Thread by @neurovium

"Second type of criticality in the brain uncovers rich multiple-neuron dynamics"...@PNASNews
pnas.org/content/116/26…

Since @neuroprinciples asked if I have changed my mind about criticality:
No I have not! Though it does not mean that I don't read and think about the work of those who are on the opposite camp. Science needs open mindedness, or else it would be faith. See the thread below...

So let me give some perspective on this. Many love the idea of "criticality" and jump the band wagon. But one has to be careful not to take analogies (sandpiles of Per Bak, criticality and phase transition in magnets, Ising) too freely. It will lead to distortion. See below...

Let's fist start with the electromagnetic field. We showed significant difference in frequency scaling between EEG and MEG, which can be explained if the extracellular medium (also including dura matter & skull) is globally non-resistive. @SpringerNature
link.springer.com/article/10.100…

Then I tested self-organized criticality in spikes and local field potential showing no clear evidence for power-law scaling or self-organized critical states in the awake and sleeping brain of cat, monkey and man @FrontPhysiol I suggested a hint to E/I
frontiersin.org/articles/10.33…

Next was to test the relationship between E/I and the oscillatory patterns
suggesting that inhibition-driven β- and γ-oscillations contribute to the reactivation of information through orchestrating highly coherent spiking activity patterns
@PNASNews
pnas.org/content/113/33…

Lately, we showed Ising ignores the inhibitory effects & dramatically overestimates synchrony among excitatory. And we point to the issues that exist with thermodynamic signatures of criticality, and why one ahould take this with a grain of salt
@PhysRevE
journals.aps.org/pre/abstract/1…

So, to sum it up, if you look at it with scrutiny and across many scales, the criticality idea just does not hold up. It does exist in vitro (neuron culture), but in vivo is truly a different beast. See below..

One has to take into account the heterogeneity of the computing components, their stochastic nature, their different timescales of excitability, the discontinuity of the excitable medium, the anisotropy of the medium, to come up with a solution that can really describe the system

So, there is no theory of cortical computation and dynamics yet...(and no it is not Bayesian or manifolds...those are just representation tools useful for cognitive tasks. But don't count as theories and mechanistic description rooted in biophysics). So the problem is still open!

Also nned to mention a tightly relevant notion: E/I dynamic balance is self generated by recurrent activity. This balance breaks down during seizures, where the temporal correlation of excitatory and inhibitory populations is disrupted. 
@SciReports
nature.com/articles/srep2…

@SciReports @threadreaderapp unroll