Owen Collumb was paralyzed in 1993, when he was 21 years old. A tire on his motorbike blew out and he fell into a ravine, breaking a single bone in his spine. When he recovered, he couldn’t move his legs and could control only the biceps in his arms, meaning that he could lift his hands but, to put them down, he had to twist his shoulders and let gravity unbend his elbows.
He spent years in an assisted living home before petitioning to move to his own place in Dublin, with the help of home aides. Living alone was liberating; he could choose what he ate and when he woke in the morning. He began working multiple jobs for foundations and advocating for people with disabilities. One of his assistants, Sylwia Filipiek, a Polish immigrant to Ireland, had been employed at a printing factory. She had no experience with home care and struggled to help Mr. Collumb into his wheelchair at first. But, over the years, they learned how to work together, and grew close.
In the summer of 2024, Mr. Collumb and Ms. Filipiek flew to Bath, England, to train for the Cybathlon, an international competition run every four years to encourage the development of assistive technologies. The competition, hosted in Switzerland by the university ETH Zurich, consists of eight races for teams and their pilots (which is what the primary competitors, with varying disabilities, are called), each targeting different innovations, such as arm prostheses, leg prostheses and vision assistance.
Each race consists of remote tasks that are supposed to simulate everyday life for the pilots: walking across a room, picking up a grocery bag, throwing a ball. One of Cybathlon’s founders, Roland Sigrist, compared it to Formula 1. Teams are encouraged to develop prototypes toward the ultimate goal of “the independence of people with disabilities,” but the competition is straightforward and real, with all its accompaniments: nerves, heartbreak, glory. The pilots are the ones that put themselves on the line. “They’re the masters of the technology, and not the other way around,” Mr. Sigrist said.
Mr. Collumb, who is now 54, has participated in the Cybathlon since its first iteration, in 2016, as a pilot in the competition’s most abstract category: brain-computer interfaces. Imagine staring at a cursor at the center of your computer and willing it to move to the right. A brain-computer interface, which allows humans to control computers with just their minds, can turn that willing into action. As someone who lacks almost all ability to move his body, a brain-computer interface could allow Mr. Collumb to play video games, use the internet and direct his wheelchair himself.
With artificial intelligence increasing the accessibility and sophistication of technological progress, the integration of organic and robot life is now a matter of degree. How tightly should we embrace these new tools? Will they make life better in the end? Can they change our idea of what people are capable of? The Cybathlon and its participants distill these questions into something concrete. “This isn’t showing your disabilities, it’s showing what you can do,” Mr. Collumb said. “You may be in a wheelchair, you may not be able to move, but you can compete.”
To implant or not implant
Mr. Collumb’s team is led by Damien Coyle, a professor of neurotechnology at the University of Bath and the director of the university’s Institute for the Augmented Human. Dr. Coyle is best known for building brain-computer interfaces with electroencephalography, or EEG, which uses electrodes placed on the scalp to measure brain activity. The idea is to see which areas of the brain spark to life when, to use the example again, someone wills a cursor to the right, and to link this activity to computer commands that push the cursor to the right. It’s not far removed from how our bodies work naturally. When you try to pick up your foot, a spot in your motor cortex activates, sending signals down your spinal cord to your foot, which then moves. The brain-computer interface replaces the spinal cord and the foot; the computer becomes the body.
This is, in some ways, the holy grail of assistive technology. Imagine integrating a prosthetic so thoroughly with your thoughts that you could control it without being aware of its otherness. It is simply part of you. Robotics and material science present limits on such possibilities — it’s one thing to effect motion in a video game, another to do so with a physical prosthesis, or avatar. But a purely virtual brain-computer interface could open many doors for people with and without disabilities. Of course, the value of further embedding ourselves in our increasingly technologized world is questionable. Are we being pressed into screens by companies that trade on our attention and personal information? Evolving from iPad children into children with hard drives in their frontal cortices and monitors in their corneas?
But it doesn’t take a transhumanist — someone who believes in technology’s power to help us “transcend ‘natural’ but harmful, confining qualities derived from our biological heritage,” as one of the movement’s leaders, Max More, has put it — to see the potential accessibility benefits. Our agency can be expanded through computers, and reliably translating brain activity so it can be read by a machine is a key to making these opportunities open to all, as the market is aware. Companies like Meta, Microsoft and Apple have all initiated brain-computer interface projects in recent years in their efforts to develop virtual reality systems.
Dr. Coyle has his own brain-computer interface company, through which he sells some of the technology he develops at the University of Bath. But for the moment, he said, “there’s a disconnect between what we can achieve and what people can get from these technologies.” Research lags behind practical application. Electrodes on the scalp are easy to use and cheap, but they are often clumsy, having to track impulses through fluid, bone, skin and hair. Activity measured from trying to move your right hand can look similar, electrically, to activity measured from trying to move your right shoulder; neurons associated with trying to move your fingers overlap closely with neurons associated with trying to move your tongue.
The alternative is to use an implant, which can pick up more granular brain activity. Neuralink, a brain implant company founded by Elon Musk, has a valuation of around $9 billion; Precision Neuroscience, a competitor, has a valuation of around $500 million. But such procedures, which involve opening the skull and placing a monitor on the surface of the brain, are invasive and expensive.
Even then, there is the problem of classification: Each person has a different neural fingerprint, so to speak, and there is no rule linking any one cluster of neurons to any one behavior. The brain is noisy, has natural variations and can be insurmountably difficult to decipher just by looking at it. A.I. has helped. Specialized programs are able to pick out patterns from data sets too large for humans to process and, from brain activity, can decode intentions better than before. However, it’s still not foolproof. And human stress or emotion can cloud a previously successful program.
