Hacker News new | past | comments | ask | show | jobs | submit login
Monkey MindPong (neuralink.com)
678 points by ag8 28 days ago | hide | past | favorite | 276 comments

For those who don’t know much about this field, this seems incredible. The reality is that this a tech demo that Neuralink is using to practice the basics. For other research groups in this field, this is tech from 20 years ago.

The current state of the art is humans, mostly quadriplegic people, with chips which let them control robot arms with enough accuracy to pick up small delicate objects, play games, and type by thinking words. Current state of the art implants let people get tactile feedback too - they can intuitively know the position of the arm they’re controlling and feel the amount of pressure on each finger tip.

I had the pleasure of meeting a man with an implant in 2016 and the technology he was using was incredible. He could pick up a grape with his arm without crushing it and then eat it. If the arm was in another room, he could feel around with it to find objects. He described how the arm just felt like a part of his body when it’s turned on.

To use SpaceX as an analogy: this isn’t a first flight. This is more like a static test fire. If Neuralink follows the SpaceX trajectory of driving down cost/increasing innovation, it results will be really out of this world. Stuff that would just sound bonkers right now.

One of the neuralink innovations is having the electrodes installed very precisely via a robot surgeon, which supposedly allows them to make things easier/better/more precise -


Previous discussion of surgical robot https://news.ycombinator.com/item?id=24311152

That's really cool! I think one of the things that will hold them back is that there are still a lot of issues on even knowing where to place the electrodes - but as those issues are solved this will have such amazing potential!

It sounds like they know exactly where to put it based on their white paper. The problem is more materials engineering for longevity and economics.

They know where to put them for the tasks they are targeting right now - tasks which have already been performed in the lab. Beyond that is a open question

It seems that the ideal way would be trhz gap if it can be done. That way no implant needed but maybe a dome.

I think that's a little unfair. I guess what they are doing is achieving the basics, but with a different interface. This is an important proof of concept.

Remember, Musk is all about scalable manufacturing and products. If this was other's tech from 20 years ago, where's the massive commercialization? They failed. Musk is focused on this aspect, and he will (given his past achievements) succeed.

This is really basic. Here is a harder task from 2015 in monkeys with only 96 implanted electrodes (Utah array): https://youtu.be/lVFWIv5lupo?t=1673

Here is a video from 2014 of someone controlling a quadcopter with their mind with zero implanted electrodes: https://www.youtube.com/watch?v=lsraC04Mm8w

That's akin to saying that phones are basic since we already had desktops in the 1990s.

Having 10x more electrodes that stay active for a long time, attached to self powered unit under skin with low latency wireless comm link is a major step forward.

It’s not that it’s nothing new - it’s obviously not identical to previous BMIs. But the thing people are reacting too is mostly the MindPong part (genuinely basic) and not the 10x electrode part (novel). It’s not like a paradigm shift, it’s a practice run for them to see if they can replicate previous successes with some changes to the tech.

That's what I'm saying. Exactly. I don't think it's "easier" to use 2000 electrodes in a microneural mesh.

Besides that, why the pushback? Is it coming from field experts who are like, "bah, old hat as shiny new. humbug!"; disappointed futurists who expected more ("we were promised flying cars and intelligent robotic spouses, and all we got was 2000 electrodes in our brains, and 140 characters"); misMuskists; people who sense unfairness in praises lauded upon this, in their mind, mediocre achievement, echoing to them a theme that the already-anointed and, in their view, undeserving, accrue all praise, while the genuinely meritful, themselves, for example, are overlooked, yet again?

I'm going to echo that this is not really as impressive as neuralink is saying. It's a good incremental improvement, but no breakthrough, and enough people have shown similar results. It's still quite far from achieving Musk's dream of uploading his brain into a robot body to fight the AI Singularity.

The professor in that first video is a consultant of Neuralink as well now, FYI.

Also! Forgot to put this in my comment but here is an interview (a couple years old) with the man I met talking about what he thinks of NeuraLink: https://www.technologyreview.com/2019/07/19/238809/nathan-co...

Would it be possible to control an extra arm somehow? Or brain plasticity doesn't go that far?

The brain is very plastic, it probably would adapt at any consistent input. A most interesting question is, what would you sacrifice?

Because the part that you would use to control your extra tentacle or whatever is probably currently used for something. If you are missing an arm is obvious what part of your brain you would use, otherwise, some sacrifice probably have to happen.

Even reading, it seems, take the space that is used in people that can't read for some detailed visual processing (1).

Maybe you could add some extra processing in the extra arm, so it's more "cheap" to use.

(1) - I learnt that in "Reading in the brain" by Stanislas Dehaene.

I don't think we necessarily have to sacrifice something. We learn new skills and knowledge all the time. Using the new limb could be a skill. Maybe it's not as dexterous as the motor cortex and real hands at first, but there might be some benefits over natural hands.

I think that you sacrifice something when you learn a new skill. The brain have a finite capacity and it just forget not used facts or skills, and the reason it's forgetting is because it's always learning.

For instance, I'm learning German and I can feel how is messing with my English, specially with the spelling (English is not my first language).

I think memory retention has more to do with simplicity / emotional consistency.

The new skill you learn may reinforce existing areas if its components are transferrable. Also, I see schemas as compression in the brain; eidetic memory shows how much some brains are capable of.

Forgetting is very good for clarification. I think your example suggests that your spelling in English hasn't been established deeply enough; I understand why you might have an adverse reaction to learning it though (German spelling is far more reasonable).

But if it has a finite capacity, when does it fill completely and with what? I guess it's entirely possible that only kids who are growing up can learn to use the extra limb and adults won't be able to.

I don't think it never fills completely, it just empties the older entries to make place for the new ones.

Also part of why voice interfaces didn't take off for complex tasks, talking about performing the task interferes too much with the mental process.

The real innovation is when you can find a method to control the physical hand and the virtual hand independently.

I suspect it would be possible if you can do some kind of real-time feedback/tuning of the ML model. Reinforcement learning stuff.

By slowly adjusting the model to reward hitting the ball back in pong, but also to penalize moving the real physical hand, you might achieve it.

Can't the brain alone use visual feedback to close the loop, separating both hands? If thinking in a particular way activates the only the robot hand (trembling), I'd guess that with sufficient plasticity you should be able to start enhancing control of it.

Obviously tactile and proprioceptive feedback through the chip would enable better / quicker control, given the sufficient plasticity hypothesis.

Absolutely. The first test subjects of this technology were chimps who weren't paraplegic, just controlling extra arms.

I bet it is possible but that does not mean that it is going to be easy. Try training yourself to wiggle an ear or to lift a single eyebrow if you cannot do that yet. It's possible but it takes quite some time.

Many monkeys use their tail as an extra arm, so, should not be a problem.

I thought the ctrl-labs guys did that.

this is why we can't have nice things... :)

If the arm was in another room, he could feel around with it

This uncanny sentence is giving me dystopian vertigo. I picture a mob of derelicts home-invading a 130 years-old tech boomer, fighting his remote-controlled spare limbs from room to room..

Heinlein’s Waldo might be a good thing to read, then. It’s the origin of the word “Waldo” in English.

It is a cool story, but I had no idea that remote control arms were called waldoes when I read it haha.

Thats a nice idea for a movie.

I had no idea these things were possible. Are they currently insanely expensive / only experimental? How available are they?

Very much insanely expensive and only experimental. The man I met had a team of like five people needed just to set the thing up and calibrate it for his "brain waves" that day (they had to constantly fine tune the neural net he used to control it because the human brain is so plastic it would change every day and become unusable after a couple days) - plus more researchers running the experiments he was doing. The team had a portable supercomputer running the neural net (this thing was maybe a two or three server cabinets on wheels) that they brought around with them. They had a safety managers to make sure the arm didn't hurt anyone and make sure he was medically okay. This was not a device he used around the house, he was an employee of the research lab and using it for experiments was his full time job.

The risks of these devices are also significant: you have to get risky brain surgery to install them and then you might still reject the implants and have to get them removed.

A very limited number of people have these - the lab this guy worked at had I think about a dozen people, all with BMIs installed using different surgery techniques (some had the implants on the areas where the brain met the nerves going to the muscles so the output to the neural net was a more "direct" representation of movement whereas others had implants on motion planning areas on the brain giving a more "high level" output to the neural net). There are several such labs in the world. My best guess is there are less than one hundred BMIs installed on people today, as well as quite a few more on animals.

Not trying to be funny here: lunch with those guys must’ve been absolutely fascinating. When I worked at Microsoft a couple of decades ago we all loved the job and we talked about work even on breaks. To see the staff of the place you’re describing just hanging out must have been extraordinary.

It was! The guy with the implant did his undergrad in nanotechnology and had pretty much became an expert in BMI stuff so they were a very cohesive team. They had a lot of funny stories too - like when they accidentally put the mind control monkeys in the cage with the cocaine addiction research monkeys and got to see how cocaine affected the brain interface (it didn’t).

