Neuronal avalanches

Lens adapted to this topic: Evidence for avalanches and near-critical neural dynamics

The mainstream research account treats neuronal avalanches as reproducible cascades observed especially in cortical preparations, with statistics and branching behavior compatible with critical branching or near-critical dynamics. Researchers propose that such dynamics can support information transmission while avoiding runaway excitation, and may reflect transient cortical cell assemblies. The interpretation remains a hypothesis rather than an established description of all brain activity.

• Experiments in rat cortical cultures and acute slices found avalanche propagation with an event-size exponent near -3/2 and a branching parameter close to 1, as predicted by critical branching theory.
• Avalanche patterns in cortical slice cultures were diverse, precise, and stable over many hours, supporting the possibility that they contribute to information storage as well as propagation.
• Avalanche statistics and related patterns have been reported beyond slices, including in resting human MEG, although this does not by itself establish whole-brain criticality.
• The proposed functional rationale is that near-critical dynamics may optimize information transmission while preventing uncontrolled network excitation.
• Some reviews interpret avalanches as transiently formed cortical cell assemblies with organization across spatial scales.

Lens adapted to this topic: Challenges to interpreting avalanches as proof of criticality

Skeptical accounts accept that avalanche-like activity can be observed but question whether its statistical signatures establish self-organized criticality or a universal operating regime of the brain. They emphasize sensitivity to detection thresholds, surrogate data, finite sampling, and the possibility that dissipative neural models are subcritical or supercritical unless parameters are tuned. Some reviews instead propose that intact cortical dynamics may be reverberating and subcritical, while in-vitro networks may behave differently.

• Power-law fits from thresholded local-field-potential signals may be unstable to detection thresholds, and surrogate stochastic signals can show similar scaling without underlying self-organized criticality.
• A model with dissipation and progressive synaptic loading is not generically critical; without parameter tuning it can be subcritical or supercritical despite a broad pseudo-critical region.
• The criticality hypothesis has unresolved conceptual and experimental objections, including concerns about whether observed scaling uniquely identifies a critical state.
• One synthesis argues that in-vitro neuronal networks can self-organize near criticality while cortical dynamics in the intact brain may instead be reverberating and subcritical.
• The relation of avalanche-like activity observed in slices to physiological processes in the intact brain remains unknown.