Skip to Main Content
Psychiatry
Make a Gift
YSM Home

INFORMATION FOR

Modeling Whole-Brain Dynamics: Turbulence as a framework for brain dynamics in health and disease

Dr. Gustavo Deco, Ph.D., Research Professor at the Institució Catalana de Recerca i Estudis Avançats (ICREA), Professor (Catedrático) at the Pompeu Fabra University (UPF), the Computational Neuroscience group.

Watch:

Presentation Slides:

This session includes 4 parts:

  1. An overview of Modeling Whole-Brain Dynamics approaches. Gustavo Deco. 10 am – 11 am: zoom link
  2. Practical tutorial for Dynamic Mean Field (DMF) / fast DMF toolbox. Leonardo Saraiva. 12 pm – 1 pm: zoom link
  3. Practical tutorial for Turbulence toolbox. Leonardo Saraiva. 1:30 pm – 2:30 pm: zoom link
  4. Q&A: practical considerations. Leonardo Saraiva. 2:30 pm: zoom link

What?

This seminar introduces whole-brain computational modeling and turbulent brain dynamics. Using models built from coupled oscillators constrained by structural connectivity, it demonstrates how large-scale brain activity can be simulated and used to reproduce empirical neural dynamics.

A key focus is the observation that brain activity operates in a turbulent regime, characterized by complex, multi-scale fluctuations. The turbulence framework provides a way for understanding how information flows efficiently across the brain and to quantify patterns of dynamic organization beyond static connectivity measures.

By combining whole-brain modeling with turbulence analysis, this approach enables the study of large-scale brain dynamics across both health and disease. It has been applied to conditions including disorders of consciousness, depression, Alzheimer’s disease, traumatic brain injury, and stroke recovery, as well as to the effects of interventions such as psychedelics and deep brain stimulation.

The framework can also be applied at the individual level, where models are fit to individual patient data to investigate and predict clinically relevant outcomes, including treatment responsiveness.

Two complementary toolboxes are introduced:

  • Dynamic Mean Field (DMF) / fast DMF toolbox: A whole-brain modeling framework that simulates neural activity across regions using biologically informed population models, constrained by structural connectivity. It allows researchers to generate and fit time-resolved brain activity and to study how changes in local circuit parameters influence large-scale dynamics.
  • The Turbulent Brain toolbox: A set of tools for quantifying the dynamical organization of whole-brain activity. Using concepts from turbulence, it characterizes multi-scale fluctuations, synchronization patterns, and the flow of activity across the brain, providing summary measures of how brain dynamics are organized in health and disease.

Why?

Whole-brain modeling combined with turbulence analysis provides a biophysically grounded framework for studying brain dynamics. Unlike purely statistical or data-driven methods, it offers mechanistic insight into how large-scale brain activity emerges from underlying interactions.

The turbulence framework captures the multi-scale, scale-free structure of brain dynamics, going beyond simpler measures such as correlations or graph-based summaries. This allows for a more complete characterization of how information is organized and transferred across the brain. An additional strength is the ability to perform in silico perturbation experiments, enabling researchers to simulate and predict the effects of interventions such as brain stimulation or pharmacological treatments.

The approach has been very well received, with key papers published in respected journals including Nature Reviews Neuroscience, Nature Human Behaviour, Cell Reports, PNAS, Neuron, Science Advances, Current Biology, and Nature Communications. There is growing interest from both the computational and clinical neuroscience communities.

How?

Prerequisites:

  • Basic knowledge of dynamical systems, differential equations, and nonlinear dynamics is helpful
  • Familiarity with neuroimaging data (e.g., fMRI, MEG) and structural connectivity (diffusion MRI)
  • Programming in Python or MATLAB

Requirements:

  • A standard computer is sufficient for basic whole-brain simulations.
  • High-performance computing (HPC) clusters are useful for large parameter sweeps in model training/estimation.

Data:

  • Resting-state fMRI combined with diffusion MRI tractography
  • Common datasets include Human Connectome Project (HCP), UK Biobank, ADNI, and OpenNeuro
  • Any dataset with both structural and functional connectivity can be used

Tools:

Workflow:

  • Construct a whole-brain model using structural connectivity
  • Simulate neural dynamics using coupled oscillator systems
  • Fit model parameters to empirical data
  • Quantify dynamical properties (e.g., turbulence, synchronization, variability)
  • Perform in silico perturbations to study effects of interventions

Tutorials and resources:

  • FastDMF documentation and implementation: Herzog, et al. "Neural mass modeling for the masses: Democratizing access to whole-brain biophysical modeling with FastDMF." Network Neuroscience 8.4 (2024): 1590-1612. https://doi.org/10.1162/netn_a_00410
  • Turbulent Brain Toolbox: Perl, Yonatan Sanz, et al. "The Turbulent Brain: a toolbox to compute the turbulent dynamical behaviour of whole-brain activity." bioRxiv (2025): 2025-11. https://doi.org/10.1101/2025.11.14.688068
  • The Virtual Brain tutorials and webinars: https://www.thevirtualbrain.org

Reference publications: