Talk by Prof. Viola Priesemann (CSN Virtual Seminar)
We hereby announce the next talk in the CSN Virtual Seminar.
Self-Organized Criticality in Neuroscience
A popular hypothesis states that the brain operates in a self-organized critical state because, in models, criticality maximizes a number of computational properties. Experimental testing of this hypothesis, however, poses a number of challenges. First and foremost, criticality is a large-scale phenomenon, whereas recordings of brain activity are limited to a small subset of all neurons, or else one must use "coarse" measures such as LFP, EEG, or BOLD signals. We show how these sampling problems can bias outcomes and interpretation. To overcome the sub-sampling problem, we developed an invariant estimator that allows us to determine the distance to a critical point quite precisely. Interestingly, in spike recordings from rats, cats, monkeys, and humans, we found that the brain does not operate exactly at the critical point, but maintains a certain distance from criticality. Why is this so? First, it allows the brain to maintain a safe margin from the supercritical regime, which is associated with epilepsy. Second, and more importantly, a system near the critical transition can take advantage of its characteristic parameter sensitivity and thereby tailor its dynamics and computations to task requirements. Indeed, in neuromorphic chips, we find that simple tasks are better solved away from the critical point, while more complex tasks benefit from criticality. Thus, in sum we show how the regime around a critical transition can be derived (a) precisely despite subsampling neural activity and (b) how this subcritical regime fosters flexible tuning of the network to task requirements.
Prof. Viola Priesemann
Max-Planck-Institute for Dynamics and Self-Organization and Georg-August-University Göttingen
You can dial in using the following zoom room information:
https://zoom.us/j/99314955130?pwd=YXkvaW1wY2s2OHJsMFRWVDNtcERiUT09
Meeting ID: 993 1495 5130
Passcode: 4pM2Z8