It was the summer of 2010, and Ian Burkhart was sizing up the waves as he swam in the ocean off the coast of North Carolina. He had traveled there on a vacation with a group of friends to unwind after wrapping up his freshman year studying video production at Ohio University. He prepared to dive into an oncoming wave and tumbled into the water. Burkhart was a capable swimmer, but the ocean is unpredictable. The wave slammed him into a sandbar—and that’s when he realized he couldn’t feel his body.
Unable to move, Burkhart was at the mercy of the ocean. His friends quickly realized something was wrong and pulled him from the water. He was brought to a nearby hospital where he underwent emergency surgery. Once he was stable, the doctors gave Burkhart the bad news: His spinal cord had been severed. He could no longer walk, the range of motion in his arms was limited to his shoulder and bicep, and he had almost completely lost his sense of touch.
After spending years working to adjust to his new reality, Burkhart enrolled in an experimental program called NeuroLife at Battelle, a nonprofit research organization in Ohio. The plan was to implant a small computer chip in his brain and use it to improve the range of motion in his arms and to artificially recreate his sense of touch. It was a long shot, but Burkhart says the potential upside was worth it. “It was a lot to consider, but paralysis wasn’t something I was ready to settle with,” he says. Now, six years after starting the study, Burkhart is able to feel objects and has enough control of his arm to shred on Guitar Hero.
Burkhart’s brain-computer interface, or BCI, was surgically implanted at Ohio State University’s Wexner Medical Center in 2014. Not much larger than a grain of rice, the chip monitors electrical signals from Burkhart’s primary motor cortex, the region of the brain responsible for voluntary movement.
A severe spinal injury impedes the signals from the brain that tell the limbs to move and sensory feedback from the limbs. In Burkhart’s case, the severity of his injury meant that there should have been a complete disconnect between his brain and his arms and legs. But recent neuroscience experiments suggest that in many “complete” spinal cord injuries—perhaps as many as half of them—a few wisps of spinal fiber survive. “Even that small contingent of fibers can lead to a reasonable signal in the brain,” says Patrick Ganzer, a neuroscientist at Battelle. Still, though the electrical signals corresponding to touch and motion are traveling to and from the brain, they’re too weak for a paralyzed person to consciously notice. They don’t feel anything, and their arm doesn’t move.
For Ganzer and his colleagues at Battelle, this raised an interesting possibility. If you extracted those weak signals from the brain, decoded their meaning, and relayed them to the limbs, you could bypass the spine and reconnect the brain and body. Researchers from other groups have demonstrated that it is possible to restore motion using a robotic hand and even send touch signals back to the user by directly stimulating their brain. But doing both at once, and with a person’s own arm, remained elusive.
The problem, says Ganzer, is that the signals for touch and movement are jumbled together in the brain. Each movement or touch generates a unique signal, and the chip in Burkhart’s head takes in around 100 different signals at a time. “We’re separating thoughts that are occurring almost simultaneously and are related to movements and sub-perceptual touch, which is a big challenge,” adds Ganzer.
To make it happen, Ganzer and his colleagues used an elaborate setup that connects Burkhart’s brain to a computer. The chip in his motor cortex sends electrical signals through a port in the back of his skull, which is delivered through a cable to a nearby PC. There, a software program decodes the brain signals and separates them into signals corresponding to intended motions and signals corresponding to a sense of touch. The signals representing intended motions are routed to a sleeve of electrodes wrapped around Burkhart’s forearm. The touch signals are routed to a vibration band around his upper arm.
First, Ganzer and his colleagues focused on restoring motion in Burkhart’s arm without the sensation of touch. Burkhart says the progress was slow at first and required him to learn how to think about moving his arm to generate electrical signals that could be picked up by the computer. “Just being able to open and close my hand was challenging, because before my injury I never had to think about what I’m actually doing to make my hand move,” he recalls.
