Dennis Degray and James Johnson are two of about three dozen people who have had brain-computer interfaces (BCIs) installed, allowing electrical signals from implants in their brains to control devices in the outside world. Degray became paralyzed from the neck down after a fall in 2007, and Johnson became paralyzed below the shoulders in 2017 in a go-cart accident. Each now has tiny arrays of electrodes placed in his brain by researchers from BrainGate, a consortium of academic scientists, engineers and physicians who study BCIs. After considerable training, both subjects can use their thoughts to accomplish tasks that would otherwise be impossible. Degray can type eighteen words per minute by thought alone. Johnson, who came out of deep depression after his accident, has now operated a robotic arm and driven a simulated car amid virtual surroundings. He says, “I am always stunned at what we are able to do.”
Neuroprosthetics for paralyzed and otherwise disabled people, and for certain brain conditions, are only part of what some see as the future of brain implants. As smart phones and algorithms shape our lives, it’s a small leap to envision connecting perfectly healthy people directly and intimately to computers by brainwaves rather than by keyboard, mouse or voice. One recent article presents brain implants as possibly the next computer mouse and a wireless one at that. In this view, brain implants could become consumer items, the “next big thing” in technology that investors and entrepreneurs support in hopes of eventually producing a profit-making product.
One investor in biomedical companies, Christian Angermayer, is confident that “in less than twenty years’ time, we will all have a BCI” as a “fundamental input-output device.” That and deeper possibilities are seen by tech entrepreneur Elon Musk, known for Tesla Motors and SpaceX and also the guiding spirit of Neuralink, his BCI company. In Musk’s first public presentation about Neuralink in 2019, he noted potential medical benefits, but also spoke of developing a “full brain-machine interface [that would] achieve a symbiosis with artificial intelligence” – a goal that seems only an aspirational dream. Synchron, a competing company, brings instead a novel approach and a focus on the medical uses of BCIs.
There is a surprisingly long history of using the brain’s electricity to help the ill and injured. In the first century CE, the Roman physician Scribonius Largus recommended placing a live fish called a torpedo, known to deliver electric shocks, on the site of a serious headache to remove the pain “forever.” In the early nineteenth century, one Giovanni Aldini claimed to cure what we would today call clinical depression by applying an electric current to the patient’s head.
It took longer to realize that these electrical interactions arise because the 86 billion neurons in the brain intercommunicate by electrochemical impulses. The first signs came when late-nineteenth century experimenters placed electrodes on the exposed brains of animals and recorded rhythmic oscillations or “brain waves.” In 1924, Hans Berger, a German physiologist, invented electroencephalography (EEG) when he recorded the first brain waves from a human by putting electrodes on a subject’s skull. EEG has become a valuable clinical tool to diagnose conditions such as epilepsy and brain tumors.
By the 1970s, EEG signals were being used to control external devices. Jacques Vidal, a UCLA computer scientist, wrote about brain-computer interfaces and coined the phrase, then described his experiment where subjects fitted with EEG contacts mentally moved a cursor through a maze on a computer screen. EEG contacts are still used; for instance, by Miguel Nicolelis, a researcher at Duke University known for his pioneering work with BCIs. In 2014 he fitted an EEG cap to the head of a paraplegic person encased in a robotic exoskeleton. After training and practice, the young man showed 50,000 spectators at the 2014 FIFA World Cup opening in São Paulo, Brazil, that by thought alone he could activate the robotic body to kick a soccer ball.
EEG contacts are non-intrusive, avoiding harm to the brain due to surgically implanting electrodes, but have limits. They sample large groups of neurons at poor signal-to-noise ratios so it is difficult to relate EEG signals to specific brain activities. They do not detect neural activity below the cortex, the brain’s outer layer, but some uses require a deeper connection. As Nicolelis explains, EEG data adequately control the lower limbs: but “inside the brain, you get much better results” for the upper limbs, because information is available there to give fine control of the arms. For these reasons, Nicolelis and most BCI researchers generally use implanted electrodes.
