Global dynamics of neural mass models

Abstract

Neural mass models are used to simulate cortical dynamics and to explain the electrical and magnetic fields measured using electro- and magnetoencephalography. Simulations evince a complex phase-space structure for these kinds of models; including stationary points and limit cycles and the possibility for bifurcations and transitions among different modes of activity. This complexity allows neural mass models to describe the itinerant features of brain dynamics. However, expressive, nonlinear neural mass models are often difficult to fit to empirical data without additional simplifying assumptions: e.g., that the system can be modelled as linear perturbations around a fixed point. In this study we offer a mathematical analysis of neural mass models, specifically the canonical microcircuit model, providing analytical solutions describing slow changes in the type of cortical activity, i.e. dynamical itinerancy. We derive a perturbation analysis up to second order of the phase flow, together with adiabatic approximations. This allows us to describe amplitude modulations in a relatively simple mathematical format providing analytic proof-of-principle for the existence of semi-stable states of cortical dynamics at the scale of a cortical column. This work allows for model inversion of neural mass models, not only around fixed points, but over regions of phase space that encompass transitions among semi or multi-stable states of oscillatory activity. Crucially, these theoretical results speak to model inversion in the context of multiple semi-stable brain states, such as the transition between interictal, pre-ictal and ictal activity in epilepsy.

Topographic variation in neurotransmitter receptor densities explains differences in intracranial EEG spectra

“…The computationally efficient implementation within DCM allows to evaluate various hypotheses or explanations effectively. However, this comes at the expense of the Variational Laplacian model fitting approach being susceptible to locally — but not globally — optimal solutions (71, 72). In addition, since parameters collapse various underlying neuronal mechanisms, one-to-one reverse mapping requires elaborate testing procedures and might only provide approximate answers, potentially limiting predictive validity; for example, for explaining effects of pharmacological interventions without further data.…”

Linking fast and slow: The case for generative models

“…Several promising directions for developing these models are worth exploring. First, HMMs of EEG and MEG signals could be extended to reflect the itinerancy between the various attractors of the (multistable) system arising from interconnected cortical population, for instance, using the escape rates introduced in ( Cooray, Rosch, & Friston, 2023a , 2023b ). Such transformation would allow one to relate the switching statistics of the Markov chain to properties of the network, effectively uncovering mechanistic explanations for state transitions.…”

A comprehensive investigation of intracortical and corticothalamic models of the alpha rhythm

“…To construct models at the intended level of granularity, there are two main approaches: 1) a ‘bottom-up’ approach, beginning at the sub-cellular level with flows of ions and action potential generation at small patches of neuronal membrane (typically using Hodgkin-Huxley or Rall model equations), or at the whole-cell level (e.g. using Izhikevich or Leaky Integrate-and-Fire model equations); or 2) a ‘top-down’ approach, which represents the collective activity of neurons sharing some common characteristics, such as the type of synapses they connect to (excitatory or inhibitory) instead of focusing on individual cells (Cook et al, 2021; Cooray et al, 2023). While the former approach is a closer representation of biological neurons with finer details, it is often inadequate for modelling empirical phenomena emerging from large-scale brain activity, as the complexity rapidly increases with the number of neurons involved, resulting in interpretability and computational issues (Cook et al, 2021).…”