabbed him is now serving 10 years in prison, he said.
Implants like the one he received had previously worked in monkeys. There have also been some tests of a simpler sensor implant in people, as well as systems using electrodes outside the scalp. And Mr. Nagle has talked before about his experience.
But the paper in Nature is the first peer-reviewed publication of an experiment in people with a more sophisticated implant, able to monitor many more brain neurons than earlier devices. The paper helps “shift the notion of such ‘implantable neuromotor prosthetics’ from science fiction towards reality,” Stephen H. Scott, professor of anatomy and cell biology at Queen’s University in Ontario, wrote in a commentary in the journal.
The sensor measures 4 millimeters by 4 millimeters — less than a fifth of an inch long and wide — and contains 100 tiny electrodes. The device was implanted in the area of Mr. Nagle’s motor cortex responsible for arm movement and was connected to a pedestal that protruded from the top of his skull.
When the device was to be used, technicians plugged a cable connected to a computer into the pedestal. So Mr. Nagle was directly wired to a computer, somewhat like a character in the “Matrix” movies.
Mr. Nagle would then imagine moving his arm to hit various targets. The implanted sensor eavesdropped on the electrical signals emitted by neurons in his motor cortex as they controlled the imaginary arm movement.
Obstacles must be overcome, though, before brain implants become practical. For one thing, the electrodes’ ability to detect brain signals begins to deteriorate after several months, for reasons not fully understood. In addition, the implant would ideally transmit signals wirelessly out of the brain, doing away with the permanent hole in the head and the accompanying risk of infection. Further, the testing involving Mr. Nagle required recalibration of the system each day, a task that took technicians about half an hour.
Still, scientists said the study was particularly important because it showed that the neurons in Mr. Nagle’s motor cortex were still active years after they had last had a role to play in moving his arms.
The implant system, known as the BrainGate, is being developed by Cyberkinetics Neurotechnology Systems Inc. of Foxborough, Mass. The company is now testing the system in three other people, who remain anonymous: one with a spinal cord injury, one with Lou Gehrig’s disease and one who had a brain stem stroke.
Timothy R. Surgenor, president and chief executive, said Cyberkinetics hoped to have an implant approved for marketing as early as 2008 or 2009. Dr. Donoghue, the chief developer, is co-founder and chief scientist of Cyberkinetics. Some of the paper’s other authors work at the company, while still others are from academic or medical institutions including Like his performance in other tasks, Mr. Nagle’s control of the computer cursor was not particularly smooth. When his goal was to guide the cursor from the center of the screen to a target on the perimeter, he hit the target 73 to 95 percent of the time. When he did, it took 2.5 seconds on average, but sometimes much longer. And the second patient tested with the implant had worse control than he, the paper said.
By contrast, healthy people moving the cursor by hand hit the target almost every time, and in only one second.
Dr. Jonathan R. Wolpaw, a researcher at the New York State Department of Health, said the BrainGate performance did not appear to be substantially better than that of a noninvasive system he is developing using electroencephalography, in which electrodes are placed outside the scalp.
“If you are going to have something implanted into your brain,” Dr. Wolpaw said, “you’d probably want it to be a lot better.”
Dr. Donoghue and other proponents of the implants say they have the potential to be a lot better, because they are much closer to the relevant neurons than are the scalp electrodes, which get signals from millions of neurons all over the brain.
One way to improve implant performance was suggested by another paper in the same issue of Nature. In a study involving monkeys, Krishna V. Shenoy and colleagues at Stanford University eavesdropped not on the neurons controlling arm movement but on those expressing the intention to move, which occurs earlier and would make the system work faster.
“Instead of sliding the cursor out to the target, we can just predict which target would be hit and the cursor simply leaps there,” said Dr. Shenoy, an assistant professor of electrical engineering and neurosciences. He said the system could operate at the equivalent of typing 15 words a minute, about four times as fast as the devices produced by Cyberkinetics and Dr. Wolpaw.
After more than a year, Mr. Nagle had his implant removed so he could undergo another operation, which allowed him to breathe without a ventilator. He can control a computer by voice, so he does not really need the implant. But he said he was happy he had volunteered for the experiment.
“It gave a lot of people hope,” he said.