Abstract:Neural mass models such as the Jansen-Rit system provide a practical framework for representing and interpreting electrophysiological activity (1-6) in both local and global brain models (7). However, they are only partly derived from first principles. While the post-synaptic potential dynamics in NMM are inferred from data and can be grounded on diffusion physics (8-10), Freeman's wave to pulse sigmoid function (11-13), used to transduce mean population membrane potential into firing rate, rests on a weaker theoretical standing. On the other hand, Montbrio et al (14, 15) derive an exact mean-field theory (MPR) from a quadratic integrate and fire neuron model under some simplifying assumptions, thereby connecting microscale neural mechanisms and meso/macroscopic phenomena. The MPR model can be seen to replace Freeman's sigmoid function with a pair of differential equations for the mean membrane potential and firing rate variables - a dynamical relation between firing rate and membrane potential (firing rate potential relation or FPR)-, providing a mechanistic interpretation of NMM semi-empirical sigmoid parameters. In doing so, we show it sheds light on the mechanisms behind enhanced network response to weak but uniform perturbations. For example, in the exact mean-field theory, intrinsic population connectivity modulates the steady-state firing rate FPR in a monotonic manner, with increasing self-connectivity leading to higher firing rates. This provides a plausible mechanism for the enhanced response of densely connected networks to weak, uniform inputs such as the electric fields produced by non-invasive brain stimulation. The new, dynamic FPR also endows the neural mass model with a form of inertia, an intrinsic delay to external inputs that depends on, e.g., self-coupling strength and state of the system. Next, we complete the MPR model by adding the second-order equations for delayed post-synaptic currents, bringing together the MPR and the usual NMM formalisms into a unified exact mean-field theory (NMM2) displaying rich dynamical features. As an example, we analyze the dynamics of a single population model, and a model of two coupled populations with a simple excitation-inhibition (E-I) architecture, showing it displays rich dynamics with limit cycles, period doubling, bursting behavior, and enhanced sensitivity to external inputs.

Dynamics of Moment Neuronal Networks

Moment Neuronal Networks: Stochastic Computation in Neuronal Systems

Moment neural network and an efficient numerical method for modeling irregular spiking activity

On Gaussian Random Neuronal Field Model: Moment Neuronal Network Approach

On a Gaussian Neuronal Field Model

From spiking neuronal networks to interpretable dynamics: a diffusion-approximation framework

Analysis and extension of exact mean-field theory with dynamic synaptic currents

Numerical Methods for Solving Moment Equations in Kinetic Theory of Neuronal Network Dynamics

A Coarse-Grained Framework for Spiking Neuronal Networks: Between Homogeneity and Synchrony

Mean-field approximations of networks of spiking neurons with short-term synaptic plasticity

KINETIC THEORY FOR NEURONAL NETWORK DYNAMICS

The determining role of covariances in large networks of stochastic neurons

Dynamics of stochastic integrate-and-fire networks

Neuronal Synfire Chain via Moment Neuronal Network Approach

Minimizing information loss reduces spiking neuronal networks to differential equations

Estimation of Spike Train Statistics in Spontaneously Active Biological Neural Networks

Toward stochastic neural computing

Dynamical mean-field theory of spiking neuron ensembles: response to a single spike with independent noises

How Can Anomalous-Diffusion Neural Networks Under Connectomics Generate Optimized Spatiotemporal Dynamics

Uncertainty Quantification in Working Memory via Moment Neural Networks