But within a year he had partially restored movement in his hand. It wasn’t long before he had enough control over his arm to play a modified version of Guitar Hero, one that required pushing the finger buttons on the neck of the guitar, but not strumming with the other hand. “Playing a video game that requires that type of multitasking— listening to the song, watching the screen for timing cues, and executing thoughts related to single finger movements—adds another level of complexity,” says Ganzer.
Burkhart says that having the ability to move objects was “fantastic,” but he was limited without a sense of touch. Without this feedback, grabbing objects required his full attention. Unless he was looking at it, he couldn’t say whether he was holding something or not. “That’s really challenging, especially if I want to grab something that’s behind me or in a bag,” Burkhart says. Even when he could see the object, the firmness of his grip was out of his control, which made handling delicate objects difficult.
Adding a sense of touch into the system proved more difficult. Neuroscientists have successfully reproduced the sensation of touch in quadrepeligic people by relaying data from sensors in a robotic prosthetic hand to a chip in the user’s brain. The problem was Burkhart’s BCI wasn’t designed for that kind of input. It wasn’t even located in the right place. Touch is registered in the somatosensory cortex, which is located behind the motor cortex, where the chip was installed. Yet Ganzer says the somatosensory cortex can be a “noisy neighbor” and some of its signals were picked up by the chip. It was just a matter of finding out what they meant.
To tease out the unique signals corresponding to touch, Ganzer and his colleagues began doing targeted stimulations on Burkhart’s thumb and forearm, parts of his limb where he still had a very weak sense of touch. By observing how Burkhart’s brain signals changed when pressure was applied to his fingers and hand, they were able to identify the weak touch signals against a background of much stronger movement signals. This meant a computer program could split the signals coming from Burkhart’s BCI so that motion signals went to the electrodes around his forearm and touch signals to an armband on his upper bicep.
Everything you wanted to know about soft, hard, and nonmurderous automatons.
Burkhart’s upper arm was also one of the few parts of his body that still had sensation after the accident. This meant that the weak pressure signals relayed from his hand to his brain could be converted into vibrations that would let him know he was touching an object. During tests with the armband, Burkhart could tell when he was touching an object with nearly perfect accuracy, even if he couldn’t see it.
At first, the Battelle touch band was a simple, on-off vibration device. But Ganzer and his colleagues further refined it so that it changes its vibration based on how hard or soft Burkhart grips an object. It’s similar to how videogame controllers and cell phones provide feedback to users, but Burkhart says it took some getting used to: “It’s definitely strange. It’s still not normal, but it’s definitely much better than not having any sensory information going back to my body.”
Robert Gaunt, a biomedical engineer at the University of Pittsburgh’s Rehab Neural Engineering Labs, contrasted Battelle’s system to the approach being developed in his own lab, where a BCI controls a robotic limb and sensors on that limb return signals that stimulate the brain to artificially recreate a sense of touch in a person’s hand. “What they’re doing is a little more like sensory substitution, rather than restoring touch to his own hand,” Gaunt says. “We all have the goal of developing devices that improve the lives of people with spinal cord injuries, but the most effective way to do that is totally unclear at this point.”
Now that Ganzer and his colleagues have demonstrated the technology in the lab, he says the next step is to improve the system for everyday use. The team already has shrunk the electronics used in the system to a box the size of VHS tape that can be mounted on Burkhart’s wheelchair. The bulky system of electrodes has also been reduced to a sleeve that is relatively easy to take on and put off. Recently, Burkhart used the system for the first time at home, controlling it through a tablet.
Given the invasive nature of BCIs, which have to be surgically implanted, it may be a while before these sorts of systems see widespread use among quadriplegics. Noninvasive BCIs that don’t require surgery are an area of active research, but it’s still very early days for the technology. Ganzer is working on a project that is funded by Darpa to develop a BCI that uses a special type of nanoparticle to wirelessly send signals to and from the brain. But none of this technology would be possible without people like Burkhart who volunteer to show what’s possible.
“My goal is to get this into the hands of other people with paralysis and see how far we can push the technology,” says Burkhart. “The biggest thing that’s motivated me is this hope for the future.”