Self-organized criticality in neural networks from activity-based rewiring

Area of Science:

• Computational Neuroscience
• Complex Systems Theory
• Network Science

Background:

• Neural systems operate in a dynamic regime between quiescence and chaotic activity.
• The criticality hypothesis posits that neural systems self-organize towards a critical point for optimal information processing.

Purpose of the Study:

• To investigate a minimal neural network model exhibiting self-organized criticality.
• To explore the role of local information and stochastic noise in achieving critical dynamics.

Main Methods:

• Development of a minimal neural network model with a local information-based rewiring rule.
• Analysis of network evolution under stochastic noise and activity-dependent link dynamics.
• Examination of avalanche statistics and power-law distributions.

Main Results:

• The model autonomously evolves towards a critical state without parameter tuning.
• Network dynamics exhibit power-law distributed avalanche statistics consistent with criticality.
• Observed exponents align with dynamical scaling theory and experimental neural avalanche data.

Conclusions:

• Local information and stochastic noise are sufficient for self-organized criticality in neural networks.
• This mechanism provides a plausible explanation for how real neural systems achieve criticality, particularly during development.
• The model's robustness suggests a fundamental principle for neural information processing.