The first such implantations were carried out in the 1950s, when researchers inserted fine metal wires into animal brains and recorded the output from individual neurons. Later, implants benefitted from techniques used in the silicon chip industry, which were adopted to produce arrays of exactly spaced silicon probes. The most widely used such unit is the Utah Array, from Blackrock Neurotech in Utah. It consists of ninety-six silicon microelectrodes up to 1.5 mm (0.06 inch) long arranged in a regular array roughly 4 mm (0.16 inch) square, each with a sharp tip covered with platinum or other metal. This arrangement can distinguish among the outputs of individual neurons, giving fine control over external devices.
Utah Arrays and similar microelectrode implants lie behind the results achieved by Dennis Degray and James Johnson as well as other recent advances in BCIs. In one, a quadriplegic person operated a robotic arm with the new feature that it sent back tactile information to the subject’s brain as he grasped an object, vastly improving his performance. In another, researchers enabled a person who had lost the ability to speak because of a stroke to communicate through a neuroprosthetic speech device. These successes exhibit remarkable progress in the medical uses of BCIs but also show that implant technology is still in its early stages.
One challenge is to find ways to implant BCIs in the brain with a minimum risk of infection or inflammation, yet that can tap the appropriate type and number of neurons. For long-term use, it is essential as well that the implants do not deteriorate over time in the brain’s unwelcoming environment. Also, in a BCI lab, scientists and physicians deal with the implant and its wired connections, but it would be ideal to have a self-contained neural interface that does not need professional support for day-to-day use. Neuralink and Synchron both seek to improve and commercialize BCI technology along these lines.
Though Elon Musk is well known, Neuralink has been called “secretive.” The Wall Street Journal first reported its existence in March 2017 and named its co-founder and president, Max Hodak. He earned a biomedical engineering bachelor’s degree in 2012 at Duke University, where he was a research assistant in Nicolelis’s lab. Little else was known until a livestreamed presentation in 2019 by Musk, Hodak and Neuralink scientists. An overview was also published that year as a peer-reviewed paper over Musk’s name, but no other papers from Neuralink have appeared since. All we know about the company’s science comes from its own releases (“neuroscience theater,” one commentator calls them) such as another livestream in 2020, and media coverage.
The 2020 presentation showed differences from what Neuralink proposed in 2019. Musk stated that the company goal is to “solve important brain and spine problems with a seamlessly implanted device,” and went on to list a dozen neurological issues he said his device could help. He claimed that the device, a “Fitbit in your skull with tiny wires,” far outdid other implants. It was a cylinder 23 mm (0.9 inch) across x 8 mm (0.3 inch) high, the size of a quarter but thicker, with 1,024 electrodes that could access or stimulate neurons. It would fit flush with the skull after implantation by an automated Neuralink-designed surgical robot that would not cause bleeding in the brain. Using a built-in wirelessly rechargeable battery, the implant would send its neural information to an iPhone app via Bluetooth.
The 2020 presentation also displayed animal testing results, after Musk announced in 2019 that Neuralink had put a chip in a monkey brain. First shown though was Gertrude, a pig. Her implant, installed two months earlier, recorded data from the neurons controlling her leg motion, which could predict actual leg movement. A similar algorithm using recordings from neurons to predict hand motion motivated a Neuralink video in 2021. It attracted 6 million views by showing a macaque monkey with implanted chips playing Pong by mind alone. Musk tweeted that the first Neuralink product would use this method to enable a paralyzed person to operate a smartphone.
These demonstrations did not wholly impress BCI researchers. They gave the chip design and the proposed surgical robot favorable comments, but pointed out that no new neuroscience was shown: monkeys mentally tied to computers and able to move objects on screens go back at least to the early 2000s including ground-breaking work by Nicolelis, and we have already seen paralyzed people like Degray and Johnson mentally control devices. Andrew Jackson, Professor of Neural Interfaces at Newcastle University, UK, tweeted that the demonstrations “didn’t show anything that hasn’t been done before … this is solid engineering but mediocre neuroscience … it is unfortunate that they are presenting their work in this way, rather than publishing peer-reviewed papers that would allow their claims to be scrutinized.”
Still, if proven safe, the chip could conceivably advance the field of medical brain implants. It could also open up the field of consumer brain implants, as either the ultimate computer mouse or as Musk’s idea of a total human-machine mind meld. This raises a big question. Currently brain surgery to implant a chip, with its associated risks, is done only for severely limited people, in the hope that it will improve their lives. Would it ever be acceptable and ethical to perform this surgery on healthy people without medical need? The question is moot however until Neuralink gains FDA approval after human clinical trials. Musk had said that these would begin in 2020, then in 2021, and now says they will begin this year.
