I want to talk about Neuralink and Musk’s brain chips, but first let’s take a trip back 50 years...
Wikipedia “The Terminal Man”
The events in the novel take place between March 9 and March 13, 1971. Harold Franklin "Harry" Benson, a computer scientist in his mid-thirties, is described as suffering from "psychomotor epilepsy"[4] following a car crash two years earlier. He often has seizures followed by blackouts, and then wakes up hours later with no knowledge of what he has done. During these seizures, he severely beats two people; the day before his admission, he was arrested after attacking a third. He is a prime candidate for an operation to implant an electronic "brain pacemaker" in the amygdala region of his brain in order to control the seizures, which will be performed in the Neuro-Psychiatric Service (NPS) of University Hospital. Two NPS surgeons, John Ellis and Robert Morris, are to perform the unprecedented surgery.
Wireheading
Michael Crichton explored the implications of implanting forty (40) electrodes in someone’s amygdala with an itty-bitty computer (barely a computer even by 1972 standards) to control them. At this same time in science fiction history (1969), Larry Niven previously explored in “Death by Ecstasy” the implications of “wire heading” where an electrode is planted in one or more brain reward “pleasure” centers wherein the lab rat person prefers electricity to food, sex, and literally anything else.
This was supposed to be better — simply a way to stimulate the brain when an epileptic episode was coming on (although what we know now about epilepsy is that it’s more of a case of insufficient inhibition to suppress a signal storm than something you can “head off via stimulation.”). Make the patient have a better life, etc.
Cautionary Tale
As a fifty-year old cautionary tale, this has merit to explore.
After resting for a day, Benson goes through "interfacing". The electrodes are activated one by one to test which ones would stop a seizure. Each electrode produces different results; one stimulates a sexual pleasure. Gerhard, one of the technicians administering the test, shows his findings to Ross, who discovers that the seizures are becoming more frequent. She explains that Benson is learning to initiate seizures voluntarily because the result of these seizures is a shock of pleasure, which leads to him having more frequent seizures. Ross further discovers that, due to a clerical error by the nurses, Benson has not been receiving thorazine. She then finds out that Benson, using the wig and disguising himself as an orderly, has evaded the police officer assigned to guard him and escaped from the hospital.
It’s all downhill from there, complete with antique IBM 360 computers (think as powerful than a Mac Plus), microwave ovens to disrupt electronics, and Chekhov’s gun.
Needless to say, the American Epilepsy Foundation didn’t care for Michael Crichton.
Deep Brain Stimulation
There has, of course, been additional research in the last 50 years.
Mayo Clinic:
Deep brain stimulation (DBS) involves implanting electrodes within areas of the brain. The electrodes produce electrical impulses that affect brain activity to treat certain medical conditions. The electrical impulses also can affect cells and chemicals within the brain that cause medical conditions.
The amount of stimulation in deep brain stimulation is controlled by a pacemaker-like device placed under the skin in the upper chest. A wire that travels under the skin connects this device to the electrodes in the brain.
Deep brain stimulation is commonly used to treat a number of conditions, such as:
- Parkinson's disease.
- Essential tremor.
- Conditions that cause dystonia, such as Meige syndrome.
- Epilepsy.
- Tourette syndrome.
- Obsessive-compulsive disorder.
While the life-expectancy of someone getting DBS (at low levels of stimulation) is on the order of 50% after 12 years, patients do report seizures, chronic pain, headaches, depression, and even dementia, as well as the risk of infection — infection beyond the blood-brain barrier — which is very life-threatening.
Neuralink
The reason I’m writing this is to lampoon Elon Musk and his chipping humans. Monkeys playing pong. Yes, we have learned a great deal about the human brain in 50 years — never-the-less we know very little. Even our attempts at AI only model one structure of the brain (Cerebral Cortex), and there are many others. Our knowledge of the human visual system is still limited to the retina and the first few layers in the occipital cortex. fMRI studies show some 30-40 active regions in the brain as part of our “operating system.” And the chemistry we use to affect depression, epilepsy as mentioned above, and addiction are also quite limited. (Basically bear skins and stone knives.)
