Functional Connectivity's Degenerate View of Brain Computation

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Systems Neuroscience

Background:

• Brain computation involves complex neural interactions quantified by functional connectivity (FC).
• The relationship between FC, static structural connectivity (SC), and neurophysiological dynamics remains debated.
• Empirical findings on FC variability and its cognitive correlates are inconsistent.

Purpose of the Study:

• To characterize the variety of patterns in FC and SC using a unified computational approach.
• To investigate the extent to which FC reflects underlying neurophysiological dynamics versus static SC.
• To assess the dimensionality and complexity of brain interaction patterns captured by FC.

Main Methods:

• Combined multivariate analysis, bootstrap methods, and computational modeling.
• Generated simulations from models with varying dynamical behaviors.
• Analyzed empirical neuroimaging data (BOLD signals) and structural connectivity (SC) estimates.

Main Results:

• Variability across FC patterns was largely explained by a low-dimensional linear subspace (1-2 dimensions).
• BOLD signal variability could not be similarly reduced, indicating higher complexity.
• FC strongly reflected SC and was influenced by a Gaussian process, with limitations in SC estimation impacting results.
• FC measures may represent brain interactions degenerately, with a common core reflecting SC and limited residual variability.

Conclusions:

• FC measures may provide a simplified or degenerate representation of complex brain interactions.
• The limited dynamical range of the BOLD signal and SC estimation methods constrain the information captured by FC.
• FC's common core reflects network capacity constrained by SC, with residual variability potentially holding meaningful information.