Mind reading technology: helmet aims to unlock brain's secrets
bloomberg.com
bloomberg.com
Depending on what condition you're suspected to have that's a perfectly reasonable question, but that aside this showcases the tendency for reductionism that has become prevalent.
The mind is not equivalent to the brain, and neither is comparable to a mechanical device. The better analogy to the statement from Johnson is, if you wanted to inspect software you certainly would not start to turn transistors on your hardware upside down, or if you wanted to get insights into a Dostoevsky novel you wouldn't start to inspect printer-ink on the pages.
For some purely physical ailment of the brain better scanning techniques may very well be a good thing, but 90% of what his project seems to be about has little to do with the physical brain and a lot to do with studying emergent phenomena of the human mind, in which case getting lost in brain chemistry is probably not that helpful, and at worst seems to lead to a sort of ML-powered craniometry.
In particular, turning transistors upside down is a good way to discover that software is fairly orthogonal to hardware orientation, if you didn't already know that.
Not that anything prevents someone from claiming otherwise...
BOLD signals are great, but they are based on oxygenation of blood, which is a latent, and at best a second order effect for metabolic activity, which is then reference to be brain activity...
I'll have to dig up the studies, but if you've been paying attention on HN and elsewhere we are starting to find that 'strengthening' neurons with a fiber like substance around it reduces the electrical activity required to achieve the same neural-result. Combine this with recent papers saying that mice smell a rose with a different, migrating, set of neurons each time... and you end up thinking, damn... this really isn't the right approach to BCI / mind reading.
We're never going to get the kind of fidelity we need for advanced diagnostics (much less mind reading) with fNIRS or fMRI. Bringing down the cost of MEG tech is the only realistic way forward.
I think you're talking about myelination, which is a result that's been around for quite a while. Myelination probably wouldn't confound this kind of "brain reading" because it's relatively stable, i.e. changes occur over the course of days at a minimum, and supposedly infrequently in the fully developed brain. Any "brain reading" will have to account for changing neural efficacies, as that's what learning is, and the tricky stuff would be learning on the scale of seconds or minutes (e.g. changes in synaptic efficacy) rather than things like myelination.
That being said, while the conventional wisdom is that myelin is stable over the course of months/years in the adult brain, perhaps it's results suggesting more ongoing myelin plasticity that are the more recent results you're referring to (a general trend in recent neuroscience). If so, I'd love to see those studies.
Disclaimer: I'm soooo skeptical of this kind of extracranial signals being used for "unlocking the brain's secrets." There's lots you can do with those signals, sure, just look at the past decades of EEG research. But they seem to be wanting to go far deeper with fundamentally the same technology, which seems like a dicey proposition to me. So while I think there are lots of potential critiques of this technology, I just don't think myelination is the critique to go with.
To add to this fNIR is only really able to penetrate a marginal depth of a few centimeters of periphery cortical tissue.
Even if it was a perfect signal, the functional information it would provide would not give us a full picture of all the sub-cortical activity.
In this case I don’t see the value proposition. Just bunch of marketing speak.
This is vacuumware.
The most interesting thing is that one of their helmets, Flow, is using IR backscatter sensing rather than the electric or magnetic sensing that most other BCI devices (including their other helmet, Flux) use.
The system is still pricey at $50-100k or $3.6k/mo, but they're also planning on selling a simplified version at $5k: https://docs.kernel.com/docs/how-much-does-a-flow-system-cos...
There are almost limitless possibilities. For example, if such a device were good and sensitive and accurate enough, I bet you could create some software that would let you quickly compose and produce music in a DAW via thoughts. And perhaps also even something more abstract like semi-implicit or subconscious "intentions" or notions. And if you might ask "okay, but what would that be good for?", I could write a whole essay I suppose.
(Disclaimer: I've been waiting for years for something to come out that would let me try to build something like this. I'm sure it'll likely still be a while before it's possible to make and practical for consumers to buy and use, but I think it could potentially happen within our lifetimes.)
It'd be much harder, but I think you could maybe even program certain things via thoughts. Maybe only a limited subset of things to start with, like visualizing a web page and having the basic skeleton, shapes, and colors quickly filled in, but if you look on the span of centuries, maybe eventually full programs, too.
I'm not sure at all if this would actually be the case, but maybe some people could potentially achieve a very high WPM if their thoughts were converted into text. Much higher than anyone could type by hand. Especially if it were something where you were reciting text/speech from memory, so that it could happen at your raw speed of thought, rather than writing an essay where you have to think about what you actually want to say as you're doing it, in which case it might be fairly close to your typing or handwriting speed.
Maybe this particular iteration of it won't be that great or that useful, but some future thing definitely will be.
The question was what’s it good for as it is today, based on people with knowledge about it.
In particular the only thing you demonstrated here is that you have absolutely no knowledge about it.
I was just matching the patronizing, brusque tone in your original post.
>The question was what’s it good for as it is today, based on people with knowledge about it.
I apologize, then, because I don't think that was clear from your post, which contained only the text "So what is this tech good for in the first place." If your question were "So what are the device's current capabilities?" or something similar, then I would've responded differently.
