Emergent complexity and rhythms in evoked and spontaneous dynamics of human whole-brain models after tuning through analysis tools

A multi-institutional research team has published a significant advance in computational neuroscience, demonstrating how to properly tune whole-brain models to produce biologically realistic dynamics. The team, led by Gianluca Gaglioti and including 12 other authors, integrated two established tools: The Virtual Brain (TVB) platform for simulating whole-brain dynamics and the Collaborative Brain Wave Analysis Pipeline (Cobrawap) for analyzing outputs with standardized metrics. They applied this framework to a 998-node human connectome using the Larter-Breakspear neural mass model, comparing TVB's default parameters against a tuned configuration.

The results, published in Neurocomputing, reveal that the tuned model exhibits several critical features absent in the default configuration. During spontaneous activity, it shows robust alpha-band oscillations (8-12 Hz), infra-slow rhythms, scale-free characteristics, greater spatio-temporal heterogeneity, and asymmetric functional connectivity. When externally perturbed, the tuned model generates non-stereotyped, complex spatio-temporal activity as measured by the perturbational complexity index—a metric used in consciousness research. This represents a substantial improvement over previous models that often produced overly simplified or biologically implausible dynamics.

This work establishes a methodological foundation for data-driven calibration of accurate whole-brain models. By providing a systematic approach to parameter tuning using complementary analysis tools, researchers can now create simulations that better approximate real brain dynamics across multiple scales. The framework enables quantitative validation of models against experimental data, moving beyond qualitative comparisons to more rigorous scientific standards in computational neuroscience.