Bit of a sensationalist title - EEG/iEEG measurements have led to worthwhile clinical outcomes but also are an avenue for bunk pseudosccience so it pays to be careful whenever you see the words 'brain waves' paired with some large claim.
More accurate, much less sexy title: Intracranial Recordings Uncover Spiral and Concentric Brain Waves During Memory Tasks" (study completed only on small cohorts of epilepsy patients performing constrained memory tasks).
No inner workings revealed or laid bare, just another piece of evidence contributing to a small part of the debate in cortical electrophysiological studies on whether or not some data measures can be discarded as the downstream, messy noise of the neurons firing or actual, high-fidelity signals of activity
Of course any science coming out that observes things that LLMs don't have is ot going to find much excitement on HN. It would mean to face reality that LLMs are merely child toys compared to the real deal which is the human brain.
I think skepticism is healthy for the vast majority of sensationalized titles especially those coming from corporate entities. I'm noting the lack of excitement here because it is, in actual fact, not a revolutionary discovery, it is a possible, un-replicated new avenue of study that could become revolutionary with enough of an evidence base. There are many dozens of these same kinds of discoveries published all the time, often leading to dead-ends (which is actually great - that's another rabbit hole you no longer have to chase - here's to null findings!) and this is one of them in spite of the title.
“The work coming out is moving this from ‘Are they relevant?’ to ‘This is a major motif of how the cortex processes information,’” said Earl K. Miller(opens a new tab), a cognitive neuroscientist at the Massachusetts Institute of Technology.
Brain waves are to biological neural populations what large-scale, low-dimensional activation modes in the transformer’s residual stream are to an LLM.
> Surprisingly Complex Waves Reveal the Brain's Inner Workings
From my ignorant point of view I can't quite make sense of the title. Brain is probably the last organ one would be surprised to find complexity in. I can see "surprisingly complex wave reveal spleen inner workings" being a good title, like "wow, we though it was a dumb blood filter, but look what we found!" kind of a thing. But for the brain isn't the default assumption that has it complexity we haven't figure out yet?
I suspect, if our conscious experience === the waves created by neuron's firing, that the hum of the GPU when running a LLM doesn't have the dynamic range to meaningfully talk about having a conscious experiences.
Bit of a sensationalist title - EEG/iEEG measurements have led to worthwhile clinical outcomes but also are an avenue for bunk pseudosccience so it pays to be careful whenever you see the words 'brain waves' paired with some large claim.
More accurate, much less sexy title: Intracranial Recordings Uncover Spiral and Concentric Brain Waves During Memory Tasks" (study completed only on small cohorts of epilepsy patients performing constrained memory tasks).
No inner workings revealed or laid bare, just another piece of evidence contributing to a small part of the debate in cortical electrophysiological studies on whether or not some data measures can be discarded as the downstream, messy noise of the neurons firing or actual, high-fidelity signals of activity
Of course any science coming out that observes things that LLMs don't have is ot going to find much excitement on HN. It would mean to face reality that LLMs are merely child toys compared to the real deal which is the human brain.
I think skepticism is healthy for the vast majority of sensationalized titles especially those coming from corporate entities. I'm noting the lack of excitement here because it is, in actual fact, not a revolutionary discovery, it is a possible, un-replicated new avenue of study that could become revolutionary with enough of an evidence base. There are many dozens of these same kinds of discoveries published all the time, often leading to dead-ends (which is actually great - that's another rabbit hole you no longer have to chase - here's to null findings!) and this is one of them in spite of the title.
Wow!!
“The work coming out is moving this from ‘Are they relevant?’ to ‘This is a major motif of how the cortex processes information,’” said Earl K. Miller(opens a new tab), a cognitive neuroscientist at the Massachusetts Institute of Technology.
Brain waves are to biological neural populations what large-scale, low-dimensional activation modes in the transformer’s residual stream are to an LLM.
This is what Walter J. Freeman III* spent his career working on, surprised he wasn't mentioned. I used to be very obsessed with him.
https://en.wikipedia.org/wiki/Walter_Jackson_Freeman_III
Nonlinear brain dynamics as macroscopic manifestation of underlying many-body field dynamics, https://arxiv.org/html/q-bio/0511037
https://web.archive.org/web/20070609235302/http://sulcus.ber...
-----
[*] No, not that guy; the son of that guy.
Edit: I've never seen a wikipedia page with so many typos.
> Surprisingly Complex Waves Reveal the Brain's Inner Workings
From my ignorant point of view I can't quite make sense of the title. Brain is probably the last organ one would be surprised to find complexity in. I can see "surprisingly complex wave reveal spleen inner workings" being a good title, like "wow, we though it was a dumb blood filter, but look what we found!" kind of a thing. But for the brain isn't the default assumption that has it complexity we haven't figure out yet?
I wonder what the "waves" in LLMs are; The global firing patterns that can help us interpret what the weights are doing en-masse
Care to define 'waves' in this context?
I suspect, if our conscious experience === the waves created by neuron's firing, that the hum of the GPU when running a LLM doesn't have the dynamic range to meaningfully talk about having a conscious experiences.
But that's armchair conjecturing.
I thought this research had already been done and published by Anthropic. https://www.anthropic.com/research/tracing-thoughts-language...