IIRC this is only experimental. The most reliable/interesting stuff are relying on ECoG (== put electrodes on the surface of the brain), and this means you have to perform a surgery.

For human applications, most of the time the researchers are using patients that requires several brains surgery (e.g severe epilepsy) so that they can do some sort of fly-by experiments (several surgery are needed so that they can also remove the electrodes).

I studied this at uni; my knowledge is probably not up to date.

Thanks - so it seems like it's actually the ease-of-implant part that Neuralink is really doing most interesting development on then, rather than the actual abilities it gives (so far at least)

Out of my range here, but I bet they have a deeper and wider base to draw on for the AI interface with the brain than most other organizations through their CEO.

I would agree, with the further caveat that musk is the guy who builds the machine that makes the machine.

You really shot this down with your comment until the end. I’m curious what you could see happening with this tech in 5/10/20/30 years? Particularly the “bonkers” stuff.

I’m asking in part because Elon’s suggested this could eventually lead to whole brain uploads and downloads. Partly asking because it’s pretty easy to be very destructive with a comment like yours and offering a glimpse of where this stands out from the crowd balances the attack.

The chips you describe, are they responding to brain activity, or are they connected to something outside it, and if so, is that remnant muscle tissue or remnant (peripheral) neural tissue? And if so, do you agree that if that's missing, this is an unprecedented (possible) solution?

Both techniques have been used! Some are connected to the top of the spinal cord where as others are on different areas of the brain. Both techniques work and it's not clear which is superior right now.

You're confusing apples and oranges here. Those implants you're referring to are at the injury site (arm stump for example), not in the brain. Completely different technology.

If they handle the software anything like they do for the tesla this will b a massive net negative.

I'm really interested in learning more about the process of putting existing research into practice as a process from an economics and innovation point of view.

Does anyone have book, talks and other material you'd suggest looking at?

Some great places to look:

The Annual BCI Research Award highlights interesting advances each year

"Search for Paradise: A Patient's Account of the Artificial Vision Experiment" is an older book (2012) about a research subject's personal experience with an attempted vision-restoring implant

The research put out by the University of Pittsburgh Human Neural Prosthetics Program and Braingate at Brown University

Sounds like you are asking about tech transfer capabilities?

As a starting point, here is a good intro article from WIRED that you might find useful:


Now imagine if someone could control an implant installed on a Mars probe.

Stupid speed of light, making it lag so much.

This seems to be very similar to the early 2000s demos at Brown University/Cyberkinetics[1]. In 2008 they demonstrated it in humans[2] after lots of work getting FDA approval. The core difference here (from my non-expert reading) seem to be electrodes are along a thread rather than on a chip.

See https://www.youtube.com/watch?v=TJJPbpHoPWo for a video of the pong demo in humans.

The motor cortex seems to have a clear, learnable signal though it is rather noisy sampling <1000 locations.

Mary Lou Jepsen at Open Water has ideas about how to do this non-invasively [3].

[1] https://www.braingate.org/publications/

[2] https://www.brown.edu/Administration/News_Bureau/2006-07/06-...

[3] https://www.ted.com/talks/mary_lou_jepsen_how_we_can_use_lig...

IIRC the major challenges was brain plasticity itself. Brain signals changed over time and the device needed new calibration.

I'm curious - does taking a signal straight from the brain reduce reaction times?

Is there a big loss of time sending the signal down to the arm, or is the reaction time mostly cognitive anyway?

Yes. Nerve signals travel at about 50-60 m/s through the limbs, so it takes about 20-30ms for a signal to travel from brain to the hand muscles.

Source: https://en.m.wikipedia.org/wiki/Nerve_conduction_velocity

Still better than Sonos Multiroom Speaker sync (70-75ms).

So Neuralink have a large secondary market of FPS players!

Primary being the army who wants soldiers to quick-fire real guns

And about to be incarcerated cyberpunk guys.

Can somebody who is knowledgeable in this field of work explain how/if neuralink is better to the previous approaches and if its easier to "scale" this tech compared to previous works ?

This is primarily an advance in the systems integration required to make such a sensor fully implantable abs capable of wireless transmission. It also ~10x’s the number of channels of the Utah array, which is the primary point of comparison as a product that can currently be used in human patients (in an investigational context as part of clinical trials). The Utah arrays do not have any active electronics however, and pass a bundle of wires through the skin, which has obvious disadvantages and scaling limitations.

NL’s approach is scalable due to their robotic insertion system which can implant a (multi-channel) thread every few seconds. It should be possible to hit a few thousand channels within the window of a few hour surgery. They do face the same challenges with size, weight, and power that everyone else does, which forces trade offs on the bandwidth, ability to isolate spikes from individual neurons, and number of active channels.

The primary limitation of this approach is that the needles cannot easily insert deeper than the outer layer of cortex (to my knowledge). This limits the application space to anything with recording or stimulation targets on the surface. Motor prostheses and gaming are perfect for this due to the anatomy, but many other medical applications require deeper targets, which their sensor cannot readily hit at the moment.

This is a great summary. I will just add that my understanding is that Neuralink is heavily focused on surgical speed, and a few seconds per thread is the slowest it will ever be. It will likely be much faster than that.

The article describes how it has advanced: it's wireless and reads relatively more electrodes. Producing the device is something that can probably scale up easilt . 1024 electrodes is still nowhere near enough to decode precise movements imho. These pong experiments have been done in humans since 20 years ago, but real fluidity seems very difficult to achieve , because signals are noisy, the brain adapts, distractors are always present etc. The article doesnt mention how long it takes until the device needs recalibration. OTOH brain surgery is not something easy to scale and i doubt people will accept surgery for implanting 2 devices just to play pong. Such a device should be precise enough to allow using a robotic arm effectively.

I think the applications of this are going to be in research for years, not as implants that paralyzed people can readily use.

In this video, it looks like they are probably only sampling ~200 locations (assuming that's a 1080p screen and they have gone for one pixel per signal)

That was difficult to watch. While the technology is impressive, it is sickening to watch. The part where they can pair the phone via bluetooth made me want to vomit. It is obviously unnatural, but I can't quite put my finger on it. Something about seeing that animal, oblivious to what's happening, and that implant with the ability to directly record thoughts and/or remote control. The whole thing doesn't sit well with me.

This is nothing compared to what we were doing in lab several years ago. Neuralink is the least invasive of the invasive ways to conduct this research. I had to put animals down, because they would be defunct after trying to remove those bulkier devices. Neuralink is relatively humane, speaking as someone who has seen this technology in its infancy.

You're right. This monkey gets to earn banana smoothies all day. I'm sure it is quite happy.

Much less happy are the animals we eat, those confined in zoos, and many of the pets people lock in tiny spaces.

The monkey probably won't be sad. I think it is entertained all day by the researchers. Playing its games, and drinking its delicious banana smoothie. As you say; Ignorance is bliss.

> The monkey probably won't be sad. I think it is entertained all day by the researchers. Playing its games, and drinking its delicious banana smoothie.

Rhesus Macaques [0] have a lifespan of thirty years. Thirty years. Thirty years in a massively controlled environment, often sedentary, with limited opportunities to socialise with peers, and the simple pleasures of mutual grooming, lying in the sun etc. Lab animals that are rehomed in rehab centres / zoos etc are often overweight, in poor health, and have a range of nervous tics and social inexperience. Many adapt to the non-lab environment, but they tend to have underlying health problems and issues with acceptance by the alpha individuals because they lack the years of experience required to understand and fit into the complex group etiquette.

Animals may still be the best model for testing drugs / devices that will go into humans, but lets not understate the massive cost to the individual animals concerned.

[Source: personal involvement with a primate rehab centre]

[0] https://animalia.bio/rhesus-macaque

Neuralink monkeys are housed with at least one other monkey for grooming / social activities, and are in the same room with visibility to other monkeys. Are they as happy as zoo monkeys? Probably not. But they're probably much happier than most other research monkeys.

> Thirty years in a massively controlled environment, often sedentary, with limited opportunities to socialise with peers, and the simple pleasures of mutual grooming, lying in the sun etc

Sounds a lot like work, to be honest.

Sounds exactly like my experience trying to reintegrate into society.

I feel like an animal trapped on Earth zoo. It’s relative.

Perhaps it is blissfully ignorant of us to think of ourselves as free on this rock. Does the monkey have a better life now than before? I don't know. The monkey will not mate and likely does not have peers. I wouldn't be surprised if those things are required for happiness. Social and sexual needs are programmed in us.

I think it's a pretty tough topic in ethics but when I was younger, I didn't think much about it because we were doing science. I remember seeing my first rat die. I didn't feel bad at the time because we did it "humanely" with gas (and it's normalized and I was only 20) and they just sort of drifted off and defecated. It's an interesting and perhaps sad use of a life.

