Whole-brain modeling of dynamic causal circuits in human cognition using amortized variational inference
Understanding dynamic mechanisms underlying cognition remains a major challenge in human neuroscience. Here, we develop, validate, and apply Multivariate Dynamical Systems Identification with Amortized Variational Inference (MDSI-AVI), a computational framework designed to address critical challenges in capturing asymmetric, context-dependent, whole-brain directed interactions while accounting for regional hemodynamic response variability in fMRI data. MDSI-AVI leverages simulation-based inference through forward and reverse variational inference to address the limitations of conventional variational methods in high-dimensional settings. By averaging over uncertainty in hemodynamic response parameters using forward simulation, MDSI-AVI provides well-calibrated posteriors of directed connectivity that scale efficiently to networks with hundreds of nodes. Applied to Human Connectome Project data ( N = 728), MDSI-AVI provides insights into working memory mechanisms, identifying the dorsal anterior insula as a critical hub influencing activity at the whole-brain level. We demonstrate task-dependent modulation of causal influences, where the salience network drives frontoparietal network activity, which differentially influences the default mode and sensorimotor networks depending on working memory load. These whole-brain causal interactions distinguish task conditions with high accuracy and predict working memory performance. Our framework demonstrates reproducible results across whole-brain parcellations, establishing MDSI-AVI as a robust tool for advancing our understanding of circuit dynamics in cognition and disease.
Authors
- Linjing Jiang (ORCID: https://orcid.org/0009-0001-6344-7841)
- Byeongwook Lee (ORCID: https://orcid.org/0000-0002-9732-5295)
- Percy K. Mistry (ORCID: https://orcid.org/0000-0002-3055-4472)
- Leticia Levin Diniz
- Luca Ambrogioni (ORCID: https://orcid.org/0000-0003-3673-0825)
- Louis Rouillard (ORCID: https://orcid.org/0000-0002-8446-3509)
- Vinod Menon (ORCID: https://orcid.org/0000-0003-1622-9857)
- Srikanth Ryali (ORCID: https://orcid.org/0000-0001-9773-0381)
- Demián Wassermann (ORCID: https://orcid.org/0000-0001-5194-6056)
- Nicholas K. Branigan (ORCID: https://orcid.org/0009-0007-7305-5992)
- Weidong Cai (ORCID: https://orcid.org/0000-0001-9581-7774)
Institutions
- Université Paris-Sud (FR)
- École Polytechnique (FR)
- Radboud University Nijmegen (NL)
- Korea Advanced Institute of Science and Technology (KR)
- Centre Inria de Saclay (FR)
- Stanford Medicine (US)
- Donders Institute for Brain, Cognition and Behaviour (NL)
- Stanford University (US)
Publication Details
- Journal
- Nature Communications
- Published
- 2026-11-17
- DOI
- https://doi.org/10.1038/s41467-026-79293-z
- Primary Topic
- Functional Brain Connectivity Studies
- Type
- article
- Field-Weighted Citation Impact
- 0.00