ELSC Seminar Series

The critical brain hypothesis suggests that the human brain operates near the threshold between two distinct dynamical states— a critical transition point that is believed to optimize information processing, representation, and adaptability. In this talk, we will examine the computational and empirical foundations of this hypothesis, along with its cognitive and clinical implications.

We will explore computational models of recurrent neural networks that naturally evolve to operate near criticality, optimizing information representation while remaining on the verge of hallucinations. These models illustrate how critical systems exhibit key properties associated with conscious experience, such as increased entropy, heightened sensitivity to internal and external stimuli, and the emergence of rich, structured activity patterns.

Empirical evidence supporting critical brain dynamics, including critical slowing and neuronal avalanches, will be presented. We will discuss how approaching criticality may shift cortical networks into a generative mode and how the proximity to criticality is distributed across the cortex. Our findings suggest that proximity to criticality increases along processing hierarchies, peaking in regions associated with the Default Mode Network.

Finally, we will introduce the ConCrit framework, which posits that conscious experience is most likely to arise near criticality and diminish as the system deviates from this optimal state. This perspective provides a unifying approach to understanding consciousness through the lens of critical brain dynamics.