Using Game Theory to Model Resting State Network Dynamics

The human brain possesses a remarkable ability to adapt its information processing to meet current goals. This ability involves activity throughout large-scale, specialized brain networks that support different functions. When the brain is at rest, network activity alternates between segregating computations to localized functional domains and integrating information across domains. Though readily observable, fluctuations in networks' tendencies to integrate and segregate information are not well understood. Given game theory's ability to describe cooperative and competitive behaviors of multiple agents, this study uses the mathematical framework to model interactions among seven large-scale brain networks. At each point in time, networks evaluate payoffs for adopting different sets of synchronization strategies, weighing both the rewards of increasing synchrony among member regions and the energetic costs of adopting these strategies. We asses the model's application to empirical data by extracting parameters from resting-state fMRI under both ketotic and glycolytic metabolic states. We examine strategies under these different metabolic states to determine whether more costly strategies are chosen when the brain has greater access to energy, and more generally determine whether the healthy brain operates at Nash equilibrium.