Zoos are bastions of conservation. They are usually only filled by rescue animals, and their funds are often used in conservation and animal welfare programs.

is that really always the case? do you have some data about it I can look into?

I stopped going to zoos because I didn't want to contribute to keeping animals like that, but would like to read information to the contrary if that's the case.

I imagine it depends on the zoo

I’ve never been to zoos outside the UK, but here often the primary function is conservation - both of the animals, which are often rescues, and of wider wildlife which is funded by ticket sales

For example, London Zoo is managed by the charity Zoological Society of London [0] and places like Monkey World are essentially rescue centres you can visit [1]

[0] https://en.m.wikipedia.org/wiki/Zoological_Society_of_London

[1] https://monkeyworld.org/

> places like Monkey World are essentially rescue centres you can visit

Can confirm this (disclaimer- I'm a supporter). Many of their animals are rescues, including more than 70 Capuchins that had been lab animals in Chile and four groups of Chimpanzees, many of whom had been rescued from use in circuses or as tourist props - the latter often with teeth knocked out so that they couldn't bite the punters. There is currently a sadly growing collection of marmosets, most rescued from the UK pet trade after tipoffs from animal protection agencies. Many of the marmosets have diseases such as rickets resulting from their owners' lack of animal husbandry skills (e.g. thinking that all they need to eat is bananas). Most of these animals lack the survival skills or health to be released back into the wild. On a more positive note, Monkey World is also a hub for breeding critically endangered species, e.g. Woolly Monkeys and Orangutans.

Sadly, if an animal is being cared for at Monkey World, it generally means that the specific individual has been abused in the past and / or the species faces functional extinction in the wild.

I think it depends on the individual zoo. It's up to them to take part in conservation programs or to stick to keeping animals exclusively for display.

The one where I live has historically been a bunch of very desolate animal display cases, but they've committed to doing what amounts to a slow U-turn, and they have done quite well in that regard.

The old enclosures were clearly built to keep animals in plain view at all times. They've remodelled a lot of them since, and they're completely rebuilding some. There are times where you don't see a single great ape during a whole visit, because their new habitats have some caves and comfy areas hidden from view, with the result that especially the gorillas now appear to have quite a bit of fun interacting with visitors at the four separate points in their enclosure where that is possible and definitely seem much more relaxed than the previous generations (as far as I can tell, I'm no gorilla myself).

Certainly still worse than a life in their natural habitat, but they are part of a multinational conservation project and all their current gorillas were essentially sent to them via this project, with the end-goal of (as I understand it) having a viable captive gorilla population in zoos around the world so there is a "backup" in case the natural populations collapse. The living conditions they provide have earned praise from experts (for what that's worth, seeing they're not gorillas themselves, either).

They still keep their lions in a tiny, cruel pen, but they're building a new one currently that is supposed to be state of the art. Most monkeys have gotten new accomodations last year I think and they're pretty involved in keeping Kunekune pigs from going almost extinct a second time and they're breeding Visayan Warty Pigs (which are absolutely amazing things, and critically endangered); I guess for a relatively provincial central European zoo with a limited budget, that's quite decent.

They've also recently completed a pretty big section with heirloom breeds of common farm animals and they do a lot of education programs and events for schools, which I feel is something my generation missed out on big time, not necessarily from a zoo, but some kind of getting in contact with animals other than the occasional cat or dog might have been quite helpful. I feel there are some really weird misconceptions about animals that are pretty widespread among people my age.

For what it's worth, I've been quite opposed to that zoo in the past, but their efforts over the last decade or so have been enough to convince me to pay for a year pass. Some zoos are much slower adopting this approach, I surely wouldn't be as supportive of one of those.

Won't someone think of the monkeys!

> The part where they can pair the phone via bluetooth made me want to vomit.

I had the same reaction. I just recently had a lot of trouble pairing Airpods to Macbook (both no more than 2 years old). If Apple can't make it work between their own devices, it's doomed. A piece of shit technology, a black stain on the whole stack.

The demo is impressive otherwise, and I don't think it's weird to control a device with your mind. Or, if weird, it's the interesting kind of weird.

I'd like to see a multi-modal GPT successor that learns not just text, image and video but also neural brain signals. It's one modality we haven't touched on yet. Maybe it will be able to extract speech directly from the brain, which is orders of magnitude harder than controlling a joystick.

> Maybe it will be able to extract speech directly from the brain, which is orders of magnitude harder than controlling a joystick.

Check out https://www.nature.com/articles/s41586-019-1119-1

And yes it is definitely orders of magnitude harder than controlling a joystick

Imagine the wonders it would do for interrogation /s

There is a big difference between decoding covert and even "imagined speech" and decoding "thoughts". In covert/imagined speech decoding, the user actually tries very hard to imagine the action of speaking and moving the corresponding vocal articulators, without actually doing it. This is similar to mental motor rehearsal done in sports training.

It works because there is direct correlation between the speech-motor cortex and how the vocal articulators (larynx, tongue, etc) move, and how that combine to produce speech.

Abstract "thoughts", on the other hand, are not so straight-forward. For one, there's no central location in the cortex where the concept of "car", for example, lives. The distributed representation of abstract thoughts within the brain makes it orders of magnitude more difficult than decoding speech for one specific individual. Then add orders of magnitude of orders of magnitude to generalize that to different people.

Maybe you're onto something. Assuming your intent is to obtain information from a captured enemy at all costs, an implant like this is probably far more humane than waterboarding or actual torture.

Also, unlike torture, it would work.

I assume your /s was meta-sarcastic.

With a few decades (years?) of refinement, this technology absolutely has the potential to capture private thoughts.

Sure, it may require an implant, but just like we all carry tracking beacons that have become indispensable to daily life today, in 20 years it may be “the done thing” to have an implant to control your home automation, etc. Or maybe the tech will improve to where it just needs a hat. And at that point, certain authorities will have no qualms about using it in interrogations, with or without “due process.”

Sounds like the point in history where we should be practicing doublethink explicitly. Is this the evolutionary niche psychopaths are meant to fill?

Btw this appears to be a common failure mode for AirPods. You need new ones :(

Yay for planned obsolescence.

Yes, it's weird in its special way. Especially the part you mentioned in which they are pairing a phone. Rest assured, this is the most humane invasive lab work on an animal I have ever seen. If you want nightmares check out how they test cochlear implants for example ... No, really, don't do that.

For me the strangest part of watching this is not the lab work or the animal, but the implications of this technology and where it might lead to in the future once applied to humans. I really like the idea and premise presented, but let's be honest: There are also way too many evil use cases ...

Yeah, it's not that hard to imagine the next bit where there's a captive warehouse of monkeys solving CAPTCHAS for food or something. It's a depressing video to be sure.

Right now, this is being done with people from India, so ... I'm not sure if your dystopia vision is a step up or down on the civilisation ladder.

By "captive", I meant they didn't choose to do so, and didn't get to go home after their shift, and weren't compensated outside of food.

I'd encourage anyone worried about the ethics of animal experiments in research like this to come up with prioritisation metrics over a number of animal welfare cause areas as an exercise. Surely a small number of well-cared for animals in high-upside experiments are lower on the list than say, pigs for pork sausages, which have worse welfare, smaller upsides and are more cost efficient to campaign against.

Would you respond the same way to ethical concerns about the treatment of a small number of humans? If we could find larger groups of humans facing worse treatment (and surely we could), you might conclude that the smaller, less-poorly treated group is simply not worth discussing. But I'd argue in that case (as I would in this one) that the conclusion is mistaken.

Also, if I may ask, do you think we have to choose between caring about these individuals and caring about those in factory farms? I personally care about both (as well as plenty of other ethical issues, naturally). And I imagine other people feel the same. So your suggestion that we measure the two causes against each other is rather confusing.

You only have a finite amount of time and resources, the way in which you choose to spend them can have enormous influence on the resulting impact. Given this, you have to prioritize. If you are not prioritizing, then somebody or something is doing it for you.

I don't think these monkeys are "not worth discussing" - but I don't think they merit more than a cursory evaluation. The same is true for say, people that are killed by falling coconuts.

There are, in all likelihood, cost-effective causes that you can donate to, in terms of your dollars, skills and capcity to give a damn. I'd suggest these monkeys shouldn't make the list.

Interesting. And you think these monkeys shouldn't make anyone's list? (I'm inferring, given that you began by addressing "anyone" with ethical concerns.)

I'd understand if you said they didn't make the cut for you personally, but I'm not sure why you'd be invested in ruling them out for everyone.

(And this a sidebar, but I think one could quite reasonably believe that advocacy here is worthwhile in exactly the bang-per-buck utilitarian sense you're invoking. For instance, people who are galvanized on behalf on these monkeys might then change their actions towards less visible, less relatable nonhumans like pigs, chickens, and fishes.)

Pretty much everyone - if you work for neuralink there's some chance of cheap interventions here, but outside of that it seems unlikely.

