Disclaimer: I'm a mathematician, well versed in algebraic topology & also have a PhD in biology.
This is one of the most disingenuous papers I've read in a while, indeed my first reaction was "wow, what a load of horse shit".
The editorial on the journal's main "selling point" is that it applies algebraic topology "in a novel way, never before seen" to neuroscience. There is absolutely no reason to use algebraic topology in this context.
They're basically counting in a variety of ways network motifs - which Uri Alon & others have been doing for years. This is entirely possible to do using directed graphs, basic combinatorics, and elementary probability.
As other commenters have pointed out, there's at best a weak correlation to actual brain activity, and the amount of data is not actually that large and within reach of off-the-shelf graph software to process (or a python script).
This is also not the first time the senior author of the paper has hyped otherwise lackluster results. It's easier to impress a TED audience than your peers. Here's a critical review on Nature:
Yup. They're treating in silico statistics as faithful to real "brain activity", because the network layout the scientists used was a detailed representation of a snapshot of a part of a real mouse brain. Don't worry if the neurons would behave differently than the simulated neurons, or if that could lead to totally different spikes globally, or if the hype irreversibly makes the story go viral and people still believe claims that were never even part of the original test years later because the story was good enough..
Also... "2.3. Topology Organizes Spike Correlations" <- "Energy makes it go!"
Microtubules? Hah! Such an outmoded and unfashionable quantum model of cognition. Here, let me introduce you to the lithium-7 spin-based phosphate-calcium ion-ball entanglement hypothesis of quantum mentation; https://www.quantamagazine.org/a-new-spin-on-the-quantum-bra...
TIL, thanks! unfortunately I think I see where they're going with it:
“I believe that if phosphorus nuclear spin is not being used for quantum processing, then quantum mechanics is not operative in longtime scales in cognition,” Fisher said. “Ruling that out is important scientifically. It would be good for science to know.”
> It's easier to impress a TED audience than your peers.
this seems like a perspective on a problem with pure theoretical research. like why print "Cycles of Time" instead of publishing in a peer reviewed journal? is Penrose being disingenuous or is he just trying to share an idea he finds interesting even if reasonably speculative?
the EPFL result is interesting even if they wanted to analyze it as network motifs in a directed graph, but that wouldn't account for the symmetry they presumably observed in the "higher dimensional polyhedrons" would it?
The University of Lausanne's Blue Brain Project has a lot of source code up at Github to take a look at [1], including some kind of specialized neuron ray-tracing renderer. Not too sure which data sources it can use. For example, is (a part of) the 31.000 neuron rat brain available somewhere?
They also seem to have developed a window system intended for large touch screens [3] that they use for "all hands" meetings on their brain research [4].
This article reminds me of this polychronization paper that describes how networks of spiking neurons result in the emergence of groups, and some discussion on how that could explain consciousness as attention to memory.
One of the simplest known organisms, the worm C. Elegans, has a brain consisting of 302 neurons that we have mapped exactly. (The entire thing is 959 cells.) We still can't simulate it correctly.
A rat's brain is O(10^7) neurons, while a human brain is O(10^11) neurons, as compared to C. Elegans with O(10^2) neurons.
Which is to say, trying to simulate a rat brain while we know we can't simulate C. Elegans is like a prehistoric human claiming he can build an F1 racecar before he's invented the wheel.
It's an interesting hypothesis for what the patterns of simulated firings "mean". I don't see any reference link to experimental methods in silico or in vivo follow up. Anyone have any reference articles for this hypothesis?
This paper (2008) describes the general approach, including the use of Betti numbers for classification of observed phenomena (in primates, not in silico):
I don't mean to be a dick but the approach they use is kinda like finding "hidden patterns" of New York City by noticing whenever small groups of conversing New Yorkers coincidentally all used the same word in a conversation.
I guess I was hoping more for something like "the brain keeps track of information as an amalgamation of partially-understood signals, due to a neural uncertainty principle describing how what neurons compute is constrained by the traits of the signals they should recover information from"
The concentrated power of mind has the ability to know without the help of the senses. In the inner world, the mind does not even need the help of the brain. The brain is only a medium for the energy called mind. It is a powerhouse, not the power; it is a distribution center but not the energy. Many modern physicists do not accept this theory, for they know only one method: applying sophisticated instruments to amplify sensory experiences. When one uses such instruments to study mental functions, they fail, for these instruments can only measure the superficial workings of the brain. Those who follow this method nevertheless contend that the brain is the mind, and that if they can develop the instruments to study the brain’s functioning they will be able to understand the mind. They are like the blind man who holds the tail of an elephant and contends that an elephant is like a snake because its tail feels like one.
Modern technology and scientific knowledge are valid as far as the external world is concerned. But there is another exact science that helps one know the unknown dimensions of life. The Samkhya system of philosophy has given birth to yogic science. It is far more advanced and methodical than modern technology and the scientific method that is being used to probe into the microscopic and macroscopic levels of the physical universe.
This is one of the most disingenuous papers I've read in a while, indeed my first reaction was "wow, what a load of horse shit".
The editorial on the journal's main "selling point" is that it applies algebraic topology "in a novel way, never before seen" to neuroscience. There is absolutely no reason to use algebraic topology in this context.
They're basically counting in a variety of ways network motifs - which Uri Alon & others have been doing for years. This is entirely possible to do using directed graphs, basic combinatorics, and elementary probability.
As other commenters have pointed out, there's at best a weak correlation to actual brain activity, and the amount of data is not actually that large and within reach of off-the-shelf graph software to process (or a python script).
This is also not the first time the senior author of the paper has hyped otherwise lackluster results. It's easier to impress a TED audience than your peers. Here's a critical review on Nature:
https://www.nature.com/news/rethinking-the-brain-1.17168