I have not yet read through this post, but I want to note that we had a post on Less Wrong about this theory over a year ago. What do you think about that post, and the comments on it? What does it get right, and what do you consider to be mistakes in it?
It was an even more recent post, “Neural Annealing” that inspired me to write this one. The most popular reaction to “Neural Annealing” was “I don’t get it. How can this be important given the relatively low bandwidth of brainwaves compared to synapses?” I wrote a comment to answer that question, and then this “Connectome-Specific Harmonic Waves” article to address the implications in full.
I think “Connectome-specific harmonic waves and meditation” has the right idea. There’s not much to say since it’s just a list of links to other sources. The most interesting thing about that post is Vaniver’s comment. I think my article makes a good point that Veniver missed. Namely, that increasing resonance within a brain increases information flow between different parts. This gives a computational advantage above pure aesthetic symmetry. The goal isn’t simplicity. It’s coordination.
Vaniver makes a good point in his final paragraph. This isn’t addressed in “Connectome-specific harmonic waves and meditation” but is addressed in my post.
[A] ‘maximize harmony’ story needs to have really strong boundary conditions to create the same sorts of conflicts.
This is accounted for by CSHW because (1) oscillations within state networks are observably contained in their state networks and (2) high frequency oscillations propagate a shorter distance than low frequency oscillations.
Mike Johnson’s “A future for neuroscience” helped me write this post. I think the biggest problem with it is that Mike Johnson fails to recognize the fractal nature of these harmonics.
The fractal nature of CSHW makes possible a self-organizing system. CSHW’s power comes from how its fractal nature makes possible a bottom-up understanding of the human brain. Since Mike Johnson misses this, he instead generalizes downward from the macroscopic structure of the brain. Mike Johnson concludes incorrectly that different harmonics would be the same between different people and then generalizes from his mistake.
By operating top-down, Mike Johnson runs into the same problems neuroscience has struggled with since its inception. Cutting a fractal black box into pieces will never tell you how the box works. All it will do is replace the black box with smaller, identical black boxes. It’s better to investigate a fractal via induction than dissection.
Under my interpretation, CSHW is a self-organizing fractal, just like a FFNN. Therefore there’s no reason to assume that one person’s harmonic signature is likely to resemble someone else’s, especially at high frequency oscillations. (I can explain this in more detail if the reasons are not clear from my original post.)
Thanks for so clearly putting your thoughts down. Honestly, I liked your comment on my LW crosspost of Neural Annealing so that I added it to the end of the post on my blog.
Briefly, I wanted to note a key section of NA where I talk about “a continuum of CSHWs with scale-free functional roles”, which depending on definitions may or may not be the same thing as CSHWs being fractal:
-------------------
Last year in A Future for Neuroscience, I shared the frame that we could split CSHWs into high-frequency and low-frequency types, and perhaps say something about how they might serve different purposes in the Bayesian brain:
The mathematics of signal propagation and the nature of emotions High frequency harmonics will tend to stop at the boundaries of brain regions, and thus will be used more for fine-grained and very local information processing; low frequency harmonics will tend to travel longer distances, much as low frequency sounds travel better through walls. This paints a possible, and I think useful, picture ofwhat emotions fundamentally are: semi-discrete conditional bundles of low(ish) frequency brain harmonics that essentially act as Bayesian priors for our limbic system. Change the harmonics, change the priors and thus the behavior. Panksepp’s seven core drives (play, panic/grief, fear, rage, seeking, lust, care) might be a decent first-pass approximation for the attractors in this system.
I would now add this roughly implies a continuum of CSHWs, with scale-free functional roles:
Region-specific harmonic waves (RSHWs) – high frequency resonances that implement the processing of cognitive particulars, and are localized to a specific brain region (much like how high-frequencies don’t travel through walls) – in theory quantifiable through simply applying Atasoy’s CSHW method to individual brain regions;
Connectome-specific harmonic waves (CSHWs) – low-frequency connectome-wide resonances that act as Bayesian priors, carrying relatively simple ‘emotional-type’ information across the brain (I note Friston makes a similar connection in Waves of Prediction);
Sensorium-specific harmonic waves (SSHWs) – very-low-frequency waves that span not just the connectome, but the larger nervous system and parts of the body. These encode somatic information – in theory, we could infer sensorium eigenmodes by applying Atasoy’s method to not only the connectome, but the nervous system, adjusting for variable nerve-lengths, and validate against something like body-emotion maps.[2][3]
These waves shade into each other – a ‘low-frequency thought’ shades into a ‘high-frequency emotion’, a ‘low-frequency emotion’ shades into somatic information. As we go further up in frequencies, these waves become more localized.
Thank you for putting my comment on your blog. Its very flattering.