The latter point (galvanised to support other welfare causes) is roughly what inspired the original post - our intuitive emotional reactions to visible harm should indeed encourage effort in investing in effective harm reduction advocacy. But if seeing these monkeys makes you sad, you really ought to think about all the less-visible, cheap to attack welfare issues that are available.

( I avoided it in the headline comment so as not to seem to shill, but if you're interested in what cause areas come out ahead, I broadly endorse, but am not affiliated with, the analysis here: https://animalcharityevaluators.org/charity-reviews/all-char... )

Thanks for sharing! It's a good resource and not one I've come across before.

So I do very much appreciate the spirit of your comment: that we should attend to less visible (and more easily addressed) harms, and not get caught up in 'celebrity causes,' so to speak.

But I think where I differ from you is that I don't view it as an either-or paradigm. I'd say to people "Go ahead and try to help these monkey individuals, and also work for, e.g., food-farmed nonhumans (who are easier to help, etc.)"

I appreciate the risk you're citing, that people could effectively "waste" time and resources on a case like this. But I'd argue there's a greater risk in approaching ethics as A) zero sum, and B) generalizable. I'll elaborate:

A) It's certainly true that we only have so many minutes in the day and so many dollars in our wallet. But I think these zero-sum resources are often not the final limit on what we can do. Rather, the limits we reach are emotional and psychological energy - which is often not zero-sum. Getting engaged in an issue (especially when it's an issue that radicalizes you) can actually increase the amount of time and resources you find for other issues. (I.e., you reclaim it from less important stuff.)

B) I'd argue ethics is patently not generalizable (in the sense that you're suggesting - i.e. that everyone should reach the same conclusion about which cases are worth effort), simply because humans are so varied. One person might have tons of money and be happy to spend on this cause in addition to whatever they give to help farmed nonhumans. Another person might feel a special bond with monkeys that makes this an easy, non-taxing (or even net-energy-positive) issue to engage in. Yet another person might currently find the plight of food-farmed nonhumans overwhelming to consider, but these monkeys will be a stepping-stone issue that help them get there. Etc.

Bluetooth (BT) is simply a wireless standard for exchanging data between devices over short distances. It is very widespread, reliable enough and easy to develop and integrate. It's an understandable and common choice in the medical devices field.

In a more developed, market society, you typically have several commoditized, replaceable products and most tend to leverage BT. The idea of replaceableness belongs to the product itself, but what we see in common for all of them is BT connectivity, so we tend to perceive BT as an identifier of replaceableness as well.

Neuralink very casually and tongue-in-cheek BT "paired" to the monkey, reinforcing the idea of replaceability and commoditization of the monkey.

Many people have empathy for animals and monkeys are seen as precursors to humans. It's very easy to see how treating a pre-human species like a replaceable product, leaves humans themselves creeped out and feeling like in the future they may also be treated like a replaceable product.

Neuralink needs an ethicist in a high exec position or board and a better PR manager. They should have done some of these to reduce that perception mess: - allude to the use of BT in serious medical applications before "pairing" with the monkey - reduce or change the common BT terms like pairing - use a computer rather than a smartphone - don't mention BT, just say wirelessly connected - don't use a Bond-villain smooth British voice

By talking about it as a PR problem, I don't mean to say that is the root cause, it is just what we can see at the surface.

Do they really care about ethics? Is it execs, engineers, video directors, everyone, no one? Do they care, but are just bad at PR? Do they not care, and this video is a reflection of that? This is what matters, for all those people to genuinely care about ethics.

Can they still genuinely care though, after becoming a corporation? Can bringing shareholder value align with ethics?

One botched video is simply one data point in the public trying to understand Neuralink's genuine stance on ethics.

Only people with close contact to Neuralink will really know. We are all hyper connected, but sadly, only superficially.

This is how progress is made. Similarly, people in DaVinci's times were sickened by his autopsies but that doesn't mean there was something wrong about them.

To counter with an extreme example: plenty of medical progress was made via Nazi and Japanese experiments on live subjects during the second world war. So no, categorically the ends does not always justify the means.

>plenty of medical progress was made via Nazi and Japanese experiments on live subjects during the second world war

[citation needed]

Plenty of unethical experiments. What was learnt? What progress was made?

I think it's 2 things:

1) The deliberately soothing British voice does not come off as soothing. It comes off as insidious, and threatening. This is in part influenced by our own cultural context with media like Black Mirror, but the effect is there nonetheless.

2) The comparison points to "pair with your iPhone" feel WILDLY misaligned with the rest of the message. The premise of Neuralink is that this is a world-changing cutting edge technology for the good of humanity. Then all of a sudden you have a situation where a living sentient being is paired to an iPhone like some sort of bluetooth speaker. It reeks of confused ethics.

> Then all of a sudden you have a situation where a living sentient being is paired to an iPhone like some sort of bluetooth speaker.

I think you put your finger on why that part was unsettling.

The ends justify the means. You're benefiting from ages of actually cruel animal experiments, they got us to where we are today.

Nobody loves them, but they're important in this case to help parralyzed humans maybe with ALS or some other disease communicate with loved ones

The assumption being that human lives are far more valuable than non-human lives.

Where it gets funky though is trying to quantify that to an extent: how many monkey disfigurements is worth fixing one human disfigurement? Ten? A thousand?

How many chimps would you blind in order to prevent humans being blinded by the latest lash extension cosmetic?

> The assumption being that human lives are far more valuable than non-human lives

Is it even controversial that yes, of course they are?

I mean, Boeing and the FAA kept 737 MAX's in the air after brown people died in the first crash when we all know they'd have grounded them if it was a crash in Kansas. We value human lives differently let alone animals.

> How many chimps would you blind in order to prevent humans being blinded by the latest lash extension cosmetic?

None. But I'd blind as many as are needed to trial human eye transplants, for example.


It's not hard to imagine them implanting Neuralink-type devices in humans to make Minority Report a reality. And that's just read-mode. With write-mode, I bet they could easily figure out a way to make wrong thoughts physically impossible to think.

It's only a matter of time before certain governments start alpha-testing this type of technology in places like Xinjiang or Myanmar.

If they haven't been already for 10-15 years.

Where is the Bluetooth pairing? I don’t see it. I’m watching the video at the top of the page... is it one of the other videos?

24 seconds in

The question is:

Who wouldn't want to ultimately be assimilated by the Borg?

It’s like social media and looking in the mirror!

you either test on animals, or you test on humans. simple as that.

On the flip side, what if this comes to pass?

First @Neuralink product will enable someone with paralysis to use a smartphone with their mind faster than someone using thumbs

Later versions will be able to shunt signals from Neuralinks in brain to Neuralinks in body motor/sensory neuron clusters, thus enabling, for example, paraplegics to walk again



Please omit snark and name-calling from your comments to HN. Your comment would be fine without the swipe at the end.


Well I can't edit it to remove it.

I initially thought GP was being overly hyperbolic, but after watching the video I can't say I fully disagree. It's not so much the animal studies aspect, but rather the fact that they used bluetooth via a mobile phone to connect to the monkey.

That aspect of connecting via bluetooth from a phone is most conventionally used to interact with replaceable commodities such as wireless speakers/headphones, but here it's being used to interact with a _live_ monkey. This framing somewhat gives the impression that this living being has been reduced to the status of a replaceable commodity, a mere peripheral that one might connect to via bluetooth.

I agree that given how much modern society relies on animal testing that it's not really a rational response - maybe this reaction could have been mitigated somewhat if they had connected to the monkey via a computer or more involved process rather than simply just the conventional bluetooth pairing flow on a phone.

> That aspect of connecting via bluetooth from a phone is most conventionally used to interact with replaceable commodities such as wireless speakers/headphones, but here it's being used to interact with a _live_ monkey.

I understand that, but I think to some degree, you're letting somewhat unrelated things influence your opinion of this.

You know what else it interacted with and controlled wirelessly? Pacemakers. Nobody thinks of those people are replaceable commodities.

Also, what about those that don't live a life of luxury and have access to lots of commoditized devices that wirelessly pair through a cellphone? People in less affluent countries might be less jaded about controlling something wirelessly and still view that as an amazing new technology associated with things they can rarely afford.

So, to what degree are those associations useful and accurate, and to what degree are they you bringing unrelated prior biases to bear?

I‘m working for a company which makes apps connecting to insulin pens and pumps via Bluetooth. Once you get into the mindset of connecting medical devices to your phone, this feeling of cheap commodity entirely disappears. Bluetooth and phones can be awesome tools, and if implemented with enough verification and validation, are sufficiently safe.

I think you articulated it well. It trivializes life. Countless animals die horrifically for science, but much of it is ultimately in the pursuit of valuing human life. This is something different. This animal has been reduced to a device accessory.

After watching macro trends in U.S. politics and tech for a while, I've held (but not articulated much) a sense that power is shifting from federal government to private interests. Previously that would have meant corporate boards, but I think increasingly it's a small number of individuals (due to realpolitik).

