Sensory-motor cortices shape functional connectivity dynamics in the human brain

Abstract

Large-scale biophysical circuit models provide mechanistic insights into the micro-scale and macro-scale properties of brain organization that shape complex patterns of spontaneous brain activity. We developed a spatially heterogeneous large-scale dynamical circuit model that allowed for variation in local synaptic properties across the human cortex. Here we show that parameterizing local circuit properties with both anatomical and functional gradients generates more realistic static and dynamic resting-state functional connectivity (FC). Furthermore, empirical and simulated FC dynamics demonstrates remarkably similar sharp transitions in FC patterns, suggesting the existence of multiple attractors. Time-varying regional fMRI amplitude may track multi-stability in FC dynamics. Causal manipulation of the large-scale circuit model suggests that sensory-motor regions are a driver of FC dynamics. Finally, the spatial distribution of sensory-motor drivers matches the principal gradient of gene expression that encompasses certain interneuron classes, suggesting that heterogeneity in excitation-inhibition balance might shape multi-stability in FC dynamics.

The impact of aging on human brain network target controllability

“…From a functional perspective the high control centrality of the sensorimotor system has been previously associated with its ability to process information not only for motor control but also for a broad range of recognition processes [41][42][43]. This is in line with the existence of gradients of cortical organization through which the sensorimotor system could boost neural activity of other association areas required for higher order functions such as cognitive control, guided attention and motivation [44].…”

Extracting Dynamical Understanding From Neural-Mass Models of Mouse Cortex

“…A common scientific goal of modeling a system is to accurately reproduce important properties of it, while also gaining an understanding of how it does so. While successful approaches have been demonstrated for maximizing goodness of fit [sometimes optimizing large numbers of parameters (Wang et al, 2019 ; Kong et al, 2021 ; Wischnewski et al, 2021 )], obtaining understanding is a key challenge for complex nonlinear models of brain dynamics. The analyses and visualizations demonstrated in this work aim to provide an understanding of the model dynamics in terms of the dynamical regimes that individual regions can access, shaped by their inputs from coupled neighbors.…”

“…An early example is the work of Chaudhuri et al ( 2015 ), which incorporated a variation in recurrent excitation corresponding to that of measured spine count in the macaque. More recent work in human has incorporated spatial heterogeneity in model parameters with: the MRI-derived T1w:T2w map (Demirtas et al, 2019 ); T1w:T2w, the first principal component of gene transcription, and an inferred excitation:inhibition ratio (Deco et al, 2021 ); a linear combination of T1w:T2w and the principal resting-state functional connectivity (FC) gradient (Kong et al, 2021 ); a fitted parametric variation that recapitulated an interpretable hierarchical variation (Wang et al, 2019 ); and a spatial variation in excitability with a spatial map of epileptogenicity in modeling seizure dynamics and spread (Jirsa et al, 2017 ; Courtiol et al, 2020 ). These papers have reported improved out-of-sample model fits to empirical data, evaluated according to a range of summary statistics of the resulting dynamics (most typically FC), and provided insights into how spatial variation in biological mechanisms (like recurrent excitation) may underpin whole-brain dynamical regimes.…”

Parameter inference on brain network models with unknown node dynamics and spatial heterogeneity

“…Cortical gradients in large scale brain modeling. Gradients of structure, connectivity, gene expression, and function across the human cortex raised considerable interest in recent year (28,29), and several modeling studies investigated their role in the large scale brain dynamics (3)(4)(5)(6). Our approach shares with them the basic principles of large-scale brain modeling based on the structural connectome, but differs in three key points: First, we do not rely on any particular neural mass model, rather, the model is derived from data.…”

“…In recent years, however, several works utilized the whole-brain modeling framework in order to explore the role of spatial heterogeneity of model parameters. Specifically, the studies found that the whole-brain models can better reproduce the features of resting-state fMRI when the regional variability is constrained by the MRI-derived estimates of intracortical myelin content (3), functional gradient (4), or gene expression profiles (5), and similar regional variability was found even without prior restrictions (6).…”