Imagine a world where blindness and deafness can be reversed. Or never forgetting a name, face, or date. Being able to augment the brainpower we’re born with to offload memories and correct human maladies. That’s what Elon Musk and the team at Neuralink are setting out to do, but is it the future of humanity or just a dream? Let’s take a closer look at Neuralink.
I’m Matt Ferrell … welcome to Undecided.
So what is Neuralink? It’s a company focused on building out a type of Brain Computer Interface (BCI), which can have many different names and classifications. But it all boils down to building an interface that can read and interpret the signals given off by the neurons firing away in our brains.
This type of research goes all the way back to 1924 when Hans Berger, a German psychiatrist, used electroencephalography (EEG) to record human brain activity. It’s a process that uses electrodes to measure the electrical activity of the brain. The first tests were probably a little … uncomfortable … given that he inserted wires under the patient’s scalp to make the recording. Thankfully, these were later replaced by attaching silver foils to the outside of the head.1 Imagine how those first patients felt when they found that bit out.
It wasn’t until 1973 that the term “BCI” was coined by UCLA Professor Jacques Vidal. In a 1977 experiment he was able to use an EEG to control an object on a computer screen and move it around a maze.2 Since that time there have been quite a few different approaches and advancements made in how to read signals from the brain, as well as write them to it. Like being able to map motor cortex neurons to control a robotic arm, which is something that was demonstrated in the 1980’s with rhesus macaque monkeys. In the late 1990’s researchers at the University of California, Berkeley were able to reproduce images seen by cats by deciphering neuron activity. In the early 2000’s 16 patients received a second generation visual cortex electrode implant to partially restore the ability to see shades of light.1 So you can start to see why so many have been excited by BCI research.
There are three basic forms of BCI: invasive, partially invasive, and non-invasive. The electrode implants that were attached to the visual cortex would be considered invasive. Partially invasive is when an implant is placed inside the skull, but it rests outside the brain and doesn’t intrude into it. And something like an EEG that attaches to the outside of the scalp is non-invasive. On that scale Neuralink would be considered invasive.
How does it work? It all starts with Neuralinks’ “Link,” which is a sealed device about 23mm in diameter and 8mm tall. Inside is a little computer that processes, stimulates, and transmits neural signals. Extending from the coin shaped Link are a series of neural threads with multiple probes/electrodes, which are carefully threaded into the brain. This invasive approach can take a far more accurate reading of brain activity than a non-invasive approach. There’s an analogy that constantly comes up around this. It’s like a stadium where each seat is a neuron firing off and making noise. An EEG is like being outside the stadium and trying to understand what’s going on with the game inside. You can only really tell if a team scored or something exciting happened. But if you move inside the stadium, the voices become clearer and you can start to make more sense of the play-by-play. Move even closer and eventually you can ask a specific neuron what they think of the game.
Built into the Neuralink system is the ability to track things like orientation, temperature, and pressure. It’s able to track a lot of the same things that your phone or smart watch can do. Or as Elon put it:
“It’s more complicated that this, but in a lot of ways it’s kind of a like a Fitbit in your skull with tiny wires.” -Elon Musk
The rationale behind this is that combining all of the collected data will someday unlock the ability to predict early signs of dangerous conditions, like a stroke or heart attack.
It can charge inductively, just like your phone or watch, so it should be easy to top off the charge every day without physical wires hooked through the body. But it’s how they install the Link that’s kind of amazing. They’ve designed a robot that can complete the very delicate threading procedure. The incredible small size and detail work needed is more than a human can pull off reliably, so automated robotics are necessary to do this quickly and safely. The ultimate goal is an automated surgery that can happen in less than an hour.
Don’t worry, I won’t get graphic with the actual details of the procedure (I don’t like that stuff myself), but it entails opening a small flap of your scalp, drilling a small hole through the skull, threading the electrodes and inserting the Link to fill the hole. After sealing the flap of skin, there’s no sign that you’ve had the procedure done at all.
During the demonstration they put on last year they showed a couple of pigs: one with an active Neuralink and the other who had one removed to show how safe it can be. The Neuralink in the active pig was connected to the neurons that controlled her snout, so they were playing real-time audio signals to represent the data they were reading from firing neurons.
“So on the screen you can see each of the spikes from the 1,024 electrodes and then if she snuffles around and touches her snout on the ground, or feed her some food, pigs love food, then you can see the neurons will fire much more than when you’re not touching the snout.” -Elon Musk
They also demonstrated being able to predict the location of the pigs limbs by interpreting the neural data that the Link was receiving with pretty high accuracy. Being able to read and understand someone’s intentions for moving a limb can potentially be applied to restore movement to someone with paralysis. But it’s not just reading data that they’re focused on. They also showed off that each data channel is capable of writing, or triggering neurons by giving off an electrical charge. Again, this could be used to treat any number of disorders.
