[throw20102010]

A key aspect for me will be whether Neuralink can enable brain control interfaces that don't require sending signals to muscles.

A key example that falls short is CRTL-labs' armband. I've seen a few demos that allow you to "control a keyboard with your brain" while wearing the armband. The only problem is that this setup requires you to move your hands as if you were actually typing on a keyboard. If my hands have to move I might as well use a real keyboard. I realize that there are some people that could use CTRL-labs' armband, such as amputees, but it's not compelling to me.

If Neuralink can let me control a computer by visualizing words or something similar instead of physically moving my hand then I'll call it a win.

[m463]

I thought the ctrl-labs stuff went beyond that. One of their demos had a phone game that you could control without moving your hand at all.

They also talked about virtual arms, and there was one guy controlling nine cursors simultaneously.

[throw20102010]

The CTRL-kit, which is the "real" product they have, uses electromyography, which needs muscle activation to work, so something must twitch at a minimum to send a signal. A good interface may enable controlling multiple things at once, but they would all require muscle activity.

They may have other things in R&D. At least they are closer to having a real product than Neuralink and they started their presentation on time at Re:MARS.

[m463]

Neural interfaces

One of my biggest “Wow!” moments of 2018 took place in the offices of neural interface company CTRL-labs. Their demo involves someone playing the old Asteroids computer game without touching a keyboard, using machine learning to interpret the nerve signals that are sent to the hands. But it isn’t quite what you think. Moving things in the digital realm without moving your hands seems startling enough (though it’s worth remembering that it was once considered remarkable to be able to read silently without moving your lips). But that’s just the first stage. Essentially, users of this technology “grow” another virtual hand, which they can move independently of their physical hands. One of the researchers bowled me over when he said he was “working on controlling nine cursors at once.” Gradually, then suddenly, our children will interface with machines in deeper and deeper ways. Humanity is already going cyborg (see trend 1); expect it to accelerate. Don’t fall into the trap of thinking that AI will replace humans when it can be used even more powerfully to augment them.

from https://www.oreilly.com/ideas/gradually-then-suddenly

[throw20102010]

In the asteriods demo (https://youtu.be/5Z5aZK2C3ew?t=711) the only reason his fingers aren't moving a bunch is because the hand is flat on the table and the table is pushing back. You can see his fingers twitch while he's playing.

This is the biggest overselling point from them. The only people that can grow a "virtual hand" are amputees. Everyone else simply has a shadow hand that follows what their real hand does (or would do if there wasn't a restraint). All of their control interfaces are coupled to what the shadow hand does- i.e., if the shadow hand's index finger curls in, the spaceship spins to the right in the game. In reality it's not a shadow hand they are sensing, but the neurons that control the movement in your hand/arm.

It's a fact of biology that if you are reading nerve impulses from the arm, then to control the computer something must be happening in your arm. Every motor neuron is connected to a muscle. The best we can get is a very sensitive system where your arm doesn't move very much. You cannot (without a million years of evolution) send signals to your arm without it moving. If you were to cut the neurons off from the muscle, you could then send activations without your arm moving. However, you would also lose physical movement ability, and nobody is doing that (except for amputees).

[colechristensen]

I was under the impression that your fact of biology was false.

Specifically that the brain-muscle mapping isn't preprogrammed and the brain learns to control what is there.

For example people with fully formed extra digits can just have a fully functional extra finger.

This does get baked in pretty well but the adult brain can re-learn after a catastrophic injury so it's not hard to believe you could figure out how to add virtual appendages or entirely novel "extremities".

[throw20102010]

It's not false. You can't send an action potential down a motor neuron connected to a muscle without a reaction from the muscle. Do not confuse the neurons in your brain (which generate new connections all the time) with motor neurons, which is what the CTRL-labs kit is sensing.

The brain does learn to control what is there, and what is there is attached to your muscles. I'll say it again: the CTRL-labs kit works on electromyography, which only works with muscle activity. You will not magically grow new neurons in your arm that are dedicated sole to the control of a computer.

[m463]

You're right. That said, surgery complicates things. It's a big commitment, it will take time to get approval and it doesn't scale as well for mass production.

[Robotbeat]

Neuralink has ambitions to address those issues, but I think it will take decades. And I don’t intend to be an early adopter.

[buboard]

it seems they are targeting motor and sensory cortex. Those do get activated by imagining the task . but it will be a very fuzzy process and require a lot of training. It remains to be seen if 1024 electrodes provide sufficient readings to be useful

[vcavallo]

with Neuralink you do not physical move.