Criticality in the cortex: what has changed
For two decades the concept that cortex sits near a critical point — a state known from physics where systems show scale-free fluctuations and maximal responsiveness — has attracted intense interest. The signature observation is the neural avalanche: cascades of spiking or local-field activity whose sizes and durations follow heavy-tailed statistics suggestive of power laws. Early work using cortical slices and microelectrode arrays established avalanche statistics in vitro and in vivo, and theoretical work proposed that such self-organized criticality (SOC) could confer computational benefits.
New convergence and refined nuance
Recent experimental and analytical advances have softened a binary view of 'critical versus not.' Large-scale recordings, calcium imaging, and careful subsampling analyses now show that cortical dynamics hover near criticality rather than sitting exactly at an idealized critical point. Studies combining multi-modal data and rigorous statistical tests have supported a picture in which the brain tunes itself to a near-critical regime that balances sensitivity and stability.
Why near-criticality might matter
- Dynamic range and sensitivity: Near a critical point, small inputs can produce large, graded responses, increasing the range of stimuli the network can represent.
- Information transmission and storage: Correlations and long-range temporal structure near criticality can boost information capacity and memory lifetimes without runaway excitation.
- Adaptability: Operating near a transition facilitates rapid switches between modes of processing via modest neuromodulatory or inhibitory changes.
Controversies and methodological pitfalls
Not everyone accepts that observed avalanches imply SOC. Critics point out several confounds: limited sampling can spuriously produce apparent power laws; common drive or filtering of Poisson processes can mimic heavy tails; and experimental thresholds and binning influence statistics. The field has responded with more rigorous null models, subsampling corrections, and alternative metrics (for example, branching ratio estimates and scaling collapse tests) that can better distinguish true critical scaling from lookalikes.
Where the debate is active now
Current work focuses on causality and control. Optogenetic perturbations that shift network excitability, or pharmacological manipulations that alter inhibition, are being used to test whether moving the network away from the inferred operating point degrades computational metrics predicted by theory. Parallel theoretical progress explores how homeostatic plasticity, inhibitory feedback, and neuromodulation can stabilize a near-critical regime without fine tuning.
Implications and next steps
The implication is pragmatic: cortex may exploit the benefits of criticality while avoiding the fragility of an exact phase transition. That view reframes criticality as a regulated operating regime rather than a singular property. Important open directions include mapping behavior-dependent shifts in criticality, linking critical metrics to perceptual and cognitive performance, and characterizing how disease states perturb the balance of excitation and inhibition that keeps cortex near the edge.
In short, neural avalanches remain a powerful lens on cortical computation: not a settled verdict but a live program of experiments and theory probing how brains tune themselves for flexible, efficient processing.
