“Matt Nagle is paralyzed. He’s also a pioneer in the new science of brain implants.”
MATTHEW NAGLE IS beating me at Pong. “O, baby,” he mutters. The creases in his forehead deepen as he moves the onscreen paddle to block the ball. “C’mon – here you go,” he says, sending a wicked angle shot ricocheting down the screen and past my defense. “Yes!” he says in triumph, his voice hoarse from the ventilator that helps him breathe. “Let’s go again, dude.”
The remarkable thing about Nagle is not that he plays skillfully; it’s that he can play at all. Nagle is a C4 quadriplegic, paralyzed from the neck down in a stabbing three years ago. He pilots a motorized wheelchair by blowing into a sip-and-puff tube, his pale hands strapped to the armrests. He’s playing Pong with his thoughts alone.
A bundle of wires as thick as a coaxial cable runs from a connector in Nagle’s scalp to a refrigerator-sized cart of electronic gear. Inside his brain, a tiny array of microelectrodes picks up the cacophony of his neural activity; processors recognize the patterns associated with arm motions and translate them into signals that control the Pong paddle, draw with a cursor, operate a TV, and open email.
Nagle, 25, is the first patient in a controversial clinical trial that seeks to prove brain-computer interfaces can return function to people paralyzed by injury or disease. His BCI is the most sophisticated ever tested on a human being, the culmination of two decades of research in neural recording and decoding. A Foxborough, Massachusetts-based company called Cyberkinetics built the system, named BrainGate.
After we play Pong for a while, I ask Nagle to try something I’d seen him do in a video: draw a circle. This is more fundamental and difficult than playing Pong. Drawing a circle freehand is a classic test of motor function, a species marker. Legend has it that Leonardo da Vinci was among the few humans who could sketch a perfect one.
Today, Nagle barely gets to imperfect. The line keeps shooting off the screen or crossing itself. Maybe it’s my presence or fatigue or some subtle shift in Nagle’s brain chemistry due to who knows what. Abe Caplan, the Cyberkinetics technician overseeing the computer gear that dominates a corner of Nagle’s room at New England Sinai Hospital, urges him on softly.
“I’m tryin’, dude,” Nagle says, cursing softly. “C’mon, you bitch.”
Caplan taps on one of his keyboards to adjust a setting, averaging the system’s motion prediction over a longer time to smooth out the line. Finally, Nagle manages to produce a collapsed half circle. He’s exhilarated but clearly exhausted. As they finish the session, Caplan nods his head toward the computers and says, “Want to hear it?”
He flicks a switch, and a loud burst of static fills the room – the music of Nagle’s cranial sphere. This is raw analog signal, Nagle’s neurons chattering. We are listening to a human being’s thoughts.
Roughly the size of a deflated volleyball, your brain weighs about 3 pounds. Its 100 billion neurons communicate via minute electrochemical impulses, shifting patterns sparking like fireflies on a summer evening, that produce movement, expression, words. From this ceaseless hubbub arose Ode to Joy, thermonuclear weapons, and Dumb and Dumber.
Nobody really knows how all that electricity and meat make a mind. Since Freud, scientists have wrangled over “the consciousness problem” to little effect. In fact, it’s only in the past 20 years that researchers have learned how to listen in on – or alter – brain waves. Neuroscientists can record and roughly translate the neural patterns of monkeys, and thousands of humans with Parkinson’s disease and epilepsy have cerebral pacemakers, which control tremors and seizures with electrical impulses.
John Donoghue, head of neuroscience at Brown University and the founder of Cyberkinetics, eventually wants to hook BrainGate up to stimulators that can activate muscle tissue, bypassing a damaged nervous system entirely. In theory, once you can control a computer cursor, you can do anything from drawing circles to piloting a battleship. With enough computational power, “everything else is just engineering,” says Gerhard Friehs, the neurosurgeon from Brown who implanted Nagle’s device.
