BrainGate, the most sophisticated brain-computer interface yet tested in humans, is being developed by John Donoghue, head of the Brain Science Program at Brown University, through a company he cofounded: Cyberkinetics Neurotechnology Systems. In June 2004, quadraplegic Matthew Nagle became the first patient to receive an experimental implant in a part of his brain that controls hand and arm movement. The implant is a square silicon chip just four millimeters (about 1/6 of an inch) wide, studded with an array of 100 hair-thin electrodes. The chip sits on the surface of the brain, while the electrodes delve midway into the brain's two-millimeter-thick cortex to eavesdrop on neurons that normally signal muscles to move. A bundle of gold wires sends those signals out through a connector affixed to the top of the skull, and to an amplifier the size of a cigar box; they then travel by fiber-optic cable to a dishwasher-size cart of computers. During training sessions for BrainGate, the computer software learns to associate patterns of neural activity with the intent to move a hand in a particular direction; it can use those intentions to pilot a computer cursor or, if all goes as planned, a motorized wheelchair.
Donoghue remembers Nagle's first training session clearly. His team asked Nagle to imagine moving his arm and hand: first to the left, then to the right, to flex his wrist, or to open and close his hand. "To me it was just incredible because you could see brain cells changing their activity," Donoghue recalls. "Then I knew that everything could go forward, that the technology could actually work."
After three years of limb paralysis, it was by no means clear that Nagle's motor cortex could still signal his intentions in a meaningful way. Donoghue's work "profoundly shows that the signals in the part of the brain you'd need to operate these neuromotor prostheses exist years after a spinal cord injury," says neuroscientist Krishna Shenoy of Stanford University, who is working with the BrainGate sensor in monkeys. That finding, he says, along with the apparent safety of the implants, "has catapulted this entire field."
The eldest son of a bricklayer, Donoghue spent several years of his Arlington, Massachusetts, childhood in a wheelchair after developing a painful degenerative bone disease that ravaged his upper leg bones where they form the hip joint. It gave him some perspective, he says, "on what it means to be limited in your mobility and not able to do the things that the rest of the world is doing."
When he started his lab at Brown in 1984—after completing a doctorate there in 1979—it was not to address mobility problems but to ask fundamental questions about the brain. How, for instance, does the brain translate intention into skilled movement? "I want to pick up my coffee cup, and my hand gets shaped and goes over there, picks it up, and grabs it. How does that happen?"
Recordings from just one, or even a few, neurons at a time in rats or monkeys weren't getting him close enough to the answer. So by 1992, he had begun looking for new technology that could detect the activity of many brain cells at once. He soon settled on a multielectrode array—now a component of BrainGate—developed by bioengineer Richard Normann at the University of Utah. Donoghue's original intent was basic brain research, but a powerful idea was taking shape among neuroscientists: coupling this type of sensor to an external device that could help the disabled.