Biology subjects

Ryali, S.

Publications and source records attributed to Ryali, S..

4 recordsLinked to original sources

bioRxiv2026

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 novel 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 reveals new 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.

neuroscience↗

bioRxiv2025

Dynamic brain states during encoding and their post-encoding reinstatement predicts episodic memory in children

Episodic memory, the ability to rapidly learn and explicitly remember past events and experiences, plays a critical role in childrens academic learning and knowledge acquisition. The formation of lasting memories relies on the brains ability to dynamically organize its activity. How these neural configurations unfold moment-by-moment across encoding and offline phases remains poorly understood. To probe these dynamics, we applied a novel Bayesian Switching Dynamic Systems approach, a hidden Markov model with automatic state detection, to fMRI data from children performing scene encoding followed by an offline post-encoding rest. We identified four distinct brain states during encoding with unique activation modes between visual, medial temporal lobe, and frontoparietal and default mode network nodes. An "active-encoding" state with integrated visual-hippocampal and frontoparietal activity dominated encoding and predicted individual memory performance, while an inactive state negatively predicted performance. State transition dynamics revealed that flexible shifts into the active-encoding state enhanced memory formation, whereas transitions toward inactive states impaired it, demonstrating that memory success depends on dynamic neural flexibility. Critically, encoding states spontaneously reemerged during post-encoding rest. A "default-mode" state characterized by enhanced default mode network activity showed sustained maintenance during rest and robustly predicted memory outcomes, an effect specific to post-encoding, not pre-encoding rest. These findings establish that episodic memory emerges from coordinated brain state sequences bridging online encoding with offline consolidation, providing a computational framework for how moment-to-moment neural dynamics support memory formation in children. This work has broad implications for optimizing educational interventions and understanding developmental disorders affecting learning and memory.

neuroscience↗

bioRxiv2025

Explainable AI techniques for dynamic functional brain imaging: validation and analysis of E/I imbalance in autism

Deep neural networks have revolutionized functional neuroimaging analysis but remain "black boxes," concealing which brain mechanisms and regions drive their predictions--a critical limitation for clinical neuroscience. Here we develop and validate an explainable AI (xAI) framework to test whether feature attribution techniques can reliably recover brain regions affected by excitation/inhibition (E/I) imbalance--a fundamental dysregulation implicated in autism, schizophrenia, and other neuropsychiatric disorders. We employed complementary simulation approaches: recurrent neural networks for controlled parameter exploration, and The Virtual Brain simulator incorporating empirically-derived human and mouse connectomes to model E/I balance alterations with unprecedented biological realism. Through systematic validation, we demonstrate that Integrated Gradients and DeepLIFT methods reliably identify brain regions affected by E/I imbalance across challenging conditions, including high noise, low prevalence, and subtle neurophysiological alterations. This performance remains robust across species and anatomical scales, from 68-region human to 426-region mouse connectomes. Application to the ABIDE autism dataset (N=834) reveal convergence between our biophysically grounded simulations and empirical findings, providing computational support for E/I imbalance mechanisms in autism. This work establishes essential tools and data for interpretation of deep learning models in functional neuroimaging, and enables hypothesis-driven analysis of cellular mechanisms across neuropsychiatric disorders.

neuroscience↗

bioRxiv2022

Neuronal dynamics of the default mode network and anterior insular cortex: Intrinsic properties and modulation by salient stimuli

The default mode network (DMN) is closely associated with self-referential mental functions and its dysfunction is implicated in many neuropsychiatric disorders. However, the neurophysiological properties and task-based functional organization of the rodent DMN are poorly understood, limiting its translational utility. Here, we combine fiber-photometry with fMRI and computational modeling to characterize dynamics of putative rodent DMN nodes and their interactions with the anterior insular cortex (AI) of the salience network. We reveal neuronal activity changes in AI and DMN nodes prior to fMRI-derived DMN activations and uncover cyclical transition patterns between spatiotemporal neuronal activity states. Finally, we demonstrate that salient oddball stimuli suppress the DMN and enhance AI neuronal activity, and that the AI causally inhibits the retrosplenial cortex, a prominent DMN node. These findings elucidate previously unknown properties regarding the neurobiological foundations of the rodent DMN and its modulation by salient stimuli, paving the way for future translational studies. HighlightsO_LIConcurrent measurement of neuronal (GCaMP) and fMRI signals in retrosplenial, cingulate, prelimbic, and anterior insula cortices C_LIO_LIGCaMP signals reveal neuronal antagonism between AI and fMRI-derived DMN activation and deactivation C_LIO_LIGCaMP signals reveal salient oddball stimuli-induced suppression of prelimbic, cingulate and retrosplenial cortices, and activation of anterior insular cortex C_LIO_LIAnterior insular cortex causally inhibits retrosplenial cortex during processing of salient oddball stimuli C_LIO_LIFindings delineate neurofunctional organization of the rodent DMN and provide a more informed model for translational studies C_LI

neuroscience↗