Fig. 4: Computation modeling explains how stroke lesions prolong intrinsic neural timescales and alter network dynamics.

A Intrinsic neural timescale is shaped by the branching ratio (\(\sigma\)), indicating that stroke can shift the brain dynamics from a slightly subcritical state (blue) toward criticality (red), with the potential to enter a supercritical state. Near a phase transition, cortical network dynamics can be modeled as a branching process, where intrinsic neural timescales peak at the critical point¹⁵. B–G Similarly, complementary measures (see “Methods” for details)—including the Hurst exponent, \({{INT}}_{0.1}\), \({{INT}}_{0.5}\) \({{ACF}}_{0}\), \({{ACF}}_{0.1}\) and \({{ACF}}_{0.5}\)—consistently peak at the critical point, reflecting signs of critical slowing down, further demonstrating the robustness and convergence of our findings across multiple temporal metrics. This result is an average of 50 trials. Each dot represents the intrinsic neural timescale (INT) or temporal correlations (TCs) value obtained from the computational model at a given branching ratio (\(\sigma\)). These values were generated by varying the synaptic propagation probability (\({\rm{\lambda }}\)) while keeping network connectivity constant. All networks \(N=\mathrm{100,000}\) neurons and a mean degree \(K=10\) with varied values of \(\lambda\) to satisfy the relationship \({\rm{\sigma }}=K\cdot {\rm{\lambda }}\). The external driving is given by \(r={10}^{-5}\). See “Methods” for details.