Musk says Neuralink will merge human brains with AI. How realistic are these ideas? Is this our future, or are Musk’s ideas fantastical sci-fi imaginations?
Tim's focus is on the bigger picture and how something like Neuralink gets us to a new paradigm in the way we communicate. He traces the history of human communication and talks about how language, despite being the best we've got, is actually an extremely slow and inefficient communication method. Then he moves on to speculating about how a future with direct brain-to-brain communication would look. It's a fascinating read, but I don't know enough about the complexities involved to evaluate which parts are realistic.
If you ever get around to reading it, I'd be curious to hear your take.
Also, now that I know that the term "fusiform face area" is a thing, my life is that much better!
Wow! When you say long-read, you really mean long-read!!
I must admit, I didn't read the entire post in detail, but I scanned the relevant sections.
My initial thoughts: Many people think that the difference between replacing lost motor function and replacing lost memories is just a difference in bandwidth. If we just increase the bandwidth by getting a whole bunch of electrodes into the brain, then we will be able to upload memories into the brain.
I think this misses the complexity of the problem. The difficulty is not just a bandwidth problem. There's a difference in kind. Tapping into the mapping of the motor cortex to read patterns of brain activity is an entirely different problem than trying to upload abstract information into the brain.
Yeah what I got from your piece is that some things are much easier to "map" to areas of your brain than others. What about Tim's premise of enabling brain-to-brain communication? I guess that's even an extra order of magnitude more complex, since we're talking about two separate brains having to encode and decode the signals?
I like the futurism aspect here, but it's quite clear that we're nowhere near to having the necessary capabilities just yet.
I never like to say never. It could be theoretically possible, one day.
A few years ago, some friends of mine developed an EEG BCI setup that allow two users to communicate with one another. It was a game set up where using joint EEG signals should have helped both players play the game. I tried it out a few times. It was a weird experiment to participant in.
Their main finding was that free communication was possible between two people, but it was much slower than just talking to each other. The participants just ended up doing their own thing (i.e make individual decisions), or they just took turns as to who would make the decisions in the game.
When you say you tried it out, you mean you had the chance to communicate with another person via the interface? Or was it set up in a way that you were only sending signal directly to a game and there was no human-to-human communication involved? (If I sound like a complete layman, it's because that's exactly what I am.)
Yes, it was similar to the Wisconsin study. Except in that study one person wore an EGG cap and the other person received TMS. The person wearing the EEG cap would think about moving their hand, the EEG system would then send that pattern of activity to a computer that would interpret that pattern of activity as "move hand" and then send (via the internet I believe) that signal to the other person who received a pulse of TMS over the motor cortex (which is just a magnetic pulse that makes neurons under the TMS coil fire) -- which causes TMS person's hand to twitch -- ultimately making them press a button.
When I say tried it out, I mean I was working in the same lab as the people running the experiment -- they were colleagues. When we set up experiments like this we like our friends (other people in the lab) to try the experiment first before we test it on naive participants -- just to iron out any bugs in the programming and setup. In this particular case I tried it out many times because I am good friends with the main researcher.
In my friends experiment, both people wore EEG caps. If I remember correctly, the game had things that gave you points and things that took points away. To play the game successfully both players should collaborate -- gaining points as well as avoiding losing points.
Yes, EEG signals from both participants were sending signals directly to the game -- the human-to-human communication happens because what happens on the screen is determined by the combination of what each person was thinking. The feedback comes through the computer screen. The setup was designed so that together the participants would gain more points if they collaborated together than if they worked independently. Perfect scores were only possible with two brains working together.
The EEG signals were measuring how much each participant was 'enhancing' the 'gaining points things' and how much they suppressed the 'lose points things'. Ultimately you'd want both, gain more points and avoid losing points. The best performance worked when both people worked together. But in this particular experiment, the participants found it easier just to take turns in controlling the game. I don't believe the experiment was ever published.
So it was less brain-to-brain and more two-brains-to-one-game situation.
This area of research is always exciting to me if you look far enough into the future and imagine what might be.
Must've been cool to have this behind-the-scenes access to studies and tests that might've not been published or widely discussed but were still fascinating to observe/participate in.
Writing about your own work and lab experiences might make great topics for some future blog post, as long as they relate to the topics you want to cover. I know I'd love to hear more.
Another amazing piece! Maybe AGI could help us map the brain better to create augmented intelligence, but for now, we’ll just have to settle for restoring limb movements I suppose :p
Wow. Absolutely loved this detailed analysis on Neuralink, what it really is, and what it's not.
Thank you! I'm happy you enjoyed it
Thanks, Suzi!
