Elon Musk enters the world of brain-computer interfaces

EVER since ENIAC, the first computer that could be operated by a single person, began flashing its ring counters in 1946, human beings and calculating machines have been on a steady march towards tighter integration. Computers entered homes in the 1980s, then migrated onto laps, into pockets and around wrists. In the laboratory, computation has found its way onto molars and into eyeballs. The logical conclusion of all this is that computers will, one day, enter the brain.

This, at least, is the bet behind a company called Neuralink, just started by Elon Musk, a serial technological entrepreneur. Information about Neuralink is sparse, but trademark filings state that it will make invasive devices for treating or diagnosing neurological ailments. Mr Musk clearly has bigger plans, though. He has often tweeted cryptic messages referring to “neural lace”, a science-fictional concept invented by Iain M. Banks, a novelist, that is, in essence, a machine interface woven into the brain.

Although devices that can read and write data to and from the brain as easily as they would to and from a computer remain firmly in the realm of imagination, that has not stopped neuroscientists (and, of course, Mr Musk) from indulging in some speculation. Theodore Berger of the University of Southern California, in Los Angeles, has proposed that brain implants might be used to store and retrieve memories. Dr Berger’s prosthesis would be intended to help those whose brains cannot form long-term memories because they are damaged. But if the idea worked, there seems little reason why those without damage should not and would not want something similar. Mr Musk himself, more ambitiously still, imagines an implant that would let the wearer tap directly into the internet, and all of the computational power available there.

Of minds and melding

Behind this suggestion lies Mr Musk’s argument, made repeatedly, that human beings need to embrace brain implants to stay relevant in a world which, he believes, will soon be dominated by artificial intelligence. Proposing the artificial augmentation of human intelligence as a response to a boom in artificial intelligence may seem a bit much. But Mr Musk’s new company is not alone. A firm called Kernel is following a similar path.

To start with, Kernel’s engineers hope to build devices for the treatment of neurological conditions such as strokes and Alzheimer’s disease. Ultimately, however, they want to create cognition-enhancing implants that anyone might care to buy. Kernel was founded in October 2016 by Bryan Johnson, an entrepreneur who, like Mr Musk, got rich by processing payments online (PayPal, which Mr Musk helped found, bought Braintree, Mr Johnson’s company, in 2013). Mr Johnson put $100m of his own money into Kernel, stating that “unlocking our brain is the most significant and consequential opportunity in history.”

In some ways, Mr Johnson and Mr Musk are merely the new boys in what is quite an old field. The first brain implants, carried out in the 1970s, were prosthetic visual systems, though they did not work well. Cochlear implants, to restore hearing, have done much better. Hundreds of thousands of people now have them—though, strictly speaking, they talk to auditory nerves rather than to the brain directly, which simplifies the task. For some people, the symptoms of Parkinson’s disease can be kept in check by electrodes the diameter of a strand of spaghetti inserted deep into the brain. And one of the latest ideas in the field is to read and interpret brain activity, in order to restore movement to the limbs of the paralysed (see article).

In one important way, however, Kernel and Neuralink are different from previous efforts. Though aimed initially at medical applications, they also explicitly nod to the possible non-medical uses of this kind of implant technology. In February Mr Musk said that he thought “meaningful” interfaces between the brain and computation were five years away. The creation of Neuralink suggests that he, like Mr Johnson, is putting his money where his mouth is.

Most neuroscientists would, it must be acknowledged, regard all this as heroically optimistic. In a review of the field, published in January in Nature Reviews Materials, Polina Anikeeva and her colleagues at the Massachusetts Institute of Technology (MIT) wrote that, although Moore’s Law and the miniaturisation of electronics have brought devices down to a size where their insertion into the brain can be considered, big challenges lie ahead.

The brain’s complexity, and researchers’ present lack of understanding of how that organ’s component cells work together to do what they do, makes designing interfaces between brain and machine hard. But, even were it simple in principle, the rigid, silicon-based tools of modern computing do not mesh easily with the squishy soft-tissue of biology. Implants often generate scars around themselves. And the surgery needed to put them in place carries risks of its own.

There may, though, be alternative approaches. One such is being tested by a group at Florida International University, in Miami, led by Sakhrat Khizroev. Dr Khizroev and his team use magnetoelectric particles so tiny that they can interact with the electric field generated by an individual nerve cell. The team inject these particles, tens of billions at a time, into a vein in a rat’s tail, then drag them into the animal’s brain using magnets. Each particle produces an electric field when stimulated by an external magnetic field. This may, in principle, permit a researcher to use such a particle to influence the electrical states of nearby nerve cells—and thus, in essence, reprogram them. How that would be done in practice, though, is obscure.

Another approach, being pioneered by Jose Carmena of the University of California, Berkeley, and his colleagues, uses devices the size of a grain of rice to convert ultrasonic energy beamed towards them into electricity that can stimulate nerve or muscle cells. Ultrasound travels through the body, so can power and control such devices without wires.

Both Dr Khizroev’s technique and Dr Carmena’s are less invasive than the current standard brain interface, a patch of needlelike electrodes known as a Utah array that is plugged into the brain’s surface. This is far too blunt an instrument to send any but the crudest signals into a brain. But, regardless of the precise approach taken to hardware, another problem the field faces is that no one understands the mechanism behind the natural equivalent of software—the way the brain encodes information. Such interfaces as do exist have to be trained, rather than instructed what to do. Instruction would be possible only if brain signals were properly understood.

It is not yet clear which technological routes Mr Musk’s and Mr Johnson’s commercial efforts will take, though Kernel recently bought Kendall Research Systems, a spin-off from MIT that builds devices which use light, rather than electricity, to stimulate the brain. But the two firms’ shared underlying premise—that medical purposes might lead to more consumer-orientated applications—does seem a sensible way to do things.

People understand that medical procedures can be risky. As long as it is done in good faith, they will tolerate experimentation on people that would be intolerable in non-medical circumstances. That will let Neuralink, Kernel and those that come along afterwards build up expertise that might be turned to more general effect in the future.

As for Mr Musk himself, Neuralink brings to five the number of ambitious technology companies in which he is involved. The others are Tesla (electric cars, batteries and solar power), which this week attracted an investment from Tencent, a Chinese tech giant; SpaceX (rocketry); the Boring Company (tunnelling); and Hyperloop (vacuum trains). It is hard to discern the connections between these ideas. But, in Mr Musk’s mind, they are presumably already laced together.