Causally informed activity flow models provide mechanistic insight into network-generated cognitive activations DOI Creative Commons
Rubén Sánchez-Romero, Takuya Ito, Ravi D. Mill

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2021, Volume and Issue: unknown

Published: April 18, 2021

Abstract Brain activity flow models estimate the movement of task-evoked over brain connections to help explain network-generated task functionality. Activity have been shown accurately generate activations across a wide variety regions and conditions. However, these had limited explanatory power, given known issues with causal interpretations standard functional connectivity measures used parameterize models. We show here that functional/effective (FC) grounded in principles facilitate mechanistic interpretation progress from simple complex FC measures, each adding algorithmic details reflecting principles. This reflects many neuroscientists’ preference for reduced measure complexity (to minimize assumptions, compute time, fully comprehend easily communicate methodological details), which potentially trades off validity. start Pearson correlation (the current field standard) remain maximally relevant field, estimating validity range using simulations empirical fMRI data. Finally, we apply causal-FC-based modeling dorsolateral prefrontal cortex region (DLPFC), demonstrating distributed network mechanisms contributing its strong activation during working memory task. Notably, this model is able account DLPFC effects traditionally thought rely primarily on within-region (i.e., not distributed) recurrent processes. Together, results reveal promise parameterizing methods identify underlying cognitive computations human brain. Highlights - provide insight into how neurocognitive are generated interactions. Functional statistical Mechanistic predict neural tasks.

Language: Английский

Regularized partial correlation provides reliable functional connectivity estimates while correcting for widespread confounding DOI Creative Commons
Kirsten L. Peterson, Rubén Sánchez-Romero, Ravi D. Mill

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2023, Volume and Issue: unknown

Published: Sept. 17, 2023

Abstract Functional connectivity (FC) has been invaluable for understanding the brain’s communication network, with strong potential enhanced FC approaches to yield additional insights. Unlike fMRI field-standard method of pairwise correlation, theory suggests that partial correlation can estimate without confounded and indirect connections. However, also display low repeat reliability, impairing accuracy individual estimates. We hypothesized reliability would be increased by adding regularization, which reduce overfitting noise in regression-based like correlation. therefore tested several regularized alternatives – graphical lasso, ridge, principal component regression against unregularized applying them empirical resting-state simulated data. As hypothesized, regularization vastly improved quantified using between-session similarity intraclass This then granted substantially more accurate estimates when validated structural (empirical data) ground truth networks (simulations). Graphical lasso showed especially high among approaches, seemingly maintaining valid underlying network structures. additionally found robust levels, data quantity, subject motion common error sources. Lastly, we demonstrated effectively predict task activations differences behavior, further establishing its external validity, ability characterize task-related functionality. recommend or similar methods calculating FC, as they unconfounded than while overcoming poor

Language: Английский

Citations

13
Neural representation dynamics reveal computational principles of cognitive task learning DOI Open Access
Ravi D. Mill, Michael W. Cole

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2023, Volume and Issue: unknown

Published: June 29, 2023

During cognitive task learning, neural representations must be rapidly constructed for novel performance, then optimized robust practiced performance. How the geometry of changes to enable this transition from performance remains unknown. We hypothesized that practice involves a shift compositional (task-general activity patterns can flexibly reused across tasks) conjunctive (task-specific specialized current task). Functional MRI during learning multiple complex tasks substantiated dynamic representations, which was associated with reduced cross-task interference (via pattern separation) and behavioral improvement. Further, we found conjunctions originated in subcortex (hippocampus cerebellum) slowly spread cortex, extending memory systems theories encompass representation learning. The formation hence serves as computational signature reflecting cortical-subcortical dynamics optimize human brain.

Language: Английский

Citations

12
Causally informed activity flow models provide mechanistic insight into network-generated cognitive activations DOI Creative Commons
Rubén Sánchez-Romero, Takuya Ito, Ravi D. Mill

et al.

NeuroImage, Journal Year: 2023, Volume and Issue: 278, P. 120300 - 120300

Published: July 29, 2023

Brain activity flow models estimate the movement of task-evoked over brain connections to help explain network-generated task functionality. Activity have been shown accurately generate activations across a wide variety regions and conditions. However, these had limited explanatory power, given known issues with causal interpretations standard functional connectivity measures used parameterize models. We show here that functional/effective (FC) grounded in principles facilitate mechanistic interpretation progress from simple complex FC measures, each adding algorithmic details reflecting principles. This reflects many neuroscientists' preference for reduced measure complexity (to minimize assumptions, compute time, fully comprehend easily communicate methodological details), which potentially trades off validity. start Pearson correlation (the current field standard) remain maximally relevant field, estimating validity range using simulations empirical fMRI data. Finally, we apply causal-FC-based modeling dorsolateral prefrontal cortex region (DLPFC), demonstrating distributed network mechanisms contributing its strong activation during working memory task. Notably, this model is able account DLPFC effects traditionally thought rely primarily on within-region (i.e., not distributed) recurrent processes. Together, results reveal promise parameterizing methods identify underlying cognitive computations human brain.

Language: Английский

Citations

10
Exploring the transmission of cognitive task information through optimal brain pathways DOI Creative Commons
Zhengdong Wang,

Yifeixue Yang,

Ziyi Huang

et al.