The overall presentation was impressive and showed off their product design methodology from creating a system that’s safe and easy to install, unnoticeable when you have one, and safe to remove or upgrade. But this is where I need to emphasize what Neuralink is not …
It’s still early days of turning decades of research into a working consumer product.
Their initial goal is to use their technology to help those with paralysis or other injury gain some independence back. Being able to think of typing out a message on your phone to have it actually do it. Think what you want to do on a computer and it happens. It’s kind of like telepathy. This has a direct link back to the research I mentioned earlier. Being able to help people with visual, auditory, or motor control issues is the ultimate goal.
Now Neuralink isn’t alone in this field by any stretch. There are companies like Paradromics in Austin, Texas that are also trying to help people with conditions like paralysis or blindness. Or Neurable, which is a Boston-based startup that’s focusing on EEG style devices to control computer interfaces with a set of brain-sensing consumer headphones.3 And an even more direct competitor is BIOS from the UK, which has a similar implant technology to help enhance and improve different health conditions.
The biggest difference between the other companies and Neuralink is really their approach and vision. Much like every other one of Elon’s companies, there’s a bit of controlled chaos around how it’s run. An engineering-focused, first principles thinking, incredible pressure cooker to push the envelope. And delivering on accelerated timelines reportedly created some internal strife, especially with the people more attuned to the methodical scientific research approach.4 After their last public event there were a few notable criticisms of the event for overselling what’s possible. MIT Technology Review published an article that called it “neuroscience theater” since none of what they saw proposed for the future is doable yet. And what Neuralink did show wasn’t anything “new” or “groundbreaking” in the realm of BCI research.
While the critiques aren’t “wrong,” many of them missed some of the nuances of the presentation. It was stated multiple times that the presentation was a recruiting tool for more researchers, doctors, and engineers. Getting those who want to be part of building the future excited to join the team. It wasn’t meant as a demo for the public to get excited about getting one implanted into their heads tomorrow. Elon also made many comparisons to other “consumer” products. Like the fact that Neuralink has over 100 times more data channels than the next best “consumer” device. Going from research to product is a huge leap, especially in the medical field, so we’re still years, if not decades away from more of the aspirational goals. What Neuralink is doing — trying to bring a product to market from the lab — is a Herculean task.
On the flip side there were also voices that were very impressed by the presentation.
“I am amazed by the rapid progress in the device architecture to enable a seamless prototype in pigs. The best part was seeing the prediction of movement vs actual movement.” -Dr Tennore Ramesh, Non-Clinical Lecturer in the Department of Neuroscience, University of Sheffield5
“In terms of their technology, 1024 channels is not that impressive these days, but the electronics to relay them wirelessly is state-of-the-art, and the robotic implantation is nice.” – Prof Andrew Jackson, Professor of Neural Interfaces, Newcastle University5
But this is where I start get a little conflicted, and I’m looking forward to seeing what you all think about this as well in the comments. I’m torn. I get so excited seeing what we as a species are capable of when we put our minds to it. The fact that we’re at the very beginning of the journey towards mapping the human brain, and potentially being able to cure diseases and disabilities, is awe inspiring. Neuralink may not be revolutionizing specific BCI research yet, but they’re tackling the engineering feat of making it a reality.
But some of what could be a potential down the road is as terrifying as it is amazing. What will it mean if we can actually offload memories, or load memories that aren’t even our own. I got flashbacks to the movie Strange Days with Angela Basset and Ralph Fiennes where there’s a black market for memory trading. And the line between what’s real and what’s not, like a fabricated memory, really starts to be blurred.
Like I said … this technology amazes and terrifies me at the same time for what it could mean decades down the road. It doesn’t help when I hear that Facebook is working on BCI technology too. Yeah, I don’t want Facebook anywhere near my neurons. If we don’t act responsibly and ethically today with the development of this technology, and put proper guidelines and controls in place, we’re creating ripples into the future that could turn into damaging waves or tsunamis. At the very least …
“This is sounding increasingly beginning to sound like a Black Mirror episode, but … uh … well they’re pretty good at predicting. Essentially, if you have a whole brain interface, everything that’s encoded in memory you could upload. You could basically store your memories as a backup and restore the memories. Ultimately you could download them into a new body or into a robot body. The future is going to be weird.” -Elon Musk
Bringing it back closer to today, there’s so much amazing potential with what Neuralink is working on. I’m looking forward to hearing and seeing future updates from the company, as well as their competitors.
- Wikipedia – Brain–computer interface ↩︎
- Real-Time Detection of Brain Events in EEG – Jacques J. Vidal ↩︎
- Neurable raises Series A to build an “everyday” brain-computer interface ↩︎
- STAT – As Elon Musk’s Neuralink prepares to draw back the curtain, ex-employees describe rushed timelines clashing with science’s slow pace ↩︎
- Science Media Centre – Expert reaction to Elon Musk’s Neuralink demonstration involving pigs ↩︎