Neuralink: unpacking the hype around Elon Musk’s brain-computer interface

Chloe Kent 12 August 2019 (Last Updated January 30th, 2020 11:05)

In July 2019, enigmatic multi-billionaire and tech idol Elon Musk made a highly anticipated announcement about his brain-computer interface company NeuraLink. Despite the ensuing excitement, the device it has developed – made up of dozens of electrode threads designed to be laced into the brain through small holes in the skull – isn’t entirely new technology.

Neuralink: unpacking the hype around Elon Musk’s brain-computer interface
NeuraLink has now applied to US regulators to begin first-in-human trials. Credit: Shutterstock

Elon Musk’s Space X project is exploring missions to Mars, his Hyperloop project aims to reinvent travel and his Boring Company is looking to build tunnels under Los Angeles. Now, he’s turning his attention to the human brain.

NeuraLink’s brain-computer interface (BCI) device uses dozens of thin wires to collect signals in the brain and transmit the information outside of the body. The tiny probe contains more than 3,000 electrodes attached to flexible threads thinner than a human hair which can monitor the activity of over 1,000 neurons.

According to Musk, a monkey was able to control a computer with just its thoughts during animal tests of the device. The firm has now applied to US regulators to begin first-in-human trials, with plans to work with paralysed people to see if it enables them to do the same.

Using BCIs to restore mobility and independence to paralysed patients has been going on for years. In a 2012 study, an aspirin-sized device implanted in the motor cortex of the brain allowed two tetraplegic patients to control a robotic arm with their minds, and in 2017, BCI technology allowed patients with locked-in syndrome to communicate, reading their brainwaves to allow them to answer a series of yes/no questions.

What is new about NeuraLink is its eventual intended application – while paralysed individuals are likely to benefit most from BCI technology, Musk is hoping to see NeuraLink’s device used by able-bodied people as well. He envisions a future of “superhuman cognition”, where his BCI is used by all people to do things like control smartphones and other devices without touching them, ultimately allowing for “symbiosis with artificial intelligence”.

Outside the lab

So far, BCI trials have been confined to laboratory settings. In their current varied formats, they have remained too cumbersome to transition into everyday life, but prototype images on NeuraLink’s website show a small, hearing-aid like device worn behind the ear. However, the device doesn’t exist in this form yet and still requires wired connections to function for the time being.

NeuraLink has also built a specialised surgical robot which is capable of inserting the electrode threads into the brain, with the early science funded by US defence funding agency DARPA, as the agency noted in a Twitter statement.

The operation would require drilling holes into a patient’s skull to implant the threads, with the automated machine making sure the correct neurons are targeted. However, NeuraLink hopes to eventually use a laser beam to pierce the skill with a series of tiny holes which wouldn’t be felt by the patient in a procedure “as safe and painless” as laser eye surgery.

NeuraLink claims that these electrodes have been able to record from around 1,000 neurons in a rat. While this does indeed sound impressive, stimulating this many neurons probably won’t be necessary to create the computer-operating BCI device Musk has in mind.

California Institute of Technology postdoctoral fellow Sumner Norman says: “Interfacing with a large number of neurons is the holy grail for many brain-machine interface scientists and electrophysiologists. However, with as few as 30 neurons, patients with chronic implants have already achieved high fidelity control with cursors on a screen and robotic prosthetics.”

So, despite NeuraLink’s claims in their literature that the BCI field has been limited by researchers employing comparatively low numbers of electrodes, this might not technically be the case, although working with larger networks of neurons could still provide important insights.

Norman says: “As the number of neurons we record from increases, there are diminishing returns in the control the patient can gain. For example, the jump in control from 10 to 100 neurons might be similar to the jump from 100 to 1000.

“We won’t know just how much performance we can squeeze out of 1000+ neurons until we try. There are countless unanswered questions surrounding how large brain networks interact; the more of those networks we can read from and write to, the more we can learn about how the brain works.”

Limitations to the technology

But one factor which is limiting BCI technology is its longevity. Cell death and scarring will occur with all brain implants, and the electrodes used in these types of technologies eventually degrade within living tissue and become useless.

“Gliosis, the reactive change of glial cells in the brain to the damage incurred by the electrode, is still a major hurdle,” says Norman. “The interface of rigid electrodes and soft brain tissue is a major factor in this effect. Gliosis actually insulates electrodes over time, reducing their signal quality. This alone can render electrodes useless over the period of months to years. Flexible electrodes, like the ones Neuralink are producing, are an attempt to minimize these effects by allowing the electrodes to move with brain tissue.”

Neuralink’s microscopically slender electrodes are designed to provide longer-term recording than current electrodes and produce less tissue damage, but it’s still not clear how this will stand up over time. In the first study cited, recordings were made from one rat implanted with the electrodes for 60 days, with the implants falling off prematurely in the other three rats studied.

By comparison, the Utah array has reported longevity of six to nine years when implanted into a primate brain.

Musk appears to be more than aware of NeuraLink’s present shortcomings.

During Neuralink’s July 2019 launch event, he said: “It’s not like suddenly we will have this incredible neural lace and will take over people’s brains. It will take a long time.”

But as it stands, NeuraLink has not explained exactly how the system translates brain activity or how the device will stimulate brain cells. The announcement of the technology was based on two white papers published in bioRxiv, reports written by NeuraLink scientists and not peer-reviewed.  In both published white papers, no attempt has been made to use the recorded neuronal data for the purposes of a BCI, and no attempts to simulate neurons through NeuraLink have been reported.

It’s fairly likely the company has achieved more than is reported in the white papers, the nature of scientific research being that breakthroughs are sometimes made years before they can be published. The company has yet to announce anything revolutionary, with the celebrity status of its founder a clear catalyst for its space in the spotlight, but its drive and ambition are clear and the opportunities a portable BCI could offer severely ill and paralysed people are second to none.

Norman says: “There is always some hyperbole when Elon Musk’s name is involved, often of his own making – but then he launches a Tesla past Mars and you have to give credit where it’s due. NeuraLink is built of a real team of card-carrying neuroscientists and engineers that are making significant steps toward the future of brain-machine interfaces. Only time will tell if Neuralink, or any other team, can build the science fiction future of seamless BCI.”