Schedule of Events | Symposia

Sketchpad Series

Individualized models connect nontrivial whole-brain dynamics across rest and task conditions

Poster Session D - Monday, March 31, 2025, 8:00 – 10:00 am EDT, Back Bay Ballroom/Republic Ballroom

Ruiqi Chen¹ (chen.ruiqi@wustl.edu), Matthew Singh², Todd Braver¹, ShiNung Ching¹; ¹Washington University in St. Louis, ²University of Illinois Urbana-Champaign

It remains debated whether whole brain dynamics reflect distinct brain “states” or noisy sampling of a monolithic state. Resting state fMRI (rfMRI) is often characterized as switching between multiple discrete states. In the N-back working memory task, single-trial neural responses were also found to be best described by multiple states. However, the mechanistic underpinnings and behavioral relevance of these nontrivial brain states remain elusive. We fit Mesoscopic Individualized NeuroDynamic (MINDy) models for rfMRI and N-back task fMRI (tfMRI) recordings in the Human Connectome Project. MINDy models consist of interconnected neural masses representing brain parcels and task-related inputs. We analyzed the “resting” and “task state” dynamics of fitted models through numerical simulations. MINDy models fit rfMRI and tfMRI data significantly better than linear null models and recapitulated key features such as functional connectivity. The models revealed nontrivial (e.g., multistable) dynamics in both resting and task states. When projected into anatomical space, the dynamical attractors mapped onto a limited set of canonical functional brain networks. We found that participants that showed a “bifurcation” between resting and task state (e.g., multistable in rest, monostable in task) performed significantly better in task. Next, we will analyze single-trial neural responses according to different attractors and associate them to trial-by-trial behavioral fluctuations. Ongoing brain dynamics contain nontrivial attractors embedded in functional brain networks, which might account for trial-by-trial variations in neural responses. The ability to modulate brain dynamics towards cognitive demand might be beneficial to cognitive performance.

Topic Area: METHODS: Neuroimaging