The good thing about messy, human models of transactions and interaction is that it can take a long time and many different voices can be heard, allowing disuptes to occur and be resolved.

Many of these successful tech corporates work to eliminate the human discussion element, and replace it with digitized (and frequently proprietary, or at least gatekept) rules.

I think I've dealt with a few difficult dominant personality types in the past, and it would not surprise me at all to see them consider humans-as-pets as a desirable future. Match that with digitized 'asset ownership' and other non-repudiable mechanisms and there could be a very dystopian and authoritarian future in the mind of some of these people.

Now I'll make sure to sound like a complete nutter (as if I hadn't already) and mention that some of these individuals and companies are now so essential to the U.S., both domestically and internationally, that they are becoming untouchable.

Meanwhile our own tech industry is busy debating and trying to determine what the future of libre software will look like. It's a pivotal moment and I'm optimistic we'll figure it out to everyone's benefit, but there is a lot at stake.

Bluetooth is just a short-range wireless technology. People use bluetooth to control their pacemakers, is there an issue with that too?

For me, I find it unsettling because it's a reminder of how ultimately we're all just compositions of atoms with electric signals in our body. While I find this immensely exciting, it unnervingly reminds me of my mortality. It's akin to what I imagine Neo would've felt when he learnt about the Matrix. I'd like to think I'd have without a doubt taken the red pill too (like I want this kind of research to be more successful) but anything that makes the Matrix more real is unsettling.

I'm not buying the idea that the animals somehow benefit from this. I'm not necessarily against animal testing either. But it's one thing to test on animals to develop lifesaving cures and another thing entirely sure to experiment with brain implants with the potential to take away a living creature's free will. Or the ability to intrude on thought. The whole thing is pretty gross.

It's great that this tech could help paralyzed people and amputees, but is it worth the cost?

To the paralyzed and the amputees, absolutely. The faster the advancement happens, the sooner animals models can be deprecated. Nobody wants to use animal models, they are a necessary evil.

Edit: This is a moot point if it turns out we can repair the damage from paralysis and amputation using bioengineering.

Absolutely worth it. It's not ideal, but how else could we advance such technology that could help the paralyzed?

Nothing says feel'n good and free will like banana smoothies and forced brain implants.

> As they said in the video, the animal is doing the activity of it of it's own free will.

Why don't they do it on humans then?

It's legally prohibited until a certain amount of animal testing has been performed. They're aiming to do the first human implants within a year.

2030 is going to be a dystopia

> It's legally prohibited until a certain amount of animal testing has been performed.


You have to conduct testing to show that the device is safe and effective before you can begin marketing it.


If you've never gone through the process of bringing a medical device to market, well, I'm not sure there's an easy FAQ that will answer your scepticism. But the basic idea is that you have to do a bunch of testing in the lab, then probably a bunch in animals, then finally a bunch in human clinical trials, then you can actually market the device to the public.

Neuralink is in the 'animal testing' phase, and it sounds like they're likely to start human testing soon.

Note that once they've gotten a device approved for humans, that doesn't mean they will stop testing on animals. There will likely be improvements to the device, new protocols, etc. that will necessitate continued testing as new features are brought to market.

Nice link, thanks.

Relevant section:

> Among the types of evidence that may be required, when appropriate, to determine that there is reasonable assurance that a device is safe are investigations using laboratory animals, investigations involving human subjects, and nonclinical investigations including in vitro studies.

Here is the ultimate question....

What smoothing algorithms did they use to guess the 'certainty' of the 'intended' movement in a Y-Plane game vs. the X,Y grid they learned from?

Having researched in this space, you don't get better results from 'training' on a X,Y space and reducing that training to a Y-grid predictor.

There is a ton of smoothing going on in the video... or the metal conductor plays a huge role in the electrical signals they get from the implanted electrodes. I once blew up a demo because of something like this metal stick as a constant I didn't think to consider.

> or the metal conductor plays a huge role in the electrical signals they get from the implanted electrodes

I highly doubt that, you must be talking about non-invasive electrodes such as EEG. When the electrodes are inside the brain and thus in the cranium, they are effectively protected from outside EM activity since they're in an effective Faraday cage, so your signal has much higher fidelity.

>When the electrodes are inside the brain and thus in the cranium, they are effectively protected from outside EM activity since they're in an effective Faraday cage, so your signal has much higher fidelity.

Pretty much this on a different and somehow related note I would still try to check if the algorithm does not just trick us into thinking that it is doing what the monkey wants it would be nice if they post some more in depth material of how the code or whatever they use looks at the end just to be sure we are not in front of a half machine clever hans (https://en.wikipedia.org/wiki/Clever_Hans)

TO clarify: we dont have the ability to know if the monkey made an error he wont say hey I didnt want to get that one right or win that pong movement, have your milkshake back please. And we kind of need to know that in fact when doing mathematical modeling and AI having a way to quantify that is a requirement and one of the first things they teach you about.

True, the post-processing algorithm could just be doing a randomish walk towards the answer, but I highly doubt that is what's happening. 2D cursor movement is well understood to be within the realm of possibility for EEG, so I imagine it is actually much easier with an in-vivo device. From what I understand, this is entirely supervised learning. The monkey is "telling" the device what neural pattern corresponds to moving the cursor. If I had to guess, then this chip is planted extremely close to the hippocampus, which is known to have cells that encode spatial information. The algorithm is probably doing a basic supervised learning on the cells that fire for a particular location of the cursor.

I'm just guessing on this, but the "training" session probably involved having the monkey stare at the screen for a while, while the cursor moved, which allowed the device to capture a spatial heatmap of which cells fired at which locations. There's probably some online optimization happening as the monkey then continues the "training" process by completing the task. Overall, this task is completely doable with the current technology, so I would not assume any foul play here. If the monkey was writing Shakespeare, I would definitely doubt it.

the chip is implanted in the motor cortex, I believe, so the training consisted of rewarding the monkey for moving the cursor to a target.

After the monkey has played the game a bunch, you can correlate the neuron firings generating motor signals to the arm with the motion of the cursor on the screen. Later, the monkey's arm need not actually move the joystick.

The hippocampus is pretty deep in the brain, and the Neuralink surgery can't go deeper than a centimeter or two.

How would it know which direction to move in if the position of the goal was not part of the input to whatever control algorithm they're using? The part of the code that detects the overlap of the cursor and the goal should be completely separate from the part that is translating the implant inputs into x,y movement so I don't see how it could be driven by anything other than input from the monkey.

> I highly doubt that, you must be talking about non-invasive electrodes such as EEG. When the electrodes are inside the brain and thus in the cranium, they are effectively protected from outside EM activity since they're in an effective Faraday cage, so your signal has much higher fidelity.

Yeah, I am speaking through non-invasive experience, however, unless your grounding is LIGO level sensitive, sticking a 3 ft metal pole into your mouth takes some significant processing to get a good ground, especially if it comes into contact with the earth and then you change your connectivity with opening and closing your mouth.

Again, low probability, but that just looks like it's asking for problems.

The skull works as a Faraday cage in such situations?

Nowhere near a perfect conductor. It, and the muscles around it, create noise and weaken the internal signal measured externally, acting as an effective cage. Measuring the signal from outside the skull is a bad idea. From inside (e.g. hippocampus), the electrodes are so much closer to the neurons, that the majority of the measured signal is not due to cranial noise.

I don't understand how you can gather such a targeted signal by sticking wires in just a few places.

Like, you can't just tape a metal coat hanger to an iPhone and start getting text messages through it.

But somehow you can put some wires just as few mm into the surface of a brain and extract fine motor signals??

The most interesting thing here is not the mind reading, it is the miniaturization of computing and sparse data transfer. Neuralink off-loaded the computation to the sensor, sending back aggregate data snapshots. Usually you will need massive ASIC/FPGA computation banks with a lot of messy to handle the data transfer and readout but they took a shortcut bye sending only counts of spikes that exceeds a certain threshold.

There is nothing stopping you from using a cloth hangar. You just need to make sure it is impedance matched and properly sanitized for subdermal use. The signal will be much worse as a cloth hangar had gauge/diameter that is several magnitudes bigger than that of a neuron. Nothing here is particularly "novel" that it cannot be trivially replicated if you have the budget of one of the FAANGs. If you have 100million in capital, you can easily find poor neuroscience and EE PhDs who will be willing to work for you for <80K a year just for the opportunity to build something like this. The reason it is difficult and out of reach for smaller companies is because small amplifiers and electronic packages that can be subdermally implanted requires significant manufacturing capital as they fall under semiconductors and biomedical devices. Once the industry settle on a design expect the price to drop dramatically as the core chips and parts become commoditized.

Is there anyway to get this signal non invasively (magnetic field detection or photonically)? Or are implanted electrodes table stakes for a neural interface?

