In a small lab at Brown University in Providence, RI, a rhesus macaque sits in a chair facing a computer screen, gripping the handle of a device that looks a lot like a sailboat’s tiller. For the moment, the monkey uses this device as if it were a computer joystick to control a simple video game: a colored dot appears on the screen, and the animal moves the cursor to meet it. Once the animal gets good at the task, though, the researchers in the adjoining room will flip a switch and it will be signals straight from the monkey’s brain, not the joystick’s movements, that drive the cursor.
This eerie feat is possible because the researchers, led by Brown neuroscientist John Donoghue, have implanted a tiny array of electrodes in the monkey’s brain. The electrodes intercept signals from individual neurons in the brain, and a specially developed computer algorithm translates these signals into trajectories and velocities for the computer cursor. The researchers’ ambitions, however, extend way beyond video-game-playing monkeys. Their hope is that their brain-machine interface system will give patients paralyzed by spinal-cord injuries or neurodegenerative diseases new abilities to interact with the world around them-using nothing more than the power of their thoughts.
Donoghue and his team launched Cyberkinetics in June 2001 to pursue that vision. The company is one of the first to arise from research into brain-machine interfaces, which has so far been relegated mainly to a handful of academic labs around the world (see “Brain-Machine Interfaces,” TR January/February 2001). And while much development remains to be done, a system like Cyberkinetics’ that taps directly into the brain could theoretically give paralyzed patients the means to control computers, robotic aids-and perhaps even their own muscles. Cyberkinetics aims to begin testing that theory in humans by the end of this year.
In the monkey studies that will pave the way for the human tests, the company-which includes cofounders Nicholas Hatsopoulos of the University of Chicago, Brown MD/PhD student Mijail Serruya and Gerhard Friehs, a neurosurgeon at Providence’s Rhode Island Hospital-is focusing on the area of the brain that issues commands to the monkey’s arm. Friehs starts by implanting a four-millimeter-square array of 100 electrodes in this region, which is located in the brain’s outermost layer, about halfway between the ear and the top of the skull. After the surgery, a small bundle of wires snakes from the array through a hole in the animal’s skull; those wires are plugged into a computer, feeding the electrical signals generated by neurons firing near each electrode into the machine.
Hatsopoulos sits at that computer as the plugged-in monkey practices a video game in the next room. The brain activity picked up by the array flashes across the screen as a jumble of hyperkinetic EKG-like graphs; rendered audible by the computer’s speakers, brain signals snap, crackle and pop like Rice Krispies in milk. Hatsopoulos turns up the volume. “I never get tired of listening to that,” he says. “This is really like reading the mind, eavesdropping on cells in the brain as the monkey’s thinking of something.” Pattern recognition software fishes the signal “spikes”-each representing a single firing of a single neuron-from the brain’s background noise and correlates them with the position of the monkey’s arm. “The amazing thing,” says Donoghue, “is that very quickly you can get a sense of the neurons’ activity and extract the hand’s trajectory.” Indeed, using only three minutes or so of data from the video game exercise, the computer can build a model capable of extrapolating the monkey’s arm movements from the brain signal alone. Once the model is fine-tuned, the computer can use the brain signal to drive either a cursor or a robotic arm in real time.
Such promising results are part of what inspired the researchers to launch Cyberkinetics and push toward clinical trials. “We already know so much; now let’s put it to use,” says Serruya. The participants in Cyberkinetics’ first human tests will be “locked-in” patients who, due to injury, stroke or neurological disease, are completely paralyzed, unable even to communicate except via subtle movements of their eyes. In those initial trials, the company will implant the electrode array, manufactured by Salt Lake City, UT-based Bionic Technologies, but the signal-processing hardware and power supply will remain outside of the body. If those first human tests bear out the promise of the monkey experiments, the company plans to further develop the technology to create an entirely implantable device.
To date, only one company has conducted human tests of a brain-recording implant with the aim of helping restore function in paralyzed patients: Atlanta, GA-based Neural Signals. Instead of an electrode array, the company implants two “neurotrophic electrodes”-glass tubes containing tiny wires and a substance that encourages brain cells to grow into the devices. Neurologist and Neural Signals founder Philip Kennedy says the studies, begun in 1997, are going more slowly than he had originally hoped, but that the company should have some clear results by the end of the year. Cyberkinetics researchers believe, however, that implanting 100 electrodes instead of just two will make their system more robust and will allow it to gather more information from the brain.
While recent work in brain-machine interfaces is encouraging, some significant hurdles remain, says William Heetderks, head of the National Institutes of Health’s Neural Prosthesis Program, which helps fund brain-machine interface research. Perhaps the biggest challenge, Heetderks says, is building an interface between the recording device (a rigid piece of hardware) and the brain (a squishy mass floating in cerebrospinal fluid) that will maintain its precise position for decades, despite small movements of the brain. While both Kennedy’s and Donoghue’s devices represent progress on that front-Kennedy’s by encouraging cells to grow into the device and stabilize the connection, Donoghue’s by gripping the brain much as golf cleats grip wet earth-Heetderks thinks that some combination of approaches might ultimately be necessary. At this point, Heetderks says, human studies “may be still a little bit premature. But obviously that’s just one opinion.”
Greg Licholai, director of ventures and business development for the neurological division of Minneapolis, MN-based Medtronic, offers a different view. “This is truly a breakthrough in approaching neurological disorders,” Licholai says of Donoghue’s efforts. “I don’t think there’s going to be a problem recruiting patients, and the system has been well proven in an animal model. The only potential holdup is how long it takes them to draw up the documents and get FDA approval of those early-stage trials.”
Cyberkinetics’ business manager and only employee, Brown undergraduate Mikhail Shapiro, is helping the company look for the management team and funding it will need to get that paperwork in order and the human tests under way. Shapiro and the company’s founders all realize they will face both business and technological challenges, but they are also convinced that, as Hatsopoulos puts it, “This is real. This is really going to help people.”
A horrifying new AI app swaps women into porn videos with a click
Deepfake researchers have long feared the day this would arrive.
We can’t afford to stop solar geoengineering research
It is the wrong time to take this strategy for combating climate change off the table.
Meet Altos Labs, Silicon Valley’s latest wild bet on living forever
Funders of a deep-pocketed new "rejuvenation" startup are said to include Jeff Bezos and Yuri Milner.
The new version of GPT-3 is much better behaved (and should be less toxic)
OpenAI has trained its flagship language model to follow instructions, making it spit out less unwanted text—but there's still a way to go.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.