Was this faked like the original Tesla self-driving promo?
This is a case of playing with something we don’t understand sufficiently — yes, it may be possible to train a monkey with electrodes to move a mouse cursor and there is some efficacy from being able to do this with human quadriplegics. But we also know that Musk has a history of overpromising and then delivering something that may be worse than nothing: Tesla Autopilot “Full Self Driving” where the customers are doing the beta testing, Tesla Solar Roof Tiles, Hyperloop (a political ploy intended to nix public transportation?), The Boring Company and flame throwers (more nixing of public transportation?), and of course colonizing Mars. (For one, you can’t get enough nitrogen there to support an ecosystem because nitrogen is needed for every single protein, not even considering radiation, perchlorates, low gravity, a one-way trip, and being a virtual company slave.)
Air Force “Wires on the Brain”
The US Air Force has been working on human/machine interfaces for decades (I first heard of it in the 1980’s at UT-Austin). Understand that pre-AI, a primary limitation to a fighter jet was the G-forces it placed upon the pilot. If the pilot can’t move the controls under high Gs, then s/he can’t maneuver the plane. Back then they were putting “cuffs” on neurons (dendrites, really) and inducing as well as monitoring signals. After awhile the dendrites would atrophy, but that was 40 years ago... As you can see, I’m a bit out-of-date as my last graduate school electrophysiology class was in 1988.
The other thing the Air Force has been working on (here’s a 2016 article from Smithsonian Magazine) is transcranial direct current stimulation, or tDCS. In this case they induce currents from an array of externally placed electrodes (not requiring surgery!) that stimulate the cortex with pulses that (might?) cause neurons to fire.
The electrode-connected subjects were then given an Air Force-specific version of the Multi-Attribute Test Battery, which was developed by NASA to evaluate people’s ability to do multiple things at a time in an environment similar to conditions experienced while in flight. They had to turn lights on and off, alter the channel and frequency of a hypothetical communications unit, target a cursor and manage resources all at the same time.
Within just four minutes, the people with ongoing stimulation were doing better on the test, and they performed better overall. The researchers noted a “significant” improvement in information processing in the subjects who got ongoing brain stimulation, and they concluded that the technology “has the ability to augment and enhance multitasking capability in a human operator.”
It might have worked better than caffeine. Maybe. At least it didn’t involve surgery.
Efficacy
Whom are we kidding? For this to have efficacy, the people treated have to be better off than they were before.
But for a business to actually function, this isn’t treatment for a small number of people — this has to be elective surgery with a large enough customer base to be profitable and therefore make Musk happy.
Consider the negative side of the equation: how dangerous is it for the alleged benefit it provides?
First, I’ve learned through 40+ years of using computers that you should never depend upon version 1.0 of anything, much less beta software. With some companies it’s more like version N.3 — wait for the second version and at least three updates before installing. Second, to have any significant statistical power in these trials it can’t be on just one or two subjects. It has to be on hundreds of patients, be run double-blind, and for a significant period of time (really long longitudinal study — or the study size has to be thousands). That isn’t going to happen quickly or cheaply (or with any kind of efficacy). There aren’t that many quadriplegic patient candidates, and how do you justify a placebo?
Don’t forget, Musk knocked up the President of Neuralink, which makes the entire prospect even more suspect. And it’s a badly run company. The red flags abound.
🚩🚩🚩🚩🚩🚩🚩🚩🚩🚩
Prediction
Fusion power will come before full-time self-driving cars (level 5), and those same self-driving cars will come before we have viable brain implants.
In Crichton’s context, a Terminal (noun) is the public facing front end of a computer, and his protagonist became such a device. In the Neuralink context, it may very well be where someone goes... The terminal is at the end of the journey.
A terminal (adjective) patient.
That’s a giant NOPE from me.