For now, that engineering remains a challenge. Cyberkinetics is just one of a dozen labs working on brain-computer interfaces, many of them funded by more than $25 million in grants from the US Department of Defense, which frankly envisions a future of soldier-controlled killer robots. Before that can happen, BCIs must become safe enough to be implanted in a human, durable enough to function reliably for years, and sensitive enough to pick up distinctive neural patterns. Many physicians doubt useful information can ever be extracted from neural activity, and some who believe in the promise of BCIs worry that putting one into Nagle’s head was premature, even reckless, considering less invasive technological options still on the table – electrode-studded skullcaps or devices that rest on the brain’s surface. They worry that a failure could set the entire field back a decade.
“The technology required is very complex,” Donoghue admits. “There are still many issues to be resolved. But it’s here. It’s going to happen. Just look at Matt.”
On July 3, 2001, Matthew Nagle and several friends went to a fireworks display at Wessagussett Beach, 20 miles south of Boston. The 6′ 2″, 180-pound Nagle had been a football standout at Weymouth High and was a devoted Patriots and Red Sox fan. That summer he was driving a van delivering kitchenware and had just passed the postal service exam. As Nagle and his buddies were leaving the beach, one of them got into a scuffle. Nagle jumped out of the car to help his friend. “The last thing I remember is sitting in the car,” Nagle says. “My friend told me I went over to this guy and he pulled a knife.”
The 8-inch blade entered the left side of Nagle’s neck just under his ear, severing his spinal cord. Nagle spent four months in rehabilitation before moving back to his parents’ house. He can’t breathe without a respirator, and though he has at times managed to wiggle a finger, doctors give him no chance of regaining the use of his limbs. Nagle’s mother ran across the BrainGate experiments while researching spinal-cord injuries online, and she brought him an article about Cyberkinetics from The Boston Globe. Nagle, who had been trying unsuccessfully to wean himself from the ventilator, begged his doctors for the chance to be the first subject. “My mother was scared of what might happen, but what else can they do to me?” Nagle rasps, jutting his chin at his wheelchair. “I was in a corner, and I had to come out fighting.”
Nagle’s doctor contacted the people running the trial. “A week later I got a call,” Nagle says. “I told them, ‘You can treat me like a lab rat, do whatever. I want this done as soon as possible.'”
Nagle turned out to be an ideal subject – young, strong-willed, and convinced that he will walk again. The only problem: Because Nagle’s brain had been cut off from his spinal cord, no one knew if he could still produce the coherent neural signals necessary for movement. It wouldn’t matter how well the BrainGate could read patterns if Nagle’s brain was broadcasting noise. Donoghue’s experiments had used healthy, fully functioning monkeys.
“That was the great unknown,” says Donoghue. “When he thinks ‘move left,’ were we going to get one neuron firing one time, 20 the next time? Or maybe not anything? Could he still imagine motion enough to make those cells modulate, to change those spikes?”
There was only one way to find out: implant the chip.
On the morning of June 22, 2004, Friehs – an expert in gamma knife surgery, which uses focused radiation to treat brain diseases like Parkinson’s – opened Nagle’s skull using a high-speed drill called a craniotome. With a number 15 scalpel, he carefully sliced through the protective membranes that surround the brain.
The living brain is a gory sponge, a mass of blood vessels shot through with a delicate mesh of fiber. Magnetic resonance imaging allowed Friehs to plot in advance the region on Nagle’s motor cortex most likely to provide readable arm-movement signals to the BrainGate. One revelation of BCI research has been that brain functions are highly distributed: Any spot within a given region can provide neural signals to operate a prosthetic. Get the BrainGate to the right place, and it would pick up signals not just from the neurons it touches, but from important neural clusters nearby as well. Using a small pneumatic inserter, Friehs tapped in the tiny array – 100 electrodes, each just 1 millimeter long and 90 microns across at its base. Friehs closed Nagle’s skull with titanium screws, leaving a tiny hole. Through that he threaded gold wires from the array to an external pedestal connector attached to Nagle’s skull. Matthew Nagle was now part biological, and part silicon, platinum, and titanium.
It took Nagle three weeks to sufficiently recover from surgery to start learning to use the BrainGate. The first session would answer Donoghue’s foremost question: Could Nagle’s brain still produce usable signals?