Not really, that we know of. To appreciate the complexity you could read this post[0] from Wait But Why? Your brain has approximately 40,000 neurons in the space of a cubic millimeter. Each saying something completely different. It’s incomprehensible noise unless you start dividing it up like Neuralink has. They get a few thousand probes in the size of a penny, which is a far greater level of resolution than you could probably ever expect from a wireless solution.

[0] https://waitbutwhy.com/2017/04/neuralink.html

Do you think something like fMRI would have enough resolution? Can’t pick up activity directly of course, but perhaps by picking up fluid flow differentials?

As mentioned by a sibling comment, fMRI have a sufficient spatial resolution (i.e. you can watch the activity of a few number of neuron at a time instead of grouping them by batch of hhundreds of them), but the temporal resolution is low (order of magnitude is seconds; a single action potential is 5ms).

Here's a graph that summarize the temporal/spatial tradeoff of the different BMI tech: https://www.researchgate.net/figure/The-various-neural-recor...

Noninvasive signal modality directly from the cortex are certainly possible (e.g. Ultrasound, fNIR). But they lack the spatial and temporal resolution of implanted electrodes. Intracortical brain signals are the most well understood and mature in terms of instrumentation as well. So in the short term, implanted electrodes offer the highest capacity for a BCI.

*messy cables

*a shortcut by

My understanding is that neurons are like a higher-order connected mesh. A bit like many fully-connected layers in a Machine Learning model.

You can't just read fine motor signals from any neuron, but given enough neurons, since they're deeply connected, you can infer a close approximation of the intended motor signals given a known output. That's why they need a lot of sampling wires in the brain; to makes sure they have enough data to make that inference. Luckily, machine-learning is very good at figuring out patterns in this sort of data.

My guess is that they train an ML model using the inputs and outputs when the joystick is connected. Then they can use the trained model to just generate the output directly from the inputs when the joystick is disconnected.

This will work fine for one particular monkeybrain/interface combo; however, you cannot use the same model on other monkeys. You have to train a new one each time.

> Like, you can't just tape a metal coat hanger to an iPhone and start getting text messages through it.

You could if we didn't encrypt connections.

On that note, does anyone know how to upgrade libssl across my entire nervous system?

Or have Bluetooth disabled by default for my brain.

The game examples were super fascinating, but since I was already essentially familiar with the fundamental principles and technology, it didn't quite impact me as much as the researcher pairing his iPhone with a Bluetooth transceiver in the monkey's brain. That sent weird shivers down my spine.

> On that note, does anyone know how to upgrade libssl across my entire nervous system?

‘Inception’ shows someone who tries to do something like that, but there seems to be a vulnerability.

This reminds me of dropout where the neural network won’t be dependent on a specific connection. Instead, it will be statistical which will actually be a good fit for this approach.


Funny you mention meshes. I’ve always thought that self-organizing maps are a good model for neurons, more so than deep neural networks.

The location of the electrodes aren't arbitrary. The region of implantation is likely the hand-motor cortex (which has thickness on the order of mm), the area associated with arm movements (check out somatotopy), so the signals acquired are actually targeted for the demo task. This relationship has been known in neuroscience for almost a century and has been validated in decades of brain-machine interface experiments.

They are also recording from up to 1000 channels, which is probably overkill for mind pong, tbh. But you'll need that many implanted electrodes to study long term electrode biocompatibility.

Even outside of primary motor/somatosensory areas, 2D cursor control is a pretty standard BMI paradigm.

Not exactly “Hello, World” but maybe “ToDo List App”.

Yeah their implant hardware is really impressive but "mind pong" is a pretty common starter project for at home EEG hacking, let alone legit BCI research. The internet is littered with examples of mind pong for the Muse like this one: https://medium.com/@nayvelt.lina/playing-ping-pong-with-my-b...

Granted it moves slower than actual Atari Pong, and takes a few minutes to get the hang of controlling the paddle. But IME it isn't much harder to control the up/down of the paddle with the EEG than it is to play a game like flappy bird, and that's using a couple random scalp electrodes from a consumer device.

So yeah thanks for bringing this up, because I feel some of the comments here are acting like this is much closer to "mind reading" than it actually is. Not that it isn't cool, but the overhyping kinda kills it for me lol. Is this how robotics people feel when Boston Dynamics releases a new video?

I'd argue it is the "Hello, World" in any BMI lab. Assuming existing implants and electronics, and non-naive monkey, advanced undergrads should get it in a semester.

That's why I went with To Do list :-)

It's not totally trivial, but a reasonably skilled person should be able to get a not-too-janky version working with a bit of effort. (Also, where are you that undergrads get to interact with monkeys?!)

I think the source of wonder here is that neural networks are very un-CPU like, in the level of correlation between the units. I'm not sure you could probe a CPU in a similar manner and be able to recover meaningful information. A binary 101X10 are entirely different numbers depending on what X is, for example, and you woulnd't necessarily expect X to correlate with the place values elsehwere (e.g. 1st and 5th bit). Similar arguments, apply, to probing different components, such as registers or instruction memory.

Funny you should compare understanding biological neural networks to a microprocessor. Check out this paper (https://journals.plos.org/ploscompbiol/article?id=10.1371/jo...), which systematically addresses how applying traditional neuroscience methods to analyzing a microprocessor would get you.

I've read that paper years ago. I'm afraid I find it unimpressive. It is not so hard, really, to fail, when you fail to try. I have a computer science degree; I've built CPUs (from all the sequential and combinational circuits on up); I also work in neuroscience, and do some of the kinds of analysis this paper tries to criticize.

For example, a very basic approach is design experimental tasks that have contrasting conditions, or events to be predicted. They don't do that. Even something as simple as pressing the controller left vs right vs not moving at all in donkeykong would have been more interesting.

In any case, brains and traditional computer circuits work on different principles as I noted in my OP. Furthermore, there is quite a lot more redundancy in brains--and partial-redundancy is a bit of how it works; for example, if you have a population of neurons with different tuning curves, then the output in response to a stimulus can be integrated, and a likelihood distribution obtained for the actual value of a stimulus; that's just fundamentally different than how, say, a 5-stage pipelined RISC cpu works.

>Like, you can't just tape a metal coat hanger to an iPhone and start getting text messages through it.

You can get a 128 key out of most electronic devices in a reasonable amount of time: https://en.wikipedia.org/wiki/Electromagnetic_attack#Known_a...

MI5 were lifting crypto keys based on keypress acoustics in the 50s, I wouldn't be so certain. If you put the phone up an expensive enough tool you can at very least find side channels

I cannot imagine any future in which this is not terribly abused or otherwise going wrong.

People invading other people's minds, Borg-like clusters of minds outsmarting anyone who is not connected with others, forcing others to play along to stay competitive. Maybe entities that even never grew up as unaugmented, unconnected persons because they were already linked in young age - one mind with different bodies.

It is absurd to try to protect humankind against a perceived existential threat by AGI by developing a technology that would also destroy humanity if it works as advertised. Add to all that the cases in which it doesn't work properly. Developing inflammation around the implants, transmitter breaking, batteries running out, losing wireless connection, ... if you think dead zones while hiking are infuriating then wait until your cognitive capabilities depend on a network connection.

I am afraid that all the good that SpaceX or Tesla have achieved so far will be undone by the damage that NeuraLink's technology will do to humankind one day.

This is honest-to-god magic. Some cyberpunk, neuromancer, hardcore reality-bending shit. The brain is the final frontier, if technology gets a foothold in there things are going to get WILD.

Also, as this stuff gets further along, there's going to be large incentives for unethical testing. Rockets can be tested with unmanned flights, but you can't really dry run brain implants.

This isn't even the start of it. Everything in this video could be done with a non-invasive EEG, and has been possible for decades. This is a teaser video. The real magic of Neuralink has not been witnessed yet.

Are you thinking bootleg neuralinks from the local ripper doc?

Well, that’s what animals are for honestly. They didn’t volunteer but we’ve deemed it “humane” nonetheless for our own moral comfort.

This particular exercise could've been done with EEG. Where Neuralink excels is in having a much higher possible bitrate.

Pong is the absolute POC. Neuralink will most likely be able to achieve higher differentiability and thus be able to complete tasks of higher complexity, so stay tuned for even more impressive tasks. EEG can't get past pong; it's limited by physics itself. The solution is to put the electrodes in the brain, and there is likely no other solution.

Looking forward to the future.

When you say "this particular exercise", are you referring to the "put the pointer in the box" exercise at the beginning? The Pong part toward the end was pretty cool. I wonder how complex of games the monkey can play :)

I believe pong has less entropy than the initial task of putting the cursor into the box. Pong is a 1D game with one degree of freedom, where you are effectively moving the cursor along the Y-axis. The previous task involved moving the cursor on the XY-plane, which involves two degrees of freedom.

Therefore, pong is less impressive than the previous task.

This is difficult to say because pong involves moving components while the box stays still. In order to play pong the monkey must first estimate where the ball will go and then give the command to move the cursor. This indicates that Neuralink may be able to handle complex multitasking, i.e. you may be able to walk or drive and the system will still be able to decode you correctly.

