Zang & Curto: Chaotic oscillations model brain's metastable states in new paper

In a new preprint, researchers Jie Zang and Carina Curto from the neuroscience community present a novel dynamical systems framework for understanding sequential metastability in brain networks. Their paper, Sequential chaotic oscillations in excitatory-inhibitory threshold-linear networks, introduces sequential chaotic oscillations (SCOs) as a candidate mechanism for the ordered transitions between metastable states observed in healthy brain function. The model operates under constant input and relies on excitatory-inhibitory threshold-linear networks (E-I TLNs), a simplified neural architecture. Crucially, the researchers discovered that the order of state transitions can be predicted by the underlying graph structure of the network, linking network topology directly to dynamic behavior.

The team identified specific parameter regimes required for SCOs: the presence of unstable singleton fixed points and sufficiently strong inhibition. They developed new graph rules for E-I TLNs and characterized fixed point structures on path and cycle networks. Additionally, they found that excitatory-inhibitory oscillations need not be synchronized, challenging common assumptions. To address this, they introduced a decomposition into z-mode (capturing excitatory differences) and mean mode (overall network activity), enabling better distinction of attractors for full-support fixed points on cycles. The work, spanning 37 pages with 12 figures, provides a rigorous mathematical foundation for chaotic itinerancy—a simple form of chaotic dynamics—in neural systems. This bridges theoretical neuroscience with observable brain dynamics, offering testable predictions for how neural circuits generate sequential activity patterns.