As Donoghue looked on, Caplan asked Nagle to think left, right, then relax. “When we watched the system monitor, we could plainly see that neurons were briskly modulating,” Donoghue recalls. “My reaction was ‘This is it!'”
Nagle had even more confidence. “I learned to use it in two or three days – it’s supposed to take 11 months,” he says. “I totally knew this was going to work.”
Four months after the operation, I watched Caplan take Nagle through a typical training session. He tracked Nagle’s mental activity on two large monitors, one of which displayed a graph of red and green spiking lines. Each spike represented the firing of clusters of neurons. As Nagle performed specific actions in his mind’s eye – move arm left, move arm up – the electrodes picked up the patterns of nearby neuron groups. Then BrainGate amplified and recorded the corresponding electrical activity. Over dozens of trials the computer built a filter that associated specific neural patterns with certain movements. Later, when Nagle again mentally pictured the motions, the computer translated the signals to guide a cursor.
Then they moved on to some more complicated neural gymnastics, with Nagle willing a large green cursor onto a picture of a money bag that popped up in different spots onscreen. Sometimes the cursor moved shakily, or shot off course; sometimes it landed almost immediately where Nagle wanted to place it. It was like watching a 3-year-old learn to use a mouse. Slowly, the cursor grew more controlled. The machine was memorizing Nagle’s characteristic neural firing patterns.
“Let’s see what you can do with this thing, Matt,” Caplan said.
Nagle turned the TV on and off and switched channels (trapped in his hospital room, he’s become a daytime-TV addict). Then he opened and read the messages in his dummy email program. “Now I’m at the point where I can bring the cursor just about anywhere,” he said. “I can make it hover off to the side, not doing anything. When I first realized I could control it I said, ‘Holy shit! I like this.'”
What are you thinking about when you move the cursor? I asked.
“For a while I was thinking about moving the mouse with my hand,” Nagle replied. “Now, I just imagine moving the cursor from place to place.” In other words, Nagle’s brain has assimilated the system. The cursor is as much a part of his self as his arms and legs were.
Fresh out of Boston University undergrad, John Donoghue went to work at the Fernald School, a facility for mentally handicapped children in Waltham, Massachusetts. His boss, Harvard neuroanatomist Paul Yakovlev, had studied in Russia under Ivan Pavlov, who turned his conditioned-reflex work with dogs and bells into the first map of the human motor cortex – left foot controlled here, right foot over there, and so on. At Fernald, Yakovlev and assistants like Donoghue spent hours a day slicing human brains, mostly damaged or defective ones, into 1/1,000-inch sections for study.
Whenever he looked up from his microscope, Donoghue could see the results of those defects roaming the halls of Fernald. Looking for ways to help people recover from cerebral injury, he became interested in plasticity, the brain’s ability to adapt and form neural pathways. “By understanding how the plasticity of the brain can be captured and controlled,” Donoghue says, “I believed we could promote the recovery of function in severely impaired patients.” He set out to learn the grammar and syntax of interneuron communication.
Ten years later, with a PhD in neuroscience from Brown, Donoghue was using microelectrodes to record neural activity in rats, one neuron at a time. It wasn’t enough. “Listening to just one neuron is like hearing only the second violinist,” says Donoghue. “With multiple neurons, it’s like hearing the whole orchestra.” The problem was, nobody knew how to record multiple neurons reliably.
In 1992, he went to the Society for Neuroscience meeting in search of a solution. “I went to every poster session that had innovative multielectrode recording methods,” Donoghue says, “and finally I found Dick Normann.”
Normann is the kind of person who shows up often in the history of scientific revolutions: more tinkerer than basic researcher, more inventor than visionary. In the early 1990s he created the Utah electrode array, a thin substrate of silicon that rests on the surface of the cortex, embedding platinum-tipped electrodes into the gray matter. Normann designed it to send signals into the brain, as part of a visual prosthetic. Donoghue realized it could also be an uplink.
A few weeks after the conference, he visited Normann’s lab in Utah. “We placed the implant in a cat, went to lunch, and let the animal recover,” Donoghue says. “When we came back we had good signals. It worked!”
