This article was taken from the May 2015 issue of WIRED magazine. Be the first to read WIRED's articles in print before they're posted online, and get your hands on loads of additional content by subscribing online.

Cathy Hutchinson is completely focused on the glass in front of her. Eyebrows knitted and lips pursed, she gradually brings it towards herself. She falters -- the glass circles the table jerkily -- but she manages to correct it and brings the vessel closer to her mouth. The room at her Massachusetts home -- a one-storey building that she shares with other disabled residents near the border with Rhode Island -- is silent, except for the hum of machines. Hutchinson positions the glass just below her lips, tilts it and sips her drink from a straw. She carefully steers the glass back to the table in front of her and puts it down with a thump. The room erupts in applause. Hutchinson sits motionless in her motorised wheelchair; she only smiles.

Hutchinson, 58, is the first almost-fully paralysed person to manoeuvre a vessel to her mouth and drink from it -- and she's completed that action solely using her mind. Hutchinson is tetraplegic; she was paralysed from the neck down, leaving her unable to speak, after a brainstem stroke on a damp May morning in 1996. Hutchinson can still control her expressive eyes, swallow, breathe and smile, which she does often, her eyes crinkling up in mirth. "I remember exactly when I had the stroke, because it was my niece's birthday, May 5," Hutchinson says in December 2014, using Tobii eye-tracking technology to type words painstakingly on a computer screen. "I remember feeling very nauseous and my right foot shaking violently, then I remember waking up to my son helping me walk up the stairs." The stroke left her "locked in" -- unable to move or speak, but still able to sense pain, heat and cold, and think lucidly. Using a blue-and-silver prosthetic arm, this is the first time in more than a decade that she has fed herself.

Protruding from the top of Hutchinson's head is a black plastic nub, resembling a tiny fez. Extending from it, a thick cable links to a refrigerator-sized cart of electronics. The machine hums loudly, screens displaying a pattern of spikes spitting out loud static. These are Hutchinson's neurons signalling to each other; the cart is reading her thoughts.

Hutchinson is participant "S3" in a clinical trial testing paralysed people's ability to manipulate computer cursors, prosthetic arms, even wheelchairs, using their brains. The company behind the trial is called BrainGate, and it makes a brain-machine interface that eavesdrops on the neurons in the brain's motor cortex, decodes their intentions and turns them into -instructions for an external device.

Hutchinson joined the BrainGate trial because of her best friend Paula, who heard about it at the Rhode Island hospital where she worked as a nurse. It has run since 2004, spearheaded by Brown University in Providence, Rhode Island, with Stanford University and Case Western Reserve University. The leaders are 65-year-old ^-neuroscientist John Donoghue and neurologist Leigh Hochberg at Brown, who have tested BrainGate's system in nine humans so far. "We asked [Hutchinson] to imagine moving her hand in specific ways," says Hochberg. "And the cells were firing faster and faster, as she thought about it. It sounds like music."

The unlikely home of the BrownInstitute for Brain bet365体育赛事 is a shingled Victorian house with large, bracketed eaves and wooden balconies at the corner of a busy street in Providence, Rhode Island. "What we're really trying to do," explains Donoghue to WIRED in his spacious, wood-panelled office, "is replace all of Cathy's connections between her brain to her brainstem, her spinal cord and ultimately back to her body."

Donoghue's first brush with aberrant human movement was at the age of eight, when he was diagnosed with Leggs-Calvé-Perthes disease, a childhood hip disorder caused by disrupted blood flow to the thigh. The illness limited his mobility until he was 13, leaving him with a slight, dragging limp that is still visible today. A decade later, his first job was at the Walter E Fernald Developmental Center, an institution for children with -developmental disabilities, where he worked with residents with far more serious movement disorders than his own, including cerebral palsy. "Both these experiences affected my thoughts on movement and the brain; they gave me insight into the burden of not being able to move," Donoghue says. "I think my interest was shaped by this, but also by an inquisitiveness about how things work and why humans are different -- what's special about our brain."

As a graduate student at Brown in Providence, Donoghue was fascinated by trying to decipher neuron circuits. He was able to record individual neurons in rats, but that wasn't enough to unravel their signals. "Leon Cooper, the Nobel Prize-winning physicist here at Brown, was really big on saying, 'If we just had enough data, we can figure it out', so he was thinking about big data in 1976," he says. "I was inspired and said, well, how do we do that? And the answer was to use multiple electrodes."

For almost two decades, Donoghue made devices to record from multiple neurons simultaneously. But he didn't hit upon the solution until 1992, when he met Richard Normann, an inventor and scientist who had just created the Utah Electrode Array, a thin sheet of silicon embedded with 100 conductive platinum-tipped needles that could sit inside the cortex of a human brain. A version of this original chip is still used in their experiments today, produced by Normann's spin-off company Blackrock Microsystems in Utah. "We started putting electrodes into monkey brains, and began to understand how thought patterns are generated. We realised we had a technology to record the brain's signals that was better than anything we had previously," Donoghue says. "That's when we thought, 'Hey, we could put this in a human'."

In 2002, Donoghue published a paper in Nature that showed a monkey playing a video game with its brain, using a rudimentary version of BrainGate. "It was the first time anyone had done this with a monkey," Donoghue says. "It was like being at the edge of a sci-fi movie." By the next year, Donoghue had tested the system in 22 monkeys and obtained approval from the US Food and Drug Administration to move ahead and test BrainGate in humans. Now all he needed was a volunteer.

Massachusetts native Matt Nagle was a 6’2" high-school footballer when he was stabbed in the neck during a brawl in 2001; his spinal cord was severed from his brain, paralysing his limbs. When his mother read about the BrainGate trial, Nagle, then 24-years-old, begged to become part of it. He was convinced it would cure his paralysis. Nagle was implanted with a chip on June 22, 2004; it remained inside his brain for a year. Although Nagle died from his injuries in 2007, he accomplished many firsts while part of the trial. He could control a computer cursor, using it to press buttons to change TV channels, check email and play games. He could even draw onscreen, and became the first person to control a rudimentary prosthetic arm with his mind. "Matt really was the six-million-dollar man -- we raised $6 million (£3.6 million) and spent it in that one year, to turn him into a bionic man," Donoghue says.

Listen to WIRED's BrainGate podcast special here.

Despite the dizzying complexity of the one kilogram of jelly-like mass that controls us, we don't need to understand it entirely to be able to interpret specific thoughts. "If you do smart sampling you can get enough of an idea to understand what's going on," Donoghue explains. "You only need part of a neuron circuit to decipher how the brain is thinking about movement." Based on his own monkey studies, and cumulative years of neuroscience research, Donoghue was able to pick out the area of the motor cortex that controlled the arm and hand. "It's really remarkable how little we know about how we move. What am I doing now? I'm holding my hand in the air, wiggling my finger back and forth -- which seems incredibly simple, but it's amazingly complicated in the brain," he says. "It's all done with just three Watts and only a few hundred million neurons."

The BrainGate setup is laborious: the voice of a dozen neurons is transmitted from Hutchinson to the cart of machines, where it is converted into spikes of data on a screen. "The general idea is you have these spike times coming in, so, since you know the exact timing of all the spikes of all these channels, you can build a model," says PhD student Jacob Komar, who works on engineering the hardware. "I'll ask the participant to, for a specified time, imagine moving their hand to the left, then to the right, and I'm going to record and look at what that neural data looks like. Once the model is built, you just look at the incoming data and then run it through the model." When the patterns match, the computer instructs an arm or cursor to move exactly as Hutchinson imagined it.

It takes practice to train a machine to understand brain commands. First, it has to decipher the unique instructions of a person's motor cortex. Hutchinson spent a few minutes at each session just imagining her arm and hand moving back and forth, flexing, rotating and grasping. In order to ensure that the researchers knew exactly what she was visualising, they asked her to watch a cursor moving on a screen and imagine her hand was moving it. The BrainGate algorithm began to learn her brain's unique signature, based on her neural signals. This became its training set.

The first device Hutchinson controlled with her mind was a cursor on a two-dimensional screen. At first, the cursor moved haltingly, sometimes wildly. But as the researchers calibrated the algorithm to Hutchinson's brain, she was able to type on a keyboard, open emails and even play games such as "Neural Pong", similar to the Atari game. By comparing data from consecutive sessions, researchers fine-tuned the machine algorithm continuously.

