Read an Excerpt
The Brain Electric
The Dramatic High-Tech Race to Merge Minds and Machines
By Malcolm Gay Farrar, Straus and Giroux
Copyright © 2015 Malcolm Gay
All rights reserved.
ISBN: 978-0-374-70962-4
CHAPTER 1
BYPASSING THE BODY
Like generations of neurosurgeons before him, Leuthardt had implanted the clinical grid of electrodes to measure the brain's action potentials, tiny pulses of electricity neurons emit each time they exchange information with nearby cells. The technique, known as electrocorticography, or ECoG, doesn't record individual neurons. Rather, the grid's electrodes pick up the collective activity of the thousands of neurons that lie beneath them, registering their summed rhythms as brain waves.
By tracking Brookman's brain waves, neurologists could observe when his normal brain activity was interrupted by the beginnings of a seizure. Instead of the normal up-and-down signal of, say, an alpha wave, Brookman's brain would become erratic as a cluster of neurons began firing in synchronic bursts. The renegade cells would inevitably recruit more neurons to their ictal cause, triggering Brookman's brain waves to grow chaotic as the epileptic storm pulsed across the brain.
It had been crucial during the implantation surgery for Leuthardt to install sensors over the entire seizure focus area. Ample coverage would enable neurologists to divine an epileptic source by noting, essentially, that brain waves first became erratic below a specific electrode. "It's kind of like a murder mystery," Leuthardt said. "You're trying to find the criminal. You can use external studies like MRI and PET scans. Those will tell you the general region — the criminal's zip code. But now we need to find his address. Electrodes give us very specific localization."
Using a similar technology, neuroscientists have long listened in on individual neurons with penetrating electrodes. Piercing their hair-thin wires into the brains of monkeys and rats, these scientists spent years searching for repeated firing patterns in individual cells. As the animals performed repetitive gestures like pressing a lever for a juice reward, the researchers found that specific neural patterns were associated with the physical action. There was a lot of room for error, but the neural patterns were often consistent. They were also repeatable: a neuron would erupt in a similar firing pattern each time the animal performed the bar-pushing action to receive its juice reward.
Around 2000, however, a handful of researchers began transforming this information into a brain-computer interface: whenever the cell produced the desired firing pattern, the computer would execute a physical command. Of course, most animals were none the wiser and would continue to press the lever to receive their juice reward. But over time, they realized they didn't need to physically press the lever to get their reward. They had only to think about it.
The clinical application for brain-computer interfaces seemed clear. Physical paralysis essentially is a communication error between the central nervous system and the branching network of peripheral nerves that radiates from the spine. In healthy bodies, the brain sends signals to the spine (Walk! Sit down!), which in turn relays the details to the peripheral nervous system (Lift the right leg! Bend at the hip!). Paralysis occurs when some relay point along the route stops working — be it through spinal cord injury, amputation, stroke, or some other form of neuron death. The brain may continue to send signals, but the message never arrives.
By the time Leuthardt entered the field in the middle of the first decade of the twenty-first century, researchers had already shown that by sinking individual electrodes into the brains of monkeys, rats, and some humans, they could tap movement at its source. A single neuron provided enough information to create basic computer commands, bypassing an animal's peripheral nervous system to give subjects modest neural control over machines. In some cases, this meant linking neural patterns associated with moving a joystick to give monkeys direct control over a cursor. In others, it meant controlling a robotic arm or mentally pressing a lever to deliver a juice reward. The control was basic, enabling the animals to move a cursor to the left or the right, or a feeding bar up or down.
But these early researchers were tapping only a handful of neurons. What sort of control could they produce if they harnessed, say, a hundred or even a thousand neurons in a person? Could they give people full control over a computer, enabling them to send e-mail or surf the Internet using only their thoughts? Could they re-create the elegant movements of the human arm? Harnessing thousands of neurons, could researchers craft a full-body exoskeleton for quadriplegics or soldiers? And how about abstract thoughts? Given ample neural access, could we bypass spoken language altogether, doing away with its ambiguities and miscommunications in favor of direct neural exchange? In the realm of memory, could brain-computer interfaces enable total recall? Could they deliver new sensory modes like infrared or X-ray vision? What was to stop these technologies from enhancing our own cognition? Could we selectively stimulate the brain to boost learning?
Those early brain-computer interfaces might have been confined to basic physical commands, but Leuthardt saw in them a union that could fundamentally change our understanding of the brain. "I saw neuroprosthetics in the very early, seminal stages," he said, "and I thought, this is it. This is the future."
Leuthardt was not alone. The field was already thick with speculation that scientists could craft a neural augment for people with paralysis. In 1998, an Irish researcher named Philip Kennedy demonstrated that he could endow a man paralyzed from the neck down with rudimentary control of a computer program. One year later, the German researcher Niels Birbaumer used EEG to enable similarly impaired patients to control basic word-processing software, and by 2001 one of the field's titans, a neuroscientist named John Donoghue, cofounded Cyberkinetics, a neurotechnology company aimed at developing commercial brain-computer interfaces. Other researchers were using electrodes to unlock the brains of monkeys. In one headline-grabbing experiment, Duke University's Miguel Nicolelis connected the motor cortex of a rhesus monkey to a robot arm in the next room. Using only its thoughts, the animal harnessed the arm to play a simple video game. "At that moment," Nicolelis wrote, "the cumulative years of research and the hopes of thousands of severely paralyzed people who dreamed of one day regaining some degree of their former mobility became deeply intertwined."
Still, there was a lot of work to do. These early efforts were a far cry from the sort of always-on commercial device Leuthardt envisioned. And that's to say nothing of crafting a brain-computer interface, or BCI, to rival the elegance and diversity of biological movement.
What's more, the interface itself was problematic. Penetrating electrodes might have enabled brain researchers to enter an intimate exchange with the brain's most basic unit — the neuron — but they were also unreliable. Like the rest of the body, the brain abhors foreign objects, and while the platinum sensors created a close union between mind and machine, it was often short-lived. The brain eventually mounted an immune response, dispatching micro-glia, astrocytes, and other proteins to cordon off the offending electrodes. Wrapped in successive layers of scar tissue, the electrodes inevitably lost their sensitivity. Signal quality degraded, sometimes in a matter of months, rendering the implant unusable. "There was no way that was going to work," Leuthardt thought. "If these microelectrodes were not lasting longer than six or seven months, there was no way a neurosurgeon would ever want to put this into a patient commercially."
Electroencephalography, or EEG, was an option, but surface electrodes had their own problems. It was a rare individual who would be willing to spend his life in what amounts to a sensor-studded swimming cap. More important, though, surface electrodes provided only a hazy portrait of the electrical storm raging inside the skull. Placed directly on the scalp, EEG electrodes can't always differentiate between the electricity inside the brain and the electrical pulses that animate the scalp. It leads to a muddy signal, adulterated with muscular electricity and even surrounding electronics.
At the time, researchers confined themselves to either EEG or penetrating electrodes. Those interfaces were fine for the research lab, but Leuthardt was convinced that if he and his fellow scientists were ever to usher in the age of neuroprostheses, they would need to enter the commercial market, crafting a highly sensitive, accurate interface that wouldn't degrade over time.
"That's what got me down the road of ECoG," he said. Unlike penetrating electrodes, the ECoG grids did not pierce the brain. Rather, they rested on its surface and would likely be more stable. Having direct contact with the brain also meant that, unlike EEG, ECoG signals weren't as likely to be contaminated by muscular artifacts from the scalp or nearby electronics. It seemed like the Goldilocks zone: more stable than penetrating electrodes, more precise than EEG. "I've always seen us as being the bed's just right in the sense that this one is too invasive, that one is too noisy, but this one is just right."
If an EEG was like listening to the muffled strains of the neural symphony behind a band shell, penetrating electrodes were like training a microphone on a sole musician or an individual string. ECoG, by contrast, was like listening to a section of the orchestral brain from the first few rows — close enough to tease out the first violins from the second violins.
But here was the real beauty of using ECoG: as a neurosurgeon, Leuthardt already had a built-in population of human research subjects. During the week or so that patients like Brookman were implanted for epilepsy monitoring, they were effectively lying in a hospital bed just waiting to have seizures. The rest of the time? The electrodes simply sat atop the brain, passively recording its electric hum. All the elements were there. Why not use the clinical setup of the epilepsy-monitoring unit to create an entirely new brain-computer interface?
At the time, all but a few neural implants were used for limited periods of time and only in the laboratory. But a neural implant that could pull detailed information from the brain while also sidestepping the glaze of signal-degrading scar tissue? A device like that could form the basis of a commercial implant that would remain in the brain for years. "It became very clear to me that this was the future," said Leuthardt. "It's a whole new universe that opens up — one that can change the human experience."
* * *
To that end, David Bundy arrived at the epilepsy-monitoring unit a few days after Brookman's surgery with a cartload of electronics. As a graduate student in Leuthardt's research lab, Bundy was hoping Brookman would don a sensor-studded glove. He wanted him to flex his fingers so he could calibrate the movement to Brookman's brain activity, the first step in building a BCI.
But Brookman was still recovering from surgery. His eyes fluttered and his head nodded lazily as he slouched semiconscious in the hospital bed. Shirtless, he wore a pair of thin cotton shorts, and his head was wrapped in a turban of gauze dressing, a Gorgon-like mane of wires exiting the right of his skull.