Following a different path, Synchron has already reached the critical benchmark of human testing after publishing all the details of its research in scientific journals. Thomas Oxley, the company’s CEO and co-founder, is a physician specializing in neurology at New York City’s Mt. Sinai Hospital, and has a PhD in neural engineering from the University of Melbourne. Oxley invented the stentrode, Synchron’s unique device that retrieves neural signals from the brain without the need for brain surgery or entry through the skull.
A stentrode (stent-electrode recording array) is a set of electrodes mounted on a stent and permanently implanted into a blood vessel in the brain. Stents are tiny hollow cylinders typically made of fine metal mesh that are inserted into an anatomical duct such as an artery to prevent it from closing. Procedures for inserting stents are “minimally invasive,” meaning they are not considered major surgery. Stents have successfully treated coronary heart disease and other conditions in large numbers of patients for years, although certain complications are possible.
Oxley’s insight was to realize that blood vessels are natural routes into the brain, and that a stent that carries electrodes and is placed in such a pathway could detect or stimulate the brain’s neural signals without brain surgery. In 2016, Oxley and a team of physicians and scientists co-authored a peer-reviewed paper as proof of concept. They inserted a commercial stent with added electrodes into a vein that overlays the motor cortex in the brain of sheep. Months of testing in numbers of sheep showed that the electrodes accurately recorded neural signals from the motor cortex, which control bodily movement.
Oxley then received medical approval in Australia to carry out initial human trials on four participants with ALS (amyotrophic lateral sclerosis) that paralyzed their arms. A Synchron stentrode 40 mm (1.6 in) long x 8 mm (0.3 in) diameter, with 16 electrodes, was placed in a vein near the motor cortex for each person, reading neural signals and wirelessly sending them to external devices. A peer-reviewed paper by Oxley’s team (2021) and a TED talk (2022) show the results. After training, participants could operate a computer through a combination of neural signals and eye-tracking to carry out daily activities such as online shopping, banking, and, most important emotionally, text messaging in their home environments without professional support. No participants experienced serious adverse events. In 2021, Synchron received FDA approval for US trials to assess permanently implanted stentrodes. The first participant has just been selected.
These trials move Synchron far closer than Neuralink to the goal of providing safe and effective human BCIs, but much remains to be learned about each company’s method. Eliminating brain surgery is a clear advantage for Synchron, especially if stentrodes can be shown to work for a variety of conditions. FDA trials and future development will reveal what Neuralink and Synchron can and cannot safely do for people.
Apart from the technologies, the company histories are important to know for a project as significant as developing human BCIs. Last January, Fortune magazine described internal conditions at Neuralink. Interviews with a half-dozen former employees revealed innate tensions due to competing goals among Neuralink’s original founders, such as performing basic research versus treating disease. Musk too paid attention to medical uses, but saw Neuralink’s true goal as empowering people by melding their minds with computers, and wanted to achieve that very quickly. Tim Hanson, an expert in animal BCIs, is one of the majority of founders who have left Neuralink. He commented in the Fortune article that the pressure to get results did not match the reality that “basic science is basically slow.” Other interviewees confirmed that the intense atmosphere contributed to problems in developing the chip such as high turnover rates.
The greatest defection occurred when Neuralink’s president, Max Hodak, announced in May 2021 “I am no longer at Neuralink,” having been forced out, according to the New York Times. Then in early 2022, a blog post from Hodak revealed that he had invested in Synchron and taken a position on its board of advisers. He described Synchron’s “quite elegant” method of accessing the brain and called the company “among the most serious” of the groups working on human BCIs. Hodak has said that his departure should not be taken as a knock on Neuralink and it is continuing as a well-funded operation, although currently without a president.
Whatever the outcome for Neuralink, after ten years Synchron’s efforts grounded in medicine and innovative BCI science have already produced real help for real people in need, without brain surgery. For a new technology that can so directly affect people’s lives and health, it is reassuring that instead of the Silicon Valley mantra “move fast and break things,” Synchron has chosen to “move deliberately and get it right” through scientific and medical openness and scrutiny. That’s exactly what I would want if I were ever to need, or choose to have, an artificial object inserted into my brain.