The best methods are the most personal: Researchers will work with the same individual participants over time, allowing them to acclimatize to the interface at the same time the program is tailored to their neural patterns. Mr. Collumb was recruited in 2010 by one of Mr. Coyle’s students for a brain-computer interface study. In preparation for the first Cybathlon, in 2016, Dr. Coyle went through his list of study participants, looking for the best performers. Mr. Collumb wasn’t even in the top five. But many participants were hesitant to sign on, and Dr. Coyle moved down the list, eventually making it to Mr. Collumb, who eagerly accepted. “It’s human nature in people, even when they’re disabled, to compete,” he said.
The two have now worked together for more than a decade, and have found that the most reliable method under stress uses only two main commands: a “left” and a “right.” To signal right, Mr. Collumb concentrates on moving his right arm as if “a huge weight” is on top of it, he said; he concentrates similarly on his left arm to signal left. These impulses are then translated, along with the timing of Mr. Collumb’s efforts, into orders for the computer.
Relying on increasingly accurate EEG readers and A.I. classifiers, Mr. Collumb and Dr. Coyle have improved since the first competition, where they placed fourth. But so have others. During their training session in the summer of 2024, the two sat together in front of a large monitor; a skintight hat with 32 honeycomb-like electrodes was wrapped around Mr. Collumb’s head.
Attila Korik, a Hungarian scientist, sat nearby, staring at his own screen. He began calibrating the brain-computer interface, asking Mr. Collumb to try moving different parts of his body while watching measurements of the voltage pulsing through the electrodes. Mr. Collumb’s head jerked to the side, his hands lifting minutely off his wheelchair. Sweat dripped down his cheek. Ms. Filipiek brought him a glass of water, and he tilted his head back as she poured it into his mouth.
There were six tasks for Mr. Collumb to complete, including pushing a virtual key into a virtual door handle and filling a virtual cup with virtual water. Three other teams — two from Italy and one from Pittsburgh — were also online, competing asynchronously. Time ticked by. Mr. Collumb missed a few queues, his eyes squinting slightly with effort. Mr. Korik put his head in his hands. The team finished in just under eight minutes. Mr. Collumb’s face was pale and waxy. He came in second place.
The winner, an American team headed by a pilot named Phillip McKenzie, who has an implanted electrode, completed the same challenge in nearly half the time. The result foretold the outcome of the actual competition, in October of that year: Mr. Collumb finished fourth overall. Mr. McKenzie came in first, completing every task perfectly, in record time.
‘We weren’t in the same league’
In November 2024, a month after his loss at the Cybathlon, Mr. Collumb confided that he was considering traveling to Thailand to get a brain implant. He said that he was “devastated” at the result.
“We weren’t in the same league as the Pittsburgh people,” he said. “They’re playing chess and we’re playing checkers.” He conceded that undergoing such a procedure in his 50s was risky, but that it would be worth it if he could contribute to science, win the Cybathlon and gain “the ultimate level of control.” He was “not going to go quietly,” he said. “Fighting is better than slowly drifting away.”
In most cases, implants do provide higher degrees of control than surface EEGs. Brain activity is much easier to interpret when it’s clear which specific regions are active. Ceci Verbaarschot, a neuroscientist at the University of Pittsburgh who led the research for the winning Cybathlon team, said that the difference between invasive and noninvasive brain-computer interface methods was like the difference between using a mouse to move a cursor on a screen and moving the cursor directly, without a physical intermediary Because of the intricacy of the electrical data from implants, classifiers can be highly specified, capturing intended movement to the level of a single finger flutter. Many people with implants report being able to navigate virtual movement similarly to how they used to navigate physical movement.
Mr. McKenzie had his implant put in during the spring of 2023. He was paralyzed in 2012, and had gone through extended bouts of poor health and depression, in which he wouldn’t eat or talk. “I felt unaccomplished,” he said. But, a month after surgery, when he picked up the brain-computer interface, he quickly made strides. “They plugged me into the robot and it was like riding a bike, brother,” he said. “It was like putting on a glove.” Within two weeks, he could control a robotic arm.
It is difficult to know the extent to which a new technology will affect your life. How easy is it to buy some new gadget that seems tailor made for you — a money clip, a robot vacuum, a vegetable slicer — just to use it once and then never again? We are all creatures of habit, and any change, even an obviously good one, is costly. It can be difficult to measure the benefits against such costs. One might outweigh the other on Tuesday and then vice versa on Friday, after you’ve had a good meal, say, or a hot shower.
Late this summer, nearly a year after the Cybathlon, Mr. Collumb said that he still hadn’t gotten the implant, and that he was leaning against it. He was content with what he could do now. “The competing is one thing,” he said. “And then life, and living, and possibilities of giving me more choice and control of my life is another thing.” He added, “An implant onto my brain cell, I don’t see it necessarily benefiting my life.”
The virtues of the implant were less ambiguous for Mr. McKenzie. “The feeling of competition, the feeling of achieving something, is what I used to live for,” he said. “So to do it again, to have that feeling again, there’s nothing like it.” He said that he had already begun training for the next Cybathlon, which will be in Singapore in 2028.
“The technology in my head, they don’t expect it to get better, but me personally, I always expect to get better,” he said. “In four years, I don’t want to hear that my technology is getting old. I want to be alpha again.”
Oliver Whang is a writer based in Brooklyn. He started writing for The Times in 2020.
The post Let the Mind-Control Games Begin! appeared first on New York Times.