That's a good point, it is nuanced for sure. My line of reasoning is that ultimately the monkey only has to control one dimension of movement of some object, so whatever complex reasoning is required to arrive at the decision to move up or down (one dimensional movement), it would still be easier for a supervised learning algorithm to learn spatial-encoding neurons over a single dimension.

I think this gets deep into the difference between premeditation and action and what that means at a neural level. I would guess that Neuralink is not processing the "logical thinking" as much as it is on the neurons that dumbly encode an association with ping-pong racket location.

As long as some set of neurons uniquely fire approximately when the pong moves along its degree of freedom, different electrodes will pick up different voltages, and that can be learned by an algorithm. So we don't need to know how the monkey arrived at a decision, but rather just be able to index the possible outcomes of the decision. Which is why I thought that the entropy of the decision space mattered. It's easier to learn a smaller space.

I would tend to think this is a dumb pattern matching cluster (e.g. see visual encoding[0]) that's being picked up by Neuralink, and not the actual "logical decision making". Although, there may be more subtle neurons that fire in a sort of "premonition" which have learned the correlation of the racket position with the ball over time.

I can link to more papers on that -- it's not just visual encoding, but rats can "plan" where they are going to go, a sort of premonition. Humans also do this. It's like an echo that happens before you yell. That's what the brain does when you plan to go to the grocery store.

This is all really interesting stuff, but I'm not an expert!

[0] https://www.youtube.com/watch?v=jw6nBWo21Zk

If you watch closely, you can see the monkey is still "playing" pong with his right hand/paw. (He's pushing and pulling on the 'smoothie straw' in sync with the brain signals / on screen actions)

But can it run Doom?

The next step for Neuralink is write mode. No, not writing memories, but reprogramming motor function. If you can capture the signals of limb movement synapses for say, a paraplegic, you can read those signals and convert them into write signals in another part of the brain responsible for the actual motor function. Over time, that neuron pathway gets reinforced and could potentially teach a person how to walk again.

It’s exciting stuff and I’m sure there are concerns to be addressed down the line wrt altering perception, but let’s not put the cart before the horse. That would be like worrying about impending AI drones in the 70s.

MIT "wrote" memories (the solution to a maze) into mice/rats using photons[0].

Neuralink in "write" mode could be engineered to do the same thing by introducing an EM current that changes over time. In the past, I've worked with "read-only", passive tetrodes, but the opposite is possible from a physics standpoint. Moving charges is just another way of saying "propagating photons". I think there's not only precedent, but there's proof that this can be done.

The hard part is figuring out the biology (for a given memory, you still need to calculate the correct I(t) that actually implants it), which is a relatively unexplored domain. Neuralink is centuries away from using "write" mode to turn everyone into Neo, but much simpler use-cases can still make a huge impact within the next couple decades.

[0] https://www.theguardian.com/science/2013/jul/25/false-memory...

>you can read those signals and convert them into write signals in another part of the brain responsible for the actual motor function.

The signals Neuralink is reading are from the "part of the brain responsible for the actual motor function". I'm not sure what you're suggesting here.

I think you may be confused on how the majority of paraplegic and NDG conditions work - it's not that the brain isn't producing the signals, it's that the signals aren't reaching the muscles they need to to create motor function - it's the link that's broken, for a variety of reasons.

Of course, the obvious corollary here is reading the signals and using electrodes implanted in the muscles to directly stimulate a contraction

Forgive my ignorance, but after writing to neurons to re-learn walking, will that render that particular Neuralink chip unless? Since the neuron structures surrounding that chip will be constructed for walking, you won't want to further disrupt/write to them. Then you would need to move the chip for any further use?

Edit: After reading another comment about the link being broken (hence the name Neuralink), this makes more sense now. Although, I still think that maybe you only get one usage out of one chip? I suppose the definition of "usage" is what is important. Presumably you won't need a chip per joint in a leg, and presumably one chip could control two legs? Or perhaps you get finer control with a chip per leg?

I was about to say that the Link (Neuralink) is already bidirectional, so they should already be able to spike each electron as well as read from it. But I had troubles finding any sources supporting this, so I'm not so sure anymore. I know that Elon talked about this in the early presentations though.

The implant definitely can write, but it's more power intensive than reading, and lower priority.

I know there's a myriad of reasons why people lose the ability to walk, but are there any in which it's a matter of what sounds like memory degradation(i.e. you need to remind a particular set of neurons what they previously "knew")?

Total non-expert here and legitimately curious.

Sometimes people with severe brain damage in the motor section can lose the ability to walk, yes, but that can also be solved with traditional physical therapy (brains are amazingly adaptive things, and can even repurpose other sections into motor function).

Butlike I said, traditional PT is a much less invasive solution to that, so I have no idea what GP is talking about.

> The next step for Neuralink is write mode. No, not writing memories, but reprogramming motor function. If you can capture the signals of limb movement synapses for say, a paraplegic, you can read those signals and convert them into write signals in another part of the brain responsible for the actual motor function. Over time, that neuron pathway gets reinforced and could potentially teach a person how to walk again.

Or - hear me out - for a small bonus, you could volunteer for a second shift in the Amazon warehouse while you sleep. An algorithm right bits to drive your body as you shuffle unevenly from station to station in the dark (no need for lights, the computer knows where things are). They can call it the Neuralink Autopilot.

Edit: would it be possible to externally trigger firing of motor neurons in someone who is brain-dead? The future might be a little Ancillary Justice-y, but with a strong "posthumous work contract" capitalist flavor to it.

Super cringy banana quip.

Also, as this comment suggests, it seems to be more of an advance in wireless data transfer than neuroscience: https://news.ycombinator.com/item?id=26746553.

I hope I'm wrong, but Muskuito's misleading marketing is never far on the horizon.

I'm convinced that this type of technology + some advancements in kinetic technology will allow for artificial telekinesis in the future. I don't know if it will be in my lifetime, but I would love to control it.

It could be billed based on the energy required to translate the objects.

Well thank god it can be billed!

More seriously, I actually yearn for the days of Star Trek and where we look at how innovations like this can make people's lives better first and then just stop there.

I was really excited to see the first goal they mentioned:

>Our first goal is to give people with paralysis their digital freedom back: to communicate more easily via text, to follow their curiosity on the web, to express their creativity through photography and art, and, yes, to play video games.

Modes of interaction in society are so poorly designed for anyone whos sensory functions are different than the norm. Bridging directly from the brain to interactions that currently require voice (normed to English Speaking White Males) or touch or movement or vision is a game changer not just for those who are paralyzed but many different groups for whom technology is not designed and is instead adapted.

> It could be billed based on the energy required to translate the objects.

It could be, but it will be billed with ads inserted directly in to your brain that you can’t skip or turn off - even with your eyes closed.

In Soviet Russia, it (neuralink) loves to control you!

Oh wait...

Anyone else feel a strange kinship with the monkey?

Absolutely. He is one of the first few through the breach for us with this tech.

Neuroscientists have been doing BCI with monkeys for years. Take a look at the work coming out of Byron Yu's and Steve Chase's labs at CMU for starters

I guess I just meant for this specific nerolink tech. He is definitely joined by a massive group of our distant cousins that have been involuntarily drafted into service for us. It's brutal, but so is life I guess. Hopefully this particular one doesn't mind it too much.

Tangent: Dating myself here, but this demo just triggered a childhood memory of the movie Project X (1987, starring Mathew Broderick).



Two years ago without a brain implant.

This has been possible for decades without an implant.

The problem is that without an implant, only this is possible. Anything significantly more complex than moving a cursor on a screen is just not possible or feasible without an implant, because of physical (electromagnetic) limitations.

Its projects like these that makes me wonder that what the actual heck am I doing with my life.

Same feelings, I started to think about dropping current ordinary app dev job and humbly get back to science. In 20 years world is going to be a completely different place.

Heh same, when I create a nice group policy I feel smart. Imagine hacking a monkey's brain.

I thought one of the biggest problems with brain-computer interfaces is our bodies creating scarring tissue around electrodes[1]. Have they solved that?

[1]: https://www.businessinsider.com/huge-problem-with-brain-impl...

"What better reward for a monkey, than a banana"? what about FREEDOM?

I am not a vegan/whatever/animal activist, but maybe we should allow terminally-ill or death-row volunteers for this kind of experiment... With good money for the family, and I think some people would like to help mankind in the end. It is better than dying without accomplishing anything in life.

So, you have at most one year with an inoperable tumor, and you will not travel because you want to be close to your family. Would you be a volunteer for this research, for 500 thousand dollars? (monkeys are cheaper, but you are the real deal)

You said you’re not a vegan, so to be clear your okay with factory farms crushing pigs to death but minimally invasive surgery on monkeys so they can play pong is too far? I don’t care about your stance on animal consumption particularly, but don’t you find that a little hypocritical?