PLoS Computational Biology, Journal Year: 2025, Volume and Issue: 21(3), P. e1012870 - e1012870

Published: March 7, 2025

Understanding the large-scale information processing that underlies complex human cognition is central goal of cognitive neuroscience. While emerging activity flow models demonstrate task transferred by interregional functional or structural connectivity, graph-theory-based typically assume neural communication occurs via shortest path brain networks. However, whether optimal route for empirical transmission remains unclear. Based on a mapping framework, we found performance prediction with was significantly lower than direct path. The routing superior to other network strategies, including search information, ensembles, and navigation. Intriguingly, outperformed in when physical distance constraint asymmetric contribution were simultaneously considered. This study not only challenges assumption through but also suggests constrained spatial embedding network.

Language: Английский

Citations

0
Generalizing prediction of task-evoked brain activity across datasets and populations DOI Creative Commons
Niv Tik,

Shachar Gal,

Asaf Madar

et al.

NeuroImage, Journal Year: 2023, Volume and Issue: 276, P. 120213 - 120213

Published: June 1, 2023

Predictions of task-based functional magnetic resonance imaging (fMRI) from task-free resting-state (rs) fMRI have gained popularity over the past decade. This method holds a great promise for studying individual variability in brain function without need to perform highly demanding tasks. However, order be broadly used, prediction models must prove generalize beyond dataset they were trained on. In this work, we test generalizability task-fMRI rs-fMRI across sites, MRI vendors and age-groups. Moreover, investigate data requirements successful prediction. We use Human Connectome Project (HCP) explore how different combinations training sample sizes number datapoints affect success various cognitive then apply on HCP predict activations site, vendor (Phillips vs. Siemens scanners) age group (children HCP-development project). demonstrate that, depending task, set approximately 20 participants with 100 timepoints each yields largest gain model performance. Nevertheless, further increasing size results significantly improved predictions, until reaching 450-600 800-1000 timepoints. Overall, influences more than size. show that adequate amounts successfully groups provide predictions are both accurate individual-specific. These findings suggest large-scale publicly available datasets may utilized study smaller, unique samples.

Language: Английский

Citations

7
Activity flow under the manipulation of cognitive load and training DOI Creative Commons

Wanyun Zhao,

Kaiqiang Su, Hengcheng Zhu

et al.

NeuroImage, Journal Year: 2024, Volume and Issue: 297, P. 120761 - 120761

Published: July 27, 2024

Flexible cognitive functions, such as working memory (WM), usually require a balance between localized and distributed information processing. However, it is challenging to uncover how local processing specifically contributes task-induced activity in region. Although the recently proposed flow mapping approach revealed relative contribution of processing, few studies have explored adaptive plastic changes that underlie manipulation. In this study, we recruited 51 healthy volunteers (31 females) investigated brain activation frontoparietal systems was modulated by WM load training. While both executive control network (ECN) dorsal attention (DAN) increased linearly with at baseline, showed linear response only DAN, which prominently attributed within-network flow. Importantly, training selectively induced an increase ECN also load, were predominantly due between-network Furthermore, demonstrated causal effect prediction through manipulation on connectivity activity. contrast classic estimation, our findings suggest provides unique insights into neural under This study offers new methodological framework for exploring integration versus segregation underlying

Language: Английский

Citations

2
AI-Powered Intraoperative nerve monitoring: a visionary method to reduce facial nerve palsy in parotid surgery-an editorial DOI Open Access
Tooba Ali,

Hibah Abid Imam,

Biya Maqsood

et al.

Annals of Medicine and Surgery, Journal Year: 2023, Volume and Issue: 86(2), P. 635 - 637

Published: Dec. 15, 2023

Ali, Tooba (MBBS); Abid Imam, Hibah Maqsood, Biya Jawed, Ifra Khan, Iman Haque, Md Ariful (MBBS, MD, MPH) Author Information

Language: Английский

Citations

2
Connectome-Based Attractor Dynamics Underlie Brain Activity in Rest, Task, and Disease DOI Creative Commons
Robert Englert, Bálint Kincses, Raviteja Kotikalapudi

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2023, Volume and Issue: unknown

Published: Nov. 5, 2023

Abstract Understanding large-scale brain dynamics is a grand challenge in neuroscience. We propose functional connectome-based Hopfield Neural Networks (fcHNNs) as model of macro-scale dynamics, arising from recurrent activity flow among regions. An fcHNN neither optimized to mimic certain characteristics, nor trained solve specific tasks; its weights are simply initialized with empirical connectivity values. In the framework, understood relation so-called attractor states, i.e. neurobiologically meaningful low-energy configurations. Analyses 7 distinct datasets demonstrate that fcHNNs can accurately reconstruct and predict under wide range conditions, including resting task states disorders. By establishing mechanistic link between activity, offer simple interpretable computational alternative conventional descriptive analyses function. Being generative yield insights hold potential uncover novel treatment targets. Key Points present yet powerful phenomenological for The uses artificial neural network (fcHNN) architecture compute “activity flow” through regions several characteristics state conceptualize both task-induced pathological changes non-linear alteration these Our approach validated using neuroimaging data seven studies offers function

Language: Английский

Citations

1
Modeling the association between functional connectivity and lateralization with the activity flow framework DOI

Xue Zhan,

Jinwei Lang,

Lizhuang Yang

et al.

Brain Research, Journal Year: 2024, Volume and Issue: 1830, P. 148831 - 148831

Published: Feb. 25, 2024

Language: Английский

Citations

0
Network modeling: The explanatory power of activity flow models of brain function DOI
Michael W. Cole

Elsevier eBooks, Journal Year: 2024, Volume and Issue: unknown, P. 81 - 117

Published: Jan. 1, 2024

Language: Английский

Citations

0