How weak links in the brain connectome can be functionally significant: dynamic mechanism and prediction of individual differences in cognitive abilities

Mammalian brains perform cognitive functions following dynamic patterns organized on a complex large- scale inter-regional network, called brain connectome. The structure-dynamics-function relationship is at the core of research in network neuroscience. Recent data in animals (e.g., mouse and macaque monkey) and humans have revealed that the brain connectome is densely connected (interregional density ~0.6), but the connection weights are highly heterogeneous, covering several orders of magnitude (e.g., from ~10!" to 10!# ). However, previous analysis of brain network has mainly considered the backbone with spare connectivity (typical density ~1~10%, depending on atlas resolution), by thresholding out and ignoring most the weak links. Strikingly, our recent analysis of large-scale data from the WU-Minn Human Connectome Project (HCP) has found that many of such ignored weak links are functionally significant: the weight variation of these weak links is correlated with cognitive abilities across individuals. This finding raised an important question: how are weak links functionally significant and do they cause individual differences in cognitive abilities?

This project proposes to address this question from the perspective of dynamic mechanisms. Weak structural connectivity could induce functional consequences through impacting dynamical interactions. Ample evidence has shown that brain dynamics is characterized by critical states, which are sensitive to small perturbations. Thus, our hypothesis is that weak links can be dynamically significant through the interplay with the critical states of neural circuits. We will test this hypothesis by analyzing how sizable signal transmission and synchronization (functional connectivity) can be realized by weak coupling between two local circuit models operating around critical states. The analysis will be based on a biologically plausible excitation-inhibition neuronal network model that we have established to study critical neural dynamics. We will then extend circuit model on the heterogeneous structural connectome in human brains to understand how the weak links and their weight variations can explain individual differences in functional connectivity patterns around the critical bifurcation point. The modeling results will be calibrated and validated by real functional connectivity from fMRI data in HCP and used to derive weak links that can predict individual differences in cognitive abilities. This project is expected to integrate two salient brain features, heterogeneous network structure and critical dynamics, in a unified framework to deepen our understanding of the structure-dynamics-function relationship. Furthermore, it will provide useful insights toward understanding neuropsychiatric disorders and toward the development of brain- inspired neural network and dynamics for efficient computing in AI.