Irreversibility as a signature of non-equilibrium phase transition in large-scale human brain networks: An fMRI study

Abstract It has been argued that the human brain, as an information-processing machine, operates near a phase transition point in a non-equilibrium state, where it violates detailed balance leading to entropy production. Thus, the assessment of irreversibility in brain networks can provide valuable insights into their non-equilibrium properties. In this study, we utilized an open-source whole-brain functional magnetic resonance imaging (fMRI) dataset from both resting and task states to evaluate the irreversibility of large-scale human brain networks. Our analysis revealed that the brain networks exhibited significant irreversibility, violating detailed balance, and generating entropy. Notably, both physical and cognitive tasks increased the extent of this violation compared to the resting state. Regardless of the state (rest or task), interactions between pairs of brain regions were the primary contributors to this irreversibility. Moreover, we observed that as global synchrony increased within brain networks, so did irreversibility. The first derivative of irreversibility with respect to synchronization peaked near the phase transition point, characterized by the moderate mean synchronization and maximized synchronization entropy of blood oxygenation level-dependent (BOLD) signals. These findings deepen our understanding of the non-equilibrium dynamics of large-scale brain networks, particularly in relation to their phase transition behaviors, and may have potential clinical applications for brain disorders.