>In particular the only thing you demonstrated here is that you have absolutely no knowledge about it.
I do, actually. Your question (ending in a period) came across as extremely condescending and dismissive, and as if you were questioning the value of BCIs in general. So my goal with that post was to try to make an argument for the potential of what BCI tech is good for in the first place, as I interpreted your post.
If you want an overview of what their company's devices can do today and what's planned, this recent podcast episode with the CEO is a good start: https://www.youtube.com/watch?v=1YbcB6b4A2U
My original post simply asked: "what is it good for". If you insist to see this as a rhetorical sarcastic question, that's your problem. I asked it as a question.
> I do, actually. Your question (ending in a period) came across as extremely condescending and dismissive
So, the fact I asked a question and didn't end it with a question mark resulted in this emotional outburst and interpreting it as "extremely condescending and dismissive". Think about that. It's illogical and outright hysterical.
I wasn't thinking about "BCI tech" in general at all. I was curious about this device. You're completely, utterly off the mark.
> If your question were "So what are the device's current capabilities?" or something similar, then I would've responded differently.
I don't want to know what are its current capabilities because I understood that from the article. But I don't know what its current capabilities are good for in practical use for their potential customers. You may want me to ask the questions you want me to ask, that's cute, but it's not how asking questions works. People ask what they want to ask, and you answer if you can and want to answer. You don't go around insulting people and inferring ill intent in their questions for not asking the questions YOU want them to ask.
You see, when you sell $50k device, you believe you're providing $50k of value. It's not an internal research device anymore. That implies they see some uses for it, and I was wondering what those are, without listening to an entire podcast.
And you clearly don't know. And that isn't my fault.
* Medical settings in which patients are slowly losing the ability to move or communicate, being able to look at a keyboard to type messages to your loved ones or perform basic actions (e.g. turning music on) make a terrible situation a tiny bit better.
* Entertainment/gimmicks: you go to Disneyland and you put on a cap/helmet and suddenly you can interact with objects just by looking at them (they have a flickering light next to them that pulses in a certain binary pattern) - it's like magic.
The second usecase is very difficult today though, the tech that I worked with (+5 years ago) needed to be trained for every new user and a training time of 10 minutes was considered very short. That of course makes the Disneyland usecase more difficult. Also you need to make sure you don't move or blink too much - maybe fNIR has less issues with that.
so . . . scammers scamming scammers?
> Devices are also going out to Harvard Medical School, the University of Texas, and the Institute for Advanced Consciousness Studies (a California lab focused on researching altered states) to study such things as Alzheimer’s and the effect of obesity on brain aging, and to refine meditation techniques.
> Once their Kernel helmets arrive, Boas and his colleagues plan to observe the brains of people who’ve had strokes or suffer from diseases such as Parkinson’s. They want to watch what the brain does as individuals try to relearn how to walk and speak and cope with their conditions.
As far as I understand the whole thing, the helmet is supposed to replace existing, clunky, not mobile version of brain scanners.
https://www.sciencedirect.com/science/article/pii/S246806722...
Looking forward for that mysterious free time to build this & play around with neuro-feedback
The fundamental question for these devices will not be one of neuroscience or practicality: the fundamental question will be how good is their signal-to-noise ratio (SNR). If they even approach clinical EEG SNR using their fNIRS, then these will be widely adopted, off the top of my head, for two uses: seizures and meditation (and the latter doesn't require anything even vaguely approaching clinical SNR).
Epilepsy is one of the most prevalent neurological conditions (the most prevalent?) in the world, affecting more than 1% of people, and it's somewhat hard to diagnose effectively due to sparsity of seizures. You can only have patients hooked up to an EEG for so long, and you just have to hope that you get a seizure during that time. Especially when doing preliminary localization in preparation to surgically intervene (the only effective treatment in ~1/3 of epilepsy cases), you need to catch that seizure when it happens. So any vaguely cheap option that allows patients to monitor for seizures in their day-to-day will see widespread adoption by hospitals. (This doesn't even account for the potential to predict seizures by a matter of seconds or minutes, which can be invaluable for someone with moderately frequent seizures.)
The meditation one sounds more out-there (especially if you read the OP article...), but is actually pretty well supported by studies. Particular results are debatable, but it's fairly well-established that meditation causes a large scale change in the frequency composition of brain waves that's pretty easy to capture on EEG. What's more, you can induce these changes consciously by providing feedback (e.g. a moving dot on a screen), a result demonstrated in monkeys (and I hope in humans, though I haven't seen that myself). There are still some lingering questions, such as the relative efficacy of inducing these changes via feedback vs via meditation, but they're questions at the level that they'll be overlooked by whatever company invites you to learn to meditate with these helmets. Intuitively, you could think of this as analogous to those posture-feedback devices I see advertised everywhere. Those notice your back is bending too much, and alert you. This would notice your thoughts are wandering, and alert you.
A third, very vague, but important consequence of a device that (finally) has good SNR is the advent of a critical-mass hacker culture around brain-computer interfaces.
If anyone knows anything about the SNR, or about more recent results on meditation, I'd love to read more.