Does the rich club control brain state transitions? A network control theoretical investigation.

1491 

Abstract Submission 

Alina Podschun¹, Richard Betzel², Sebastian Markett³

¹Humboldt-Universität zu Berlin, Berlin, Germany, ²Indiana University, Bloomington, IN, ³Humboldt-Universtität zu Berlin, Berlin, Germany

Alina Podschun  
Humboldt-Universität zu Berlin
Berlin, Germany

Richard Betzel  
Indiana University
Bloomington, IN

Sebastian Markett  
Humboldt-Universtität zu Berlin
Berlin, Germany

The brain seamlessly transitions between functionally relevant patterns of activity, often referred to as "brain states." Extensive research indicates that the underlying dynamics of these processes are linked to the structural connectome of the brain (Cole et al., 2016). The rich club-a network comprising densely interconnected brain regions-emerges as particularly pivotal for efficient integration of information (van den Heuvel & Sporns, 2011). Disruption of its organizational integrity results in an increase in energetic cost during transitions between brain states (Betzel et al., 2016), prompting some to designate its regions as potential control centers (Gu et al., 2014). In contrast, compelling evidence also underscores the significance of peripheral, sparsely connected regions in governing control processes (Betzel et al., 2016; Senden et al., 2018). In this study, we systematically delved into the specific role played by the rich club in overseeing transitions between behaviorally constrained brain states. Leveraging a network control theoretical (NCT) framework and openly available data from the Human Connectome Project (HCP), our investigation sought to elucidate potential control functions carried out by the rich club.

Analyses were grounded in individually reconstructed FA-weighted, undirected structural brain networks, constructed using the CATO pipeline (de Lange et al., 2023) and Lausanne sub-parcellation (219 ROI) of the Desikan Killiany atlas (Cammoun et al., 2021). Brain states were delineated using preprocessed contrasts from all seven tasks provided by the HCP. Individual rich clubs for each subject were identified based on the maximal normalized rich club coefficient (Riedel et al., 2022). On average, 21 regions (9.69%) of an individual's nodes were categorized as rich club members. The group-level rich club definition was consequently established by selecting the top-ranking 9.69% of nodes across subjects-those consistently identified as rich club members.
For the analysis of brain state stability and energetic cost of brain state transitions, we applied an optimal control framework (see Braun and colleagues, 2021). Linear approximations were used to represent brain state dynamics. The rich club's control contribution was assessed by excluding group-level rich club nodes from a matrix of potential control regions, preventing them from influencing brain state dynamics. Results were compared to a spin-test-based null model (Váša et al., 2018), where a size-matched set of random nodes was excluded instead. Statistical comparisons were conducted using repeated measures ANOVA.

Figure 1 shows an overview of constituent regions of the group-level rich club.
When these regions were excluded from the matrix of control nodes, the impact on control metrics was significantly less pronounced compared to excluding a size-matched set of random regions (see Figure 2). Brain state stability consistently remained higher when rich club nodes were excluded (all p < 0.01), and the energy required for transitioning between brain states was higher when excluding random regions than when excluding the rich club (all p < 0.01). These findings suggest that, contrary to expectations, the rich club consistently contributed less to the stabilization of brain states and the energetic control of state transitions than would be anticipated by chance.

In excluding rich club nodes as controllers within the optimal control energy framework, we consistently observed less pronounced impacts on control measures compared to the exclusion of random regions. Our findings challenge the notion of the rich club as a potential control center in the human brain, indicating that its influence on controlling brain state dynamics is less significant than previously believed. Instead, our results underscore the critical role of weakly connected peripheral nodes in efficiently controlling the brain's activity landscape.

Executive Function, Cognitive Control and Decision Making

Higher Cognitive Functions Other ²

Connectivity (eg. functional, effective, structural) ¹

fMRI Connectivity and Network Modeling

Other Methods

Cognition

FUNCTIONAL MRI

WHITE MATTER IMAGING - DTI, HARDI, DSI, ETC

Other - network control theory; rich club; graph theory; network neuroscience