Decomposing the modulation of interactions between neuronal populations

Multiplicative interaction channels (MICs) decomposes how a measured modulator reshapes communication subspace interactions, recovering modulation geometry in high-dimensional, low-sample simulations.

• Why it matters: Across-brain recordings need interpretable components showing how behavioral state or other population activity changes which neuron subspaces interact, beyond additive interaction models.
• What they did: MICs extends the communication subspace framework using bilinear perturbation of reduced-rank regression to fit a low-rank tensor, then derives MIC triplets via hierarchical fitting and closed-form decomposition.
• The result: In simulations MICs reliably recovers ground-truth modulation, and in macaque and mