Task-Dependent Reconfiguration of Global Functional Connectivity and Complexity
Poster No:
1125
Submission Type:
Abstract Submission
Authors:
Maren Wehrheim1, Joshua Faskowitz2, Christian Fiebach1
Institutions:
1Goethe University Frankfurt, Frankfurt am Main, Hesse, 2Indiana University Bloomington, Bloomington, IN
First Author:
Co-Author(s):
Introduction:
Brain activity continuously fluctuates over time, even if the brain is in controlled (e.g., experimentally induced) states. Recent years have seen an increasing interest in understanding the complexity of these temporal variations, particularly also with respect to between-person differences in brain function. Here, we study how the variability and complexity of task-elicited BOLD activation, as well as global functional network properties, adjust to increasing cognitive demands and how these dynamic reconfigurations relate to individual differences in task performance.
Methods:
We use data from the Human Connectome Project (van Essen et al., 2013) from the working memory task (N-back). We select a subset of unrelated low-motion individuals (see Ito et al., 2020) for whom complete scan protocols were available (N=330). We compute time-resolved global functional connectivity across the complete scan following Esfhalani et al. (2020). We then measure the temporal variability of global connectivity (standard deviation and mean squared successive difference) within each task block. Additionally, we measure the complexity of the BOLD activation time series as the extrinsic (linear) dimensionality and intrinsic (non-linear) dimensionality of the signal within each block. To determine reliability of the derived measures of neural variability and complexity used in the present study, we compute reliability across consecutive runs (within-session Spearman-Brown-corrected split-half correlations).
Results:
As cognitive load increases, brain-wide functional connectivity is reduced (global functional decoupling). While this result may appear unexpected at first glance (under the assumption that higher cognitive load requires increased functional integration), we observed that load-dependent functional decoupling is directly associated with an increase in complexity (non-linear dimensionality) of the BOLD signal across all brain networks, reflecting a fundamental reconfiguration of the whole functional connectome. Both components of reconfiguration, functional decoupling and dimensionality increase, are directly associated (r = -.77) and directly behaviorally relevant, as indicated by significant across-participant correlations with working memory performance (r = .29 and r = -.24, respectively). Thus, better performance goes hand in hand with greater functional decoupling and increased network complexity. We furthermore observed that functional decoupling evolves over the course of task blocks and that better performance is associated with less variable connectivity dynamics over the course of the task (r = -.26). Lastly, we demonstrate that intrinsic BOLD dimensionality and the variability of global functional connectivity strength have good split-half reliability (r > .76), thus supporting the use as robust markers of individual differences at the neuro-functional level.
Conclusions:
Our results demonstrate the dynamic reconfiguration of brain networks in response to varying cognitive task demands. Whereas previous literature showed heterogeneous results with respect to how functional connectivity is adjusted under increasing working memory demands, our results indicate that global connectivity is reduced because the brain is reconfigured dynamically in a task-dependent manner. Interestingly, reconfiguration is on the one hand more pronounced in persons with better working memory performance (and thus 'more dynamic'), while at the same time more consistent (i.e., less variable over time). More generally, our results demonstrate that variability and complexity in the brain are reliable measures with high behavioral relevance, which will contribute important insights into understanding the nature of individual differences in cognitive abilities.
Learning and Memory:
Working Memory 1
Modeling and Analysis Methods:
Activation (eg. BOLD task-fMRI)
Connectivity (eg. functional, effective, structural) 2
Keywords:
Cognition
Computational Neuroscience
Data analysis
FUNCTIONAL MRI
Memory
1|2Indicates the priority used for review
Provide references using author date format
Ito, T., Brincat, S. L., Siegel, M., Mill, R. D., He, B. J., Miller, E. K., ... & Cole, M. W. (2020). Task-evoked activity quenches neural correlations and variability across cortical areas. PLoS computational biology, 16(8), e1007983.
Van Essen, D. C., Smith, S. M., Barch, D. M., Behrens, T. E., Yacoub, E., Ugurbil, K., & Wu-Minn HCP Consortium. (2013). The WU-Minn human connectome project: an overview. Neuroimage, 80, 62-79.