Related Experiment Video
Updated: Jan 4, 2026
08:31
Concurrent Recording of Co-localized Electroencephalography and Local Field Potential in Rodent
Published on: November 30, 2017
12.8K
Neural field models for latent state inference: Application to large-scale neuronal recordings
Michael E Rule1, David Schnoerr2, Matthias H Hennig3
1Department of Engineering, University of Cambridge, Cambridge, United Kingdom.
Plos Computational Biology
|November 5, 2019
Summary
We developed a new statistical method to analyze large neural recordings, inferring neuron states from spiking activity. This approach reveals the crucial role of refractory periods in shaping neural population dynamics, particularly in the developing retina.
Area of Science:
- Computational Neuroscience
- Systems Neuroscience
- Statistical Modeling
Background:
- Large-scale neural recordings offer unprecedented insights into neural population dynamics.
- Extracting theories of collective dynamics from population data presents significant statistical challenges.
- Neural field models are established tools for mathematical population modeling but require statistical inference methods for large datasets.
Purpose of the Study:
- To develop a statistical inference method for interpreting neural field models using large-scale neural recordings.
- To enable the inference of intrinsic neuronal states (e.g., active, refractory) directly from spiking activity.
- To connect mathematical neural theory with point-process state-space models for neural data analysis.
Main Methods:
- A novel moment closure technique was employed to interpret neural field models as latent state-space point-process models.
- This method facilitates statistical inference, allowing for the estimation of unobserved neuronal states.
- The approach was validated using synthetic data and subsequently applied to high-density recordings from the developing mouse retina.
Main Results:
- The method successfully infers intrinsic neuronal states, such as active and refractory periods, from population spiking data.
- Application to neonatal mouse retinal waves confirmed the significant impact of long-lasting refractory states on spatiotemporal dynamics.
- The study demonstrates the utility of the developed method in analyzing complex neural population activity.
Conclusions:
- This work introduces a powerful conceptual and methodological advance for analyzing large-scale neural recordings.
- The developed method bridges mathematical neural modeling with statistical inference techniques for point-process data.
- It highlights the critical role of neuronal refractory states in shaping emergent neural population dynamics.
