A reduction in energy costs induces integrated states of brain dynamics.

In the human brain, interactions between multiple regions organize stable dynamics that enable enhanced cognitive processes. However, it is unclear how collective activities in the brain network can generate stable states while preserving unity across the whole brain scale under successive environmental changes. Herein, a network model was introduced in which network connections were adjusted to reduce the energy consumption level by avoiding excess changes in the activated states of each region during successive interactions. For time series data obtained from fMRI images, a connection matrix was generated by a simulation, and the predictions made by this matrix yielded accurate results relative to the real data. In this simulation, the adjustment process was activity-dependent, in which the interregional connections between intense active regions were reinforced to prohibit free behaviours. This resulted in a reduced excess energy loss and the integration of multiple regional activities into integrated dynamic states under constraints imposed by other regions. It was suggested that the simple rule of saving excess energy costs plays an important role in the mechanism that regulates large-scale brain networks and dynamics.