Developing the decoding algorithm that reads Hutchinson's mind is the job of Beata Jarosiewicz. An associate professor at Brown University, Jarosiewicz's goal is a self-calibrating algorithm that automatically adjusts to a person's brain and records data instantly, without needing a technician. "The neural tuning and the firing rates continuously change, because the brain is this squishy thing -- there's blood flowing through it, there's respiration, all these things make the brain move a little," she says. "When the neurons that we're recording are in the order of ten microns in size, even tiny little motions of the brain will make the signals change drastically, so we have to keep recalibrating the decoder." Once the same experiment -- say, controlling a cursor to type on a virtual keyboard -- is repeated four or five times, the decoder has been calibrated and can be used for the rest of the day, but each day requires a freshly calibrated algorithm.

Going from a 2D cursor to a prosthetic arm moving freely in 3D space was a leap. To calibrate the system, the researchers asked Hutchinson to watch a humanoid robotic arm -- made by the German space agency DLR -- perform a range of actions with a shoulder, elbow and wrist, and imagine doing the same. "Our first steps were to define a set of possible grasps: a whole-hand grip versus a key grip which is like a pinch, versus a chug grip which might be three fingers for throwing a ball," says Jarosiewicz. "We can use that classifier to figure out first if the person is attempting a grasp, and then which of the possible grasps it is by seeing what the neural activity is most similar to among those patterns." In three weekly sessions, Jarosiewicz would watch closely as Hutchinson played various games. Her favourite, Hutchinson says, involved little pink foam balls in black cones that sat at the edge of a rod. When activated, they would pop up one by one at different heights, and she had to move the robot hand to grasp them. The results were published in May 2012 in Nature. They showed that Hutchinson and a second BrainGate participant -- 69-year-old Connecticut resident Robert Veillette, another stroke survivor -- were able to grasp the balls a significant proportion of the time.

Jarosiewicz's next step is to make the algorithm more precise by decoding the brain's intention to move before it actually gives the command. From participants' data she was able to figure out at what point their brains started planning a certain movement. This would ultimately give the BrainGate user smoother control. "When you're involved in the nitty-gritty details of code writing and calibrating, you lose sense of the big picture, which is enabling people to control things with their brains," she says. "Witnessing that, interacting with Cathy, it puts the whole thing into perspective."

The BrainGate technology works in a rudimentary form, but there is one major drawback -- the person must be connected via a wire. This means it's unsuitable for daily, domestic use. "It needs a trained individual to attach it and monitor it," Donoghue says. "Once we have a wireless, implanted system, people can work with it all the time, because the safety concern of the tethering has gone away. They can be practising all the time, probably getting better and better at neural control."

Brown University neuro-engineer Arto Nurmikko may have the solution. In his office, the Finnish ex-naval commander holds out a small, implantable device that would replace the bulky plug that sticks up through Hutchinson's skull. "We have designed microchips that take weak neuron signals from a monkey's brain, and amplify, sort them out and then wirelessly string them out in real time," Nurmikko says. "We are in the hundred-megabits-per-second wireless data range, which is possibly faster than your internet speed at home."

The data is only transported over a few metres at most. "So we have already reached a level of wireless data transmission which exceeds anything that anybody else has done in a freely moving monkey," Nurmikko says. "This means that it's not cabled and so, hypothetically, a human can move around the house in a wheelchair while using this." In Nurmikko's lab, two research students demonstrate the titanium pin. "It has all the same guts as that external device, except it is screwed to the top of the skull and the skin closed," says Komar. "You no longer have an open port, which is a large risk for infection. Someone could have one of these implanted and use it for a long time and not worry."

The team is working to achieve its safety approvals from the US Food and Drug Administration before it can test it on humans. "Is the wireless chip available yet?" Hutchinson asks. Not yet, she is told. But when it is, would she have it implanted? A few seconds tick by as she types rapidly with her eyes. "Absolutely."

BrainGate is one of three groups testing out a brain-machine interface for prosthetic control. Similar clinical trials are being run by the California Institute of Technology and the University of Pittsburgh. The Pittsburgh group uses the same microelectrode chip as BrainGate, but has developed its own decoding algorithm. It has had one participant so far, who was enrolled for two-and-a-half years -- Jan Scheuermann, a 55-year-old producer of murder-mystery parties and mother of two from Pittsburgh. Ten years ago, Scheuermann developed spinocerebellar degeneration, which left her paralysed from the neck down.