Other researchers were chasing the same goal. In 2002, Miguel Nicolelis, a neurobiologist at Duke, provided the best evidence yet of the brain’s plasticity. He and his team plugged 86 microwires into the brain of a monkey and taught the animal to use a joystick to move an onscreen cursor (the reward: a sip of juice). After the computer had learned to interpret the animal’s brain activity, Nicolelis disconnected the joystick. For a while, the monkey kept working it. But he eventually figured it out. The monkey dropped the joystick and stopped moving his arm; the cursor still moved to the target. As the monkey calmly downed another swallow of juice, Nicolelis’ lab fell silent in awe. The mammalian brain could assimilate a device – a machine.
Still, the Utah array had a few advantages over other designs. It “floats” with the movement of respiration and blood pumping, remaining stable in relation to the surrounding neurons. Years of refinement stripped much of the metal away, making it more-biocompatible and less likely to cause scarring. By 2003, Donoghue and Normann had tested the device, now called BrainéGate, in 22émonkeys. That got them the FDA approval they’d been looking for: a small trial with five human subjects. Their first was Matthew Nagle.
At a conference in 2002, Anthony Tether, the director of Darpa, envisioned the military outcome of BCI research. “Imagine 25 years from now where old guys like me put on a pair of glasses or a helmet and open our eyes,” Tether said. “Somewhere there will be a robot that will open its eyes, and we will be able to see what the robot sees. We will be able to remotely look down on a cave and think to ourselves, ‘Let’s go down there and kick some butt.’ And the robots will respond, controlled by our thoughts. Imagine a warrior with the intellect of a human and the immortality of a machine.”
Some scientists have further suggested that implants designed to restore cognitive abilities to Alzheimer’s and stroke victims could enhance the brainpower of healthy people. There’s talk of using BCIs to stifle antisocial tendencies and “program” acceptable behavior.
Of course, as spooky as these scenarios may be, they first require that BCIs actually work. A few months into the trial, Donoghue’s device looks safe enough, and it clearly reads useful (albeit rudimentary) signals from Nagle’s brain. No one knows if it will still work a year from now, much less five or ten. If Nagle’s brain rejects the implant, other researchers might shy away from implants altogether. “The key is what functions you want to restore,” says Duke’s Nicolelis. “If you only want to play a videogame or turn on your TV, you don’t need to get into the brain. It’s really a question of cost-benefit and risk. That’s why I’m concerned about things moving too fast; we need to know more about the risks of leaving these things in people’s heads for long periods. We don’t know much yet.”
That’s why some BCI researchers are looking for less invasive methods. A team of Austrian researchers taught a quadriplegic patient to open and close a prosthetic hand using an electrode-studded skullcap that picked up electroencephalograms, waves of electricity generated by the entire brain. It was impressive, but the patient required five months of training to pull it off. And this past December, researchers at the Wadsworth Center in Albany, New York, reported that a patient was able to move a cursor around on a monitor using externally detected signals – no implant. Another group has tried an electrode array that rests on the surface of the brain without penetrating the cortex. “I don’t think anybody really knows which method is going to be the safest and still give detailed pictures of brain activity,” says Wadsworth’s Gerwin Schalk.
Donoghue remains convinced that the only way to give people with immobile bodies full interaction with their environment is through embedded electrodes. “No other method gives you the power and clarity you need to transform this noisy signal into something that a patient can use,” he says. “The people who question whether this will really work, I don’t think they realize how much has already been done. We’ve got a 1,098-day monkey, who had a working BCI for almost three years. The question is, how long do you want to keep doing this in monkeys?”
Nagle has the fervor of the saved. He’s convinced that BrainGate will restore him to movement. “It’s just around the corner,” he says. “I know I’m going to beat this.” Already he can control a prosthetic hand – it’s an eerie sight, rubberized, disembodied fingers grasping and relaxing on a tabletop a few feet from the motionless Nagle. “Thirty-nine months I’ve been paralyzed,” he says. “I can stick with it another two years, till they get this thing perfected.”
But controlling a robotic hand is a long way from walking. Near the end of our interview, I ask Donoghue if he thinks he’s giving Nagle false hope for a cure.
“I don’t know that it’s false,” he replies. “It’s hope.”