Your piece is a great companion to this long-read by Tim Urban from 2017: https://waitbutwhy.com/2017/04/neuralink.html
Tim's focus is on the bigger picture and how something like Neuralink gets us to a new paradigm in the way we communicate. He traces the history of human communication and talks about how language, despite being the best we've got, is actually an extremely slow and inefficient communication method. Then he moves on to speculating about how a future with direct brain-to-brain communication would look. It's a fascinating read, but I don't know enough about the complexities involved to evaluate which parts are realistic.
If you ever get around to reading it, I'd be curious to hear your take.
Also, now that I know that the term "fusiform face area" is a thing, my life is that much better!
Wow! When you say long-read, you really mean long-read!!
I must admit, I didn't read the entire post in detail, but I scanned the relevant sections.
My initial thoughts: Many people think that the difference between replacing lost motor function and replacing lost memories is just a difference in bandwidth. If we just increase the bandwidth by getting a whole bunch of electrodes into the brain, then we will be able to upload memories into the brain.
I think this misses the complexity of the problem. The difficulty is not just a bandwidth problem. There's a difference in kind. Tapping into the mapping of the motor cortex to read patterns of brain activity is an entirely different problem than trying to upload abstract information into the brain.
Yeah what I got from your piece is that some things are much easier to "map" to areas of your brain than others. What about Tim's premise of enabling brain-to-brain communication? I guess that's even an extra order of magnitude more complex, since we're talking about two separate brains having to encode and decode the signals?
I like the futurism aspect here, but it's quite clear that we're nowhere near to having the necessary capabilities just yet.
I never like to say never. It could be theoretically possible, one day.
A few years ago, some friends of mine developed an EEG BCI setup that allow two users to communicate with one another. It was a game set up where using joint EEG signals should have helped both players play the game. I tried it out a few times. It was a weird experiment to participant in.
Their main finding was that free communication was possible between two people, but it was much slower than just talking to each other. The participants just ended up doing their own thing (i.e make individual decisions), or they just took turns as to who would make the decisions in the game.
That's really fascinating! Sounds quite similar to this University of Wisconsin experiment from 2014: https://www.washington.edu/news/2014/11/05/uw-study-shows-direct-brain-interface-between-humans/
When you say you tried it out, you mean you had the chance to communicate with another person via the interface? Or was it set up in a way that you were only sending signal directly to a game and there was no human-to-human communication involved? (If I sound like a complete layman, it's because that's exactly what I am.)
Yes, it was similar to the Wisconsin study. Except in that study one person wore an EGG cap and the other person received TMS. The person wearing the EEG cap would think about moving their hand, the EEG system would then send that pattern of activity to a computer that would interpret that pattern of activity as "move hand" and then send (via the internet I believe) that signal to the other person who received a pulse of TMS over the motor cortex (which is just a magnetic pulse that makes neurons under the TMS coil fire) -- which causes TMS person's hand to twitch -- ultimately making them press a button.
When I say tried it out, I mean I was working in the same lab as the people running the experiment -- they were colleagues. When we set up experiments like this we like our friends (other people in the lab) to try the experiment first before we test it on naive participants -- just to iron out any bugs in the programming and setup. In this particular case I tried it out many times because I am good friends with the main researcher.
In my friends experiment, both people wore EEG caps. If I remember correctly, the game had things that gave you points and things that took points away. To play the game successfully both players should collaborate -- gaining points as well as avoiding losing points.
Yes, EEG signals from both participants were sending signals directly to the game -- the human-to-human communication happens because what happens on the screen is determined by the combination of what each person was thinking. The feedback comes through the computer screen. The setup was designed so that together the participants would gain more points if they collaborated together than if they worked independently. Perfect scores were only possible with two brains working together.
The EEG signals were measuring how much each participant was 'enhancing' the 'gaining points things' and how much they suppressed the 'lose points things'. Ultimately you'd want both, gain more points and avoid losing points. The best performance worked when both people worked together. But in this particular experiment, the participants found it easier just to take turns in controlling the game. I don't believe the experiment was ever published.
Thank you for sharing all the details!
So it was less brain-to-brain and more two-brains-to-one-game situation.
This area of research is always exciting to me if you look far enough into the future and imagine what might be.
Must've been cool to have this behind-the-scenes access to studies and tests that might've not been published or widely discussed but were still fascinating to observe/participate in.
Writing about your own work and lab experiences might make great topics for some future blog post, as long as they relate to the topics you want to cover. I know I'd love to hear more.
Another amazing piece! Maybe AGI could help us map the brain better to create augmented intelligence, but for now, we’ll just have to settle for restoring limb movements I suppose :p
hahaha maybe.
It's remarkable, isn't it, that restoring basic limb function is considered just the baseline achievement!
Amazing times we live in!