In video interviews, she is full of praise. "I used to have to think, 'Up, clockwise, down, forward, back'; now I just look at the target and [the arm] goes there. It's not a matter of thinking about direction any more, my brain knows what to move," she says. Scheuermann was able to pick up objects of different shapes and sizes and pour liquid into a glass. "She could also grab a bar of chocolate out of a researcher's hand and feed it to herself," says lead investigator Jennifer Collinger, 33. Her team published their findings in the Journal of Neural Engineering in December 2014, showing that Scheuermann had shown ten unique grasps.

As more patients test the system, scientists are learning more about the damaged nervous system. The nine BrainGate patients, who have a range of disabilities, have each taught Donoghue's team something significant. "I can say it works not only in the two stroke patients we looked at, but also in ALS (amyotrophic lateral sclerosis), and in the two spinal-cord-injured people we saw," he says. "You can record their neurons; they look as they would do when you were really moving. And this is true even in ALS patients, where there is neural degeneration going on. It's remarkable."

If you ask Donoghue about his ultimate dream, the answer is immediate: to allow a person not only to control a prosthetic arm by thought, but to control their real arm. Robert Kirsch could make this possible. The 56-year-old bioengineer at Case Western Reserve University is part of the BrainGate family. He directs the Functional Electrical Stimulation Center in Cleveland, Ohio, which focuses on restoring abilities to people with stroke, spinal cord and other neurological injuries. His big idea: stimulating muscles with electricity to reactivate paralysed limbs and organs.

The FES system, which is implanted in thousands of patients worldwide, resembles a pacemaker that is embedded in the upper chest or lower abdomen. From here, tiny wires go out to muscles that need stimulation, either coiling up outside the muscle or wrapped around a nerve trunk in the limb. "We can apply electrical stimulation to muscles so they contract," he explains. But the stimulation is usually controlled by a switch or an external trigger. "We want to give control to the person. That's where BrainGate comes in."

This year, Kirsch's team will try out this system in parallel with BrainGate, on an army veteran from Ohio; the patient has a severe spinal cord injury and complete arm paralysis. Using $2.5 million from the US National Institutes of Health, Kirsch's team will inject electrode wires into his arm. He already has the BrainGate chip implanted in his motor cortex. "We are going to try and give him a very simple hand grasp. We are also going to give him elbow flexion and extension, some shoulder motion and the ability to rotate the forearm muscles so he can position the hand in different ways -- palm up to palm down," Kirsch says. This system will be entirely controlled through BrainGate's system -- so all the patient will have to do is think about moving, and his arm should obey in real time. "This is the moonshoot," Kirsch says.

The team has already simulated the experiment on Hutchinson. "We made a complete model of an arm, so all the muscles behave like the muscles in your arm, but it's an animation driven by a complex mathematical model," Donoghue explains. "We hooked Hutchinson up to that and her job was to drive that animated arm from point to point. It worked!" This simulation has now been run with multiple BrainGate participants, according to Kirsch -- and worked every time. "If this technology works by really being able to connect the brain back to a limb, we will have something that impacts the lives of huge numbers of people," Donoghue says. "That's our vision, the long-term hope."

On a wet December afternoon, WIRED drives to see Hutchinson at her home. The house is inviting, with a Christmas tree at the door, two garden gnomes and cushioned rocking chairs set out on the porch. Hutchinson is waiting in the living room, wearing a red jumper, bright pink socks and shoes with neon laces. A camera mounted on her computer screen tracks a round, white target on the bridge of her glasses, and every time her gaze rests on a letter on the on-screen keyboard, the cursor clicks. A few moments after WIRED arrives, her mechanical voice reads out, "Hello. I'm so glad to meet you."

Being the first non-verbal participant in the trial didn't hold Hutchinson back -- she completed the BrainGate trial in 2012, after almost 1,000 days of testing, the longest-running participant in the programme. "I felt I reconnected with my body. It was a comfortable return to the way I was before my [stroke]," Hutchinson says.

If the BrainGate system becomes commercially available to her again, her goal is to use it to regain some part of her independence. "When I was told I would never [move] again I wanted to prove them wrong. After the BrainGate trial, I felt a sense of accomplishment."

Madhumita Venkataramanan is WIRED's associate editor. She wrote about privacy and data in 11.14

This article was originally published by WIRED UK