Correlated Ensemble Solver Reveals Criticality and Hidden Attractor States in Neural Population Codes: Complete Analysis and Replication Guide
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
This paper presents a comprehensive analysis of neural population coding in the salamander retina using the Correlated Ensemble Solver (CES), a mathematical framework that computes exact joint firing probabilities with linear computational complexity. The analysis is applied to publicly available recordings of 160 retinal ganglion cells responding to natural movie stimuli across 283,041 time samples. The study establishes five principal findings. First, the CES substantially outperforms both independent and pairwise maximum entropy models in predicting synchronous firing statistics, achieving the lowest Bayesian Information Criterion among all models tested. Second, the retina exhibits signatures of critical dynamics, with avalanche size distributions following power-law statistics close to theoretical predictions for systems at criticality. Third, neural activity patterns follow Zipf's law, indicating near-optimal utilization of the available pattern space. Fourth, state entropy is maximized when the system operates near the critical point, providing direct evidence that the retina tunes its operating point to maximize information capacity. Fifth, population dynamics traverse seven discrete attractor states, with causal hub neurons coordinating activity across the population. The paper provides detailed methodology for all statistical analyses, comprehensive numerical results with uncertainty estimates, and complete replication instructions including parameter specifications.
Fields of Study & Subjects
Neural dynamics and brain function
Theoretical and Computational Physics
stochastic dynamics and bifurcation
Publication details
- Journal / source
- Zenodo (CERN European Organization for Nuclear Research)
- Publication date
- 30 March 2026
- Language
- EN
- Document type
- Preprint