Hacker News Reader

Comment by vansul | Hacker News Reader

To me (neuro undergrad, comp sci masters), the demo seemed underwhelming but encouraging. As far as I can tell the tech was not doing anything particularly novel and the 'fitbit in your skull' analogy was personally uninspiring. [edit: the volume of reads /form factor is a step forward]

That said, the demo was mostly for attracting talent - lots of money and a great team will likely get them somewhere. In terms of their long term goals I expect the area that will give them the most trouble is reading/writing interesting stuff on the cortex. Neural coding is really hard and poorly understood.

View on HN

I have problem with the two different narratives.

Neuralink has two goals:

1) short term. better BMIs to to treat serious brain diseases with more bandwidth and more and better electrodes. I can see this happening.

2) long term. Invasive BMI for something other than treating diseases. Better electronics is not going to make invasive brain implants safe for consumer use, they just reduce certain risks.

To solve (2) Neuralink must solve fundamental issues in medicine related to body implants and brain surgery. It would be groundbreaking and probably revolutionize medicine even outside brain implants. Neuralink researchers might get Nobel in Medicine.

Breakthrough in brain surgery. To get FDA to approve brain surgery (making holes in the dura mater) even brain surgery to open and close the brain for no good medical reason would be insanely hard. Any form of brain surgery has lots of risks. Bleeding in the brain, seizure, scarring of the brain, immediate infection risk and late infection risk. Inserting lace with thousands of electrodes is huge operation even if you could do with with endoscope.

Breakthrough with implants. With the BMI ther are issues with coagulation and constant low level inflammation. It's both health risk and gradually degrades the effectiveness of electrodes. Electrode must be in contact with tissue and interact with it. There is risks even with teeth implants, artificial knees and hips. They collect bacteria around them and they are just inert objects.

Cochlear implants are established technology, and they also interface with nerves, just not with brain matter. Is this really so much harder?

The device stimulates cochlear nerve it's not brain surgery.

Just like with any other implant, the risk must justify the benefit. Cochlear implants have 3% rate for major complications, like life threatening meningitis or facial paralysis (caused by nerve damage). Potential benefit justifies the risk.

To expand a bit here: Cochlear Implants are inserted into the cochlea, the curly organ in the ear that is responsible for hearing.

The other organs in the ear are the vestubular organs which give you a sense of balance.

The little wire in the implant has electrodes along it's length that shock the 'damaged' cells in the cochlea, bypassing how those cells sense sound, and just directly stimulating them.

The reason that it's a bit 'crude', is that the shocking is not specific to the nerves, it just shocks a bunch of those sensing cells. The reason for this is a bit involved and is due to the tonotopy (sound-map) of the cochlea.

The implantation is relatively simple for a trained surgeon. It can be done on small children.

https://www.cochlear.com/us/en/home/diagnosis-and-treatment/...

Well articulated points - pushign back slightly regarding your breakthroughs on implants, there's good data to suggest that the size scale that Neuralink is now operating at does not elicit the same gliosis (reactive tissue growth) that electrodes larger than 15-20 µm do. The infection risk is heavily mitigated by the device being fully implantable, but you're absolutely right that there is some risk there.

Smaller size reduces the penetration trauma. I wonder how they are going to reduce chronic tissue response?

Even if the surgery does not result bacterial infection, insertion injury and the pretense of a foreign object often leads to persistent low level inflammation.

There is so much basic research questions that simple engineering is not going to solve.

> the tech was not doing anything particularly novel

Where are all these people apparently doing long term, low latency wireless 1024 channel ephys with stim capability?

Apologies I spoke too soon, you're absolutely right that the volume is an impressive achievement.

Reply on news.ycombinator.com