An Evolutionary Game Theory Model of Spontaneous Brain Functioning

Dario Madeo,Agostino Talarico,Chiara Mocenni,Alvaro Pascual-Leone,Emiliano Santarnecchi

doi:10.1038/s41598-017-15865-w

Abstract

Our brain is a complex system of interconnected regions spontaneously organized into distinct networks. The integration of information between and within these networks is a continuous process that can be observed even when the brain is at rest, i.e. not engaged in any particular task. Moreover, such spontaneous dynamics show predictive value over individual cognitive profile and constitute a potential marker in neurological and psychiatric conditions, making its understanding of fundamental importance in modern neuroscience. Here we present a theoretical and mathematical model based on an extension of evolutionary game theory on networks (EGN), able to capture brain's interregional dynamics by balancing emulative and non-emulative attitudes among brain regions. This results in the net behavior of nodes composing resting-state networks identified using functional magnetic resonance imaging (fMRI), determining their moment-to-moment level of activation and inhibition as expressed by positive and negative shifts in BOLD fMRI signal. By spontaneously generating low-frequency oscillatory behaviors, the EGN model is able to mimic functional connectivity dynamics, approximate fMRI time series on the basis of initial subset of available data, as well as simulate the impact of network lesions and provide evidence of compensation mechanisms across networks. Results suggest evolutionary game theory on networks as a new potential framework for the understanding of human brain network dynamics.

Highlights

The principles determining the moment-to-moment local vascular/metabolic supply giving rise to the functional connectivity patterns observed via functional magnetic resonance imaging in the primate and non-primate brain are still under discussion
The analysis of spontaneous and evoked activity recorded using blood oxygenation level dependent (BOLD) signal is mostly based on functional connectivity - inferred by means of linear (e.g Pearson correlation, partial correlation) and nonlinear tools27,28 - and effective connectivity (e.g. Granger causality)[29] metrics to provide a quantification of interregional interplay, while neglecting brain’s non-linear dynamics and stationarity of time series[30]
Statistical mechanics has been used to describe the activity of nervous nets[41] by using “replicator equations”, a mathematical tool widely used in the context of evolutionary game theory (EGT)[42,43,44]

Summary

Introduction

The principles determining the moment-to-moment local vascular/metabolic supply giving rise to the functional connectivity patterns observed via functional magnetic resonance imaging (fMRI) in the primate and non-primate brain are still under discussion. The analysis of spontaneous and evoked activity recorded using blood oxygenation level dependent (BOLD) signal is mostly based on functional connectivity - inferred by means of linear (e.g Pearson correlation, partial correlation) and nonlinear tools (e.g. mutual infomation)27,28 - and effective connectivity (e.g. Granger causality)[29] metrics to provide a quantification of interregional interplay, while neglecting brain’s non-linear dynamics and stationarity of time series[30] This might prevent the testing and identification of organizational principles underlying brain’s spontaneous functioning, as those commonly applied to the study of complex network dynamics. The following sections will introduce the rationale and specifics of the EGN model for fMRI BOLD data (Evolutionary Games for Brain Networks - EGN-B hereafter), where each brain region, assumed as an assembly of neurons composing anatomically or functionally defined regions, is modeled as a player in an evolutionary game Under this assumption, we derived a model that incorporates the mechanisms of cooperation and competition among network nodes. Additional details about the model, neuroimaging datasets and fMRI preprocessing are included in Supplementary Information

Methods

Results

Conclusion