Bayesian generative modeling reveals a multi-modal hierarchical architecture in the mouse functional connectome

Area of Science:

• Systems neuroscience
• Connectomics
• Computational neuroscience

Background:

• Understanding large-scale brain organization is a key challenge.
• Variability in analytical approaches suggests functional networks may not have a single structure.

Purpose of the Study:

• To test the hypothesis that brain connectome organization is multi-modal.
• To characterize the full posterior distribution over network partitions.

Main Methods:

• Bayesian generative modeling framework
• Stochastic block models
• Analysis of resting-state fMRI data in mice

Main Results:

• A non-degree-corrected hierarchical architecture best describes the functional connectome.
• Multiple distinct and co-dominant organizational schemes exist.
• Higher-order association networks are fractionated, while sensory systems remain cohesive.

Conclusions:

• The resting-state connectome exhibits a distribution over alternative, co-dominant organizational configurations.
• Brain organization is inherently degenerate, offering a repertoire of network configurations.
• This perspective reconciles inconsistencies in previous studies.