For context, the debate in the field has been whether these are epiphenomena of the actual neuronal activity, or a meaningful driver of further activity.
I don’t think these studies quite rest the case. As Buzsaki comments in the article, the action is in the cells to create these waves. Do the waves themselves get measured and impact what happens next? The issue at hand is: synaptic currents are stronger and we know neurons respond to them. The wave behavior is in the extracellular fluid as well, and that part we have evidence neurons don’t respond much to.
That said, thing the article doesn’t talk about is astrocytes. Especially in the cortex, most synapses are tripartite. There’s an astrocytic end foot that sheathes the synapse and regulates how much of the bulk extracellular space has access to the synapse. And one astrocyte can touch thousands of synapses and in humans, hundreds of thousands. They also form a syncitium, through gap junctions/electrical synapses.
Do astrocytes respond to the extracellular wave behavior, which potentially changes the extent and behavior of the syncitium, and thus impact the neurons they touch?
That’s still an open question but that’s what I’d be interested to find out more about.
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
I wouldn't be so sure that the brain necessarily clocks faster than that. I think it's more likely that faster chains of thought are defined by the flow of the wave rather than its frequency.
Of course any science coming out that observes things that LLMs don't have is not 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. You'll very often find me in those OAI/Anthrop threads being very critical of their claims.
I like applying that same rigor to articles like this one. This is not a revolutionary discovery, it is a possible, un-replicated new avenue that could become revolutionary with enough of an evidence base. But there are many dozens of these same kinds of discoveries published all the time often leading to dead-ends (which is actually great since that's another rabbit hole you no longer have to chase - here's to null findings!)
In the same way you shouldn't accept PR regurgitated by so-called 'independent' journals when they come from corporate and government institutions, the same rigor should be applied to PR from universities seeking grants
It’d be interesting to try and model the various waveforms observed as time crystals and see if the mathematics of the latter can (partially) predict the former.
> 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?
Brain waves are to biological neural populations what large-scale, low-dimensional activation modes in the transformer’s residual stream are to an LLM.
“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.
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.
For context, the debate in the field has been whether these are epiphenomena of the actual neuronal activity, or a meaningful driver of further activity.
I don’t think these studies quite rest the case. As Buzsaki comments in the article, the action is in the cells to create these waves. Do the waves themselves get measured and impact what happens next? The issue at hand is: synaptic currents are stronger and we know neurons respond to them. The wave behavior is in the extracellular fluid as well, and that part we have evidence neurons don’t respond much to.
That said, thing the article doesn’t talk about is astrocytes. Especially in the cortex, most synapses are tripartite. There’s an astrocytic end foot that sheathes the synapse and regulates how much of the bulk extracellular space has access to the synapse. And one astrocyte can touch thousands of synapses and in humans, hundreds of thousands. They also form a syncitium, through gap junctions/electrical synapses.
Do astrocytes respond to the extracellular wave behavior, which potentially changes the extent and behavior of the syncitium, and thus impact the neurons they touch?
That’s still an open question but that’s what I’d be interested to find out more about.
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
And still all slow stuff, 4 to 12 Hz. There must be something in the brain that clocks faster, but nobody seems to have found it yet.
I wouldn't be so sure that the brain necessarily clocks faster than that. I think it's more likely that faster chains of thought are defined by the flow of the wave rather than its frequency.
Of course any science coming out that observes things that LLMs don't have is not 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. You'll very often find me in those OAI/Anthrop threads being very critical of their claims.
I like applying that same rigor to articles like this one. This is not a revolutionary discovery, it is a possible, un-replicated new avenue that could become revolutionary with enough of an evidence base. But there are many dozens of these same kinds of discoveries published all the time often leading to dead-ends (which is actually great since that's another rabbit hole you no longer have to chase - here's to null findings!)
In the same way you shouldn't accept PR regurgitated by so-called 'independent' journals when they come from corporate and government institutions, the same rigor should be applied to PR from universities seeking grants
It’d be interesting to try and model the various waveforms observed as time crystals and see if the mathematics of the latter can (partially) predict the former.
> 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?
> But for the brain isn't the default assumption that has it complexity we haven't figure out yet?
I thought AGI, um I mean "super intelligence", is just around the corner, so we have essentially already cracked it, just needs some fine tuning? /s
Brain waves are to biological neural populations what large-scale, low-dimensional activation modes in the transformer’s residual stream are to an LLM.
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.
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.
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...