I would now add this roughly implies a continuum of CSHWs, with scale-free functional roles:
One of the most important most important implications of CSHWs’ is what you call their “scale-free functional roles” and what I call their fractal “scale invariance”. Terms like RSHW, CSHW and SSHW are just markers for arbitrary scales, like “gamma rays” and “infra-red” on the electromagnetic spectrum. I just finished an article attaching equations to this idea.
I have not yet read through this post, but I want to note that we had a post on Less Wrong about this theory over a year ago. What do you think about that post, and the comments on it? What does it get right, and what do you consider to be mistakes in it?
EDIT: Since you mention meditation, consider my questions also to apply to this more recent post on CSHW.
It was an even more recent post, “Neural Annealing” that inspired me to write this one. The most popular reaction to “Neural Annealing” was “I don’t get it. How can this be important given the relatively low bandwidth of brainwaves compared to synapses?” I wrote a comment to answer that question, and then this “Connectome-Specific Harmonic Waves” article to address the implications in full.
I think “Connectome-specific harmonic waves and meditation” has the right idea. There’s not much to say since it’s just a list of links to other sources. The most interesting thing about that post is Vaniver’s comment. I think my article makes a good point that Veniver missed. Namely, that increasing resonance within a brain increases information flow between different parts. This gives a computational advantage above pure aesthetic symmetry. The goal isn’t simplicity. It’s coordination.
Vaniver makes a good point in his final paragraph. This isn’t addressed in “Connectome-specific harmonic waves and meditation” but is addressed in my post.
This is accounted for by CSHW because (1) oscillations within state networks are observably contained in their state networks and (2) high frequency oscillations propagate a shorter distance than low frequency oscillations.
Mike Johnson’s “A future for neuroscience” helped me write this post. I think the biggest problem with it is that Mike Johnson fails to recognize the fractal nature of these harmonics.
The fractal nature of CSHW makes possible a self-organizing system. CSHW’s power comes from how its fractal nature makes possible a bottom-up understanding of the human brain. Since Mike Johnson misses this, he instead generalizes downward from the macroscopic structure of the brain. Mike Johnson concludes incorrectly that different harmonics would be the same between different people and then generalizes from his mistake.
By operating top-down, Mike Johnson runs into the same problems neuroscience has struggled with since its inception. Cutting a fractal black box into pieces will never tell you how the box works. All it will do is replace the black box with smaller, identical black boxes. It’s better to investigate a fractal via induction than dissection.
Under my interpretation, CSHW is a self-organizing fractal, just like a FFNN. Therefore there’s no reason to assume that one person’s harmonic signature is likely to resemble someone else’s, especially at high frequency oscillations. (I can explain this in more detail if the reasons are not clear from my original post.)
Thanks for so clearly putting your thoughts down. Honestly, I liked your comment on my LW crosspost of Neural Annealing so that I added it to the end of the post on my blog.
Briefly, I wanted to note a key section of NA where I talk about “a continuum of CSHWs with scale-free functional roles”, which depending on definitions may or may not be the same thing as CSHWs being fractal:
-------------------
Last year in A Future for Neuroscience, I shared the frame that we could split CSHWs into high-frequency and low-frequency types, and perhaps say something about how they might serve different purposes in the Bayesian brain:
I would now add this roughly implies a continuum of CSHWs, with scale-free functional roles:
Region-specific harmonic waves (RSHWs) – high frequency resonances that implement the processing of cognitive particulars, and are localized to a specific brain region (much like how high-frequencies don’t travel through walls) – in theory quantifiable through simply applying Atasoy’s CSHW method to individual brain regions;
Connectome-specific harmonic waves (CSHWs) – low-frequency connectome-wide resonances that act as Bayesian priors, carrying relatively simple ‘emotional-type’ information across the brain (I note Friston makes a similar connection in Waves of Prediction);
Sensorium-specific harmonic waves (SSHWs) – very-low-frequency waves that span not just the connectome, but the larger nervous system and parts of the body. These encode somatic information – in theory, we could infer sensorium eigenmodes by applying Atasoy’s method to not only the connectome, but the nervous system, adjusting for variable nerve-lengths, and validate against something like body-emotion maps.[2][3]
These waves shade into each other – a ‘low-frequency thought’ shades into a ‘high-frequency emotion’, a ‘low-frequency emotion’ shades into somatic information. As we go further up in frequencies, these waves become more localized.
-------------------
Thank you for putting my comment on your blog. Its very flattering.
One of the most important most important implications of CSHWs’ is what you call their “scale-free functional roles” and what I call their fractal “scale invariance”. Terms like RSHW, CSHW and SSHW are just markers for arbitrary scales, like “gamma rays” and “infra-red” on the electromagnetic spectrum. I just finished an article attaching equations to this idea.