Temporal Coordination and Criticality in Human Neural Dynamics - Bridging Insights from rhythmicity, synchronization, and computational modeling

The neural processing of exogenous and endogenous information is organized into highly specialized anatomically distinct neural populations. These populations undergo rhythmic fluctuations in excitability, commonly referred to as neural oscillations. Neural oscillations have been studied for decades and are thought to be fundamental to the temporal coordination of information flow throughout the cortical processing hierarchy. However, after decades of research, data analysis methods for neural oscillations are still under active development. In the first study of this thesis, we extended the operationalization of oscillations beyond the power-based lens and developed a novel method that quantifies rhythmicity directly. The results showed that the actual rhythmicity of cortical oscillations is well delineated in both anatomical and spectral properties, with the mesoscale architecture organized into small frequency islands. However, no neuron operates independently of others. In the second study, we investigated long-range phase synchronization between high gamma frequencies in humans. We demonstrated that this synchronization has a modular architecture, segregates into healthy and epileptic zones, is characterized by laminar profiles, and is transiently enhanced and suppressed in separate frequency bands during a response inhibition task. Observables of neural dynamics vary between individuals and cognitive states. The critical brain hypothesis offers an explanation for this, and in Study III, we tested whether the level of variability is associated with the critical state of a neural system. We argue that in the human brain in vivo, a hypothetical critical point is stretched into a wider regime of critical-like dynamics—Griffith's phase—and that healthy brain regions operate on the subcritical and critical sides of this extended critical regime, while epileptogenic areas are located on the critical-supercritical side. Computational models have shown great utility in discovering the mechanistic principles of observed phenomena. In the fourth study, we developed a hierarchical model of critical-like oscillatory activity based on the Kuramoto model. We found that the model observables are physiologically plausible and most similar to real human recordings in the subcritical-critical range, demonstrating that the model can be fitted to reconstruct individual behavior with a high degree of similarity. Taken together, these studies provide new insights into the analysis of neural oscillations through complementary approaches, including connectivity in vivo, computational studies, phase-synchronization networks, and the criticality properties of individual nodes. In addition to extending the boundaries of known neural dynamics observables, we delved deeper into the Terra Incognita and found a way to directly quantify the rhythmicity of oscillations in alignment with the brain clock's timing. This may provide new insights into the analysis of information processing in the brain.