Huaizuo Zheng, Yifei Liu

(School of Information and Management, Guangxi Medical University)

Abstract：

The concepts and theories of neuroscience are rooted in different fields such as information theory, dynamical systems theory, and cognitive psychology. Not all of these concepts and theories can be organically connected, and some concepts cannot be directly compared with each other. Specific terminology within the field also poses obstacles to cross domain integration. However, the integration at the conceptual level can provide intuitive and consolidated forms of understanding, forming important guidance for the progress of neuroscience. This article integrates deterministic and stochastic dynamic processes within the framework of information theory, thereby linking information entropy and free energy with emergent dynamics and self-organization mechanisms in brain networks. We have identified the fundamental properties of the population of neurons that lead to equivariant matrices in the network, and their complex behavior can be naturally represented by structured flows on manifolds, thus establishing an intrinsic model of brain function theory. We used the brain network simulation platform The Virtual Brain to demonstrate how to translate these concepts into practical applications, and illustrated its use through examples of healthy aging and epilepsy.

Key Words：

virtual brain; brain network; connection group; emergence; structured flow on manifolds