Normally, the tangle of wires that spilled from his head would transmit Brookman's brain waves to a bank of computers down the hall. But Brookman had agreed to be one of Leuthardt's research subjects, and for an hour each day grad students like Bundy connected his cables to their own cart of amplifiers, digitizers, and computers.
Leuthardt kept the amplifier in what's known as a Faraday cage, a wood-framed box wrapped in copper mesh to isolate the device from surrounding electronics. From one side of the cage tumbled a rainbow-colored cascade of wires that linked the amplifier with the leads exiting Brookman's brain. From the other, the amplifier connected to a computer whose screen showed a graph of brain waves from each electrode.
It was a surprisingly ad hoc affair, with single-serving cups of orange juice, travel-sized bottles of Listerine, and toothbrushes still in their wrapping strewn across the room. These signs of the family's vigil were everywhere: a half-eaten cluster of grapes sat on a table next to individual-sized bottles of body wash and shampoo.
Meanwhile, Brookman's mother and aunt watched warily from a pair of vinyl-covered lounge chairs. The room's blinds had been drawn against the morning sun, and a nurse in maroon scrubs sat quietly in the corner, prepared to intervene should Brookman seize during the testing.
It was no empty measure. One day earlier, Brookman's eyes had rolled back in his head and his body stiffened just as the grad students were setting up their equipment. They beat a quick path to the door as Brookman fell into convulsions, the day's research session scuttled.
Now Brookman seemed only slightly awake. "We want to see what your brain signals are doing when you're moving your hand in different ways," said Bundy, explaining how they wanted to correlate the movements of the sensor-laden glove to specific brain waves. "The goal is to help people that maybe have spinal cord injury or amputation so they can have a prosthetic hand."
Bundy told Brookman he'd be following a series of simple prompts to link, or calibrate, his brain waves to the movement of his hand. He'd need to flex his thumb, extend his index finger, and pinch with his thumb and index finger. Once they'd calibrated the glove, they would move on to the task itself: Brookman would think about making a specific hand gesture to mentally control the up-and-down movement of a column on the monitor.
"Does that sound all right?" Bundy asked after explaining the task.
"Yeah," drawled Brookman, only half-awake.
"Does that make sense?"
"Yeeeeaaaah."
But it didn't make sense. Brookman lagged behind the simple prompts, incorrectly pinching or extending his forefinger five seconds after the computer prompt. By then, the computer had moved on to the next prompt, and Bundy had to start the program again after a few failed tries.
"Just flex your thumb and then extend it out," Bundy said. "A pinch would be just bringing your thumb and your finger together — just like this," he said, making an "okay" sign with his right hand.
"Just your thumb, baby," Brookman's aunt interjected. "Keep your hand open and just do your thumb. Are you awake, baby?"
Brookman was, but only barely. Though he normally took a host of antiseizure drugs, neurologists had taken him off his medication to better locate his seizure focus. "We want to make sure we get the seizures, because occasionally we'll put all these electrodes on, and they won't have any seizures," said Brookman's neurologist, Hogan. "We were pretty aggressive."
The neurologist needn't have worried: without medication, Brookman had suffered some twenty-five seizures in the first twenty-four hours following the surgery, more than Hogan had ever seen. Emerging disoriented from these rolling convulsions, Brookman didn't know where he was or what had happened. He'd been wild in his panic, trying to rip the wires from his head and lashing out. It got so bad that at one point the hospital staff restrained him with leather straps.
But now Brookman was dazed and docile with pain relievers. He was meek, eager to work with the researchers, and fearful he would disappoint them.
"Can you understand what he's saying?" his aunt asked.
"Yeeeeaaaah," Brookman moaned as the computer prompted him to make a fist.
"Can you make a fist?" she coaxed as he brought his fingertips slowly to his palm. "Good job!"
"Can you flex your thumb?" Bundy jumped in, following the computer prompt.
"Just your thumb, baby," said his aunt, a woman with spiky brown hair and a peach-colored blouse. "Do it with your thumb."
But Brookman moved both his index finger and his thumb, moving them slowly in unison.
He clearly wasn't up to the task. Brookman's seizures, coupled with the pain medication, kept him semiconscious. He was easily confused, nodding off in the middle of tasks and unable to follow the simple instructions.
"I think we want to just let you rest," Bundy finally said after several failed attempts. "We might try to come back in a little bit."
"I'm ripped up," Brookman apologized.
"We'll let you rest."
"Let me rest to where I can at least see straight," Brookman said. "I'm so tore up right now."
(Continues...)
Excerpted from The Brain Electric by Malcolm Gay. Copyright © 2015 Malcolm Gay. Excerpted by permission of Farrar, Straus and Giroux.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.