Yep, I agree this sounds hypocritical,but everybody has different "lines" to cross. I guess mine is how close they are to humans. (98%), and I also differentiate "food" from "experiment".

They crush pigs to death? What? Nothing would surprise me at this point, but I couldnt find a citation for that at first glance.

OP might be referring to pigs crushing each other to death because of overcrowding in badly-run farms and during transportation. Farmers don't throw pigs in trash compactors (and why would they? Non-compacted pigs is how they do earn their money.)

Patients undergoing brain surgeries are routinely asked to volunteer for neuroscientific studies and they often do so. Terminally ill patients also donate their brain tissue for tons of studies more often than you would think. But the scope of experiments that one can do with them and the available multimodal data available from parallel studies in human cortex is extremely limited.

On the other hand, non-human primates are readily available in a controlled environment, can be used to generate huge amounts of data specific to a diverse set of tasks and experiments that you just can not do on the human patients.

The brain is a complex machine that is extremely hard to decipher just by studying a single species or one kind of experiments. The need for studies across species with a hundred different ways of collecting and analysing data is essential to first understand the mechnasims of the brain and only then (hopefully) be able to causally alter it for specific applications in medicine and so on.

Yeah I'd volunteer for implant testing if I had a 6 month life expectancy due to a terminal illness.

I don't think it's a good idea to allow death-row inmates to do this though; They'd likely be pressured into volunteering as the only way to delay the execution.

This sounds like the plot of a movie: death-row inmate forced to volunteer for implant testing, but he manages to connect to some kind of military robot and fights for his freedom.

Death Rowbot (2022)

I did not thought about this, but at the same time, "punishment" cannot coexist with "reparation"? Money would go to victims.

Our brain interface tech is so primitive, but it holds incredible promise if we're able to solve it.

Imagine being able to virtualize consciousness, back up memories, live forever...

The problem is how invasive it is, and how little reward there is initially. We have to solve the chicken and egg problem, but unfortunately cracking open the skull and sticking electrodes into the brain is not something one typically wants to do if they're conscious about their health.

I hope we can gain traction with medical applications and then begin to make steady advances. It'd be neat to have fully virtual, synthetic senses before we die.

Digitizing consciousness is a much harder problem, because we don't really understand how much of the brain works, the areas involved with conscious thought in particular, and the exact learning mechanisms the brain employs. The way we do deep learning these days is a very rough approximation of how the brain might work, and backpropagation is not really biologically plausible. So IMO, digitized consciousness is probably two to three orders of magnitude harder than this demo we're seeing here. Could unfortunately be a century away or more.

However, the medical applications should come much sooner. What they've shown is amazing. The monkey's control of the paddle seems to be precise and fast. If people who are paralyzed and wheelchair bound just had a way to operate the wheelchair, a robot arm, and communicate with the outside world reliably, that would already be amazing. If we could somehow reconnect their nerves or give them some kind of mechanized suit, it could be even better... And if Neuralink's tech works, that could be only 10-15 years away.

The thing i always think about with digital consciousness is that it would just be a clone that can’t distinguish itself from the OG consciousness that’s still resident in the brain.

The way this idea came to me was in a weird day dream thinking about the first experiment to attempt it... Patient lying on the table all wired up, a Dr. Frankenstein moment of ‘the upload’ followed by a screen witnessing the boot of the relocated consciousness. The doctor interacts with the persona, asking what it’s like and getting direct responses saying how amazing it is.

Meanwhile, amidst the commotion and celebration, you see the ’donor’ patient wake up in a reflection on the monitor and hear them faintly ask ‘did it work?’

I would argue that there is one consciousness, which is split into differing perspectives. I call that the multi-perspective being. This solves the teleport paradox (if you clone someone then where is the consciousness). The consciousness is everywhere, but split into differing perspectives.

Kudos to Neuralink engineers and doctors. We are still far away from human trials but this looks nothing short of SciFi

Really amazing demo. The moment that they just unplug the controller is impressive.

>>> Our first goal is to give people with paralysis their digital freedom back

PKD's "We Can Remember It For You Wholesale" notwithstanding, "Consumer Neurotech" feels like one of those categories that even the most visionary of sci-fi writers failed to build into all aspects of human existence in the future ;)

Ian Banks / Culture Series and the neural lace.. [1]

[1] https://www.economist.com/1843/2017/03/31/the-novelist-who-i...

It looked like the dexterity of the mind was quicker than the dexterity of the hand.

No brain signal to hand latency

That's just replaced with brain signal to chip then "decoder" then game software, which I would think is greater than nerve conduction velocity[1]. But then again, there's no "latency" of hand movement + joystick signal -> game software, so maybe it evens out?

[1] https://en.wikipedia.org/wiki/Nerve_conduction_velocity

Mind cursors seems amazing but I believe it must be very quickly exhausting hence preventing useful adoption for the mainstream.

To be fair, the same was said about touchscreens and VR.

Isn't the idea that currently we already do the "mind cursor" plus the actual muscle movements?

Holy poop that was amazing

So we directly connect to the blood and tissue CPU and enable things not possible before like bringing mind control to paraplegics and avenues for brain ransomware. Imagine the future.

What is the metal stick the monkey puts into its mouth?

it's a straw thru which he is drinking a banana smoothie.

I can't wait to see this technology be immediately used for nefarious purposes by dictatorships and power drunk governments.

No need to waterboard prisoners for information anymore when you can just suck data from their brains.

In most judicial systems, you cannot be compelled to testify against yourself. With this technology, I really worry that that right will slip away.

This feels like the precipice before dystopian hell, to me.

I know I'm conforming to a stereotype by saying this, but I cant stop thinking about how this could be the premise of a Black Mirror Episode.

I also think we should restrict such technology. Without restrictions we can play God, and I am in no way approving of doing anyhting similar to that.

The part about pairing the monkeys NeuraLink Device "like you do with a speaker" made me laugh. Not becuase it's funny, but because it is so incredibly absurd. I dont know wether to feel good about how far technology has come, or to feel terrified about all the implications.

(I can't help myself; Nietzsche once said (or wrote) "God is dead. God remains dead. And we have killed him. How shall we comfort ourselves, the murderers of all murderers? What was holiest and mightiest of all that the world has yet owned has bled to death under our knives: who will wipe this blood off us? What water is there for us to clean ourselves? What festivals of atonement, what sacred games shall we have to invent? Is not the greatness of this deed too great for us? Must we ourselves not become gods simply to appear worthy of it?")

A lot of people are focusing on the Bluetooth aspect of this which seems to be the most inconsequential part of it. Of course you need to connect to the brain device remotely; what do you want, a USB port on top of your head? Or should we go back to PS/2 or parallel ports and everyone can hide a 2in thick wire in their ponytail?

If we want the brain to be an input device then we must treat it like an input device. That means interfacing with it with the most modern/common standards we have, which, for better or worse (mostly worse), is currently Bluetooth.

Do you also think it's weird that people can put on VR headsets to alter their sight, or put in AirPods to subvert their hearing? 'Cause it is, but I think we've all normalized it to the point that it feels like something we're allowed to do. I think the same will happen with Neuralink at quite a fast pace.

I wasn't saying that using Bluetooth is absurd. The Absurd part, in my opinion, is treating the brain as an "input device". Ones brain should not be tinkered with.

You can not compare VR Headsets and airpods to a chip that is, literally, implanted IN your brain. - These devices generate Extrinsic Stimuli. The Neuralink however directly READS the Neuron Activity (as I understood it).

> Without restrictions we can play God

We've been playing God for about as long as human civilization has existed. Agriculture is playing God - overwriting the byproducts of nature. Medicine is playing God - overwriting natural selection.

I don't understand your core point: why is it inherently wrong to attempt to augment the brain? It seems like a purely instinctual response.

Is it God's will for paralyzed people to remain unable to interact with the world, and live the rest of their lives in misery? If we have a way to do a lot of good for a lot of people, isn't it inherently immoral to not develop the technology further?

youre right, but I fear a future where such techonolgy can be abused. I haven't carfully considered all the positives such technology will bring into our world. I was starteled by the potential of abuse and how far technology has come.

Let's stop changing organs then

The most amazing thing in this video is NOT how we can read signals from the monkey brain, but how a monkey brain can learn to generate signals to control the cursor. I think it is remarkable that a brain can learn to have such fine control of its neural activities.

Poor monkey. It should belong to the forest or other natural environment it was took from. I don't get it why there are still tests run on animals. The first what technology should bring is a replacement for every animal test subject.

we torture and kill so many more animals to eat meat and dairy, than we do to conduct science.

While I share your view, I think it's important to point out that this really isn't the best place to start if you want to reduce animal suffering.

Precisely. This is on the contrary one of the least bad areas to conduct animal experiments. Humanity would be nowhere without all kinds of animal experiments and I challenge everyone who thinks it's cruel to think about how cruel it would be for humans to erase all the advances made thanks to them.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact