Programa del congreso

Whole-brain models provide a valuable framework to investigate the mechanisms underlying large-scale brain dynamics. In this work, we introduce methodological and modeling contributions that enhance their performance and interpretability. First, we implement the Heun integration scheme in neurolib, achieving significantly better alignment with empirical functional connectivity compared to the classical Euler method, as measured by a geodesic distance on symmetric positive definite matrices. Second, we develop a fast multimodel framework, 600× faster than the original one, enabling the coupling of cortical and thalamic models at scale. Using this framework, we compare a phenomenological corticothalamic model (Hopf supercritical in cortex, subcritical in thalamus) with a biophysical corticothalamic model (ALN–Costa). Results show that ALN–Costa reproduces empirical functional connectivity more

accurately. We identify the main limitation of Hopf in this context: the Balloon–Windkessel model only receives the real part of the oscillatory signal, oversimplifying its dynamics and yielding flattened Fmats. Nevertheless, propagation experiments reveal that Hopf has strong potential to describe how oscillations spread across brain networks, with coupling strength emerging as the key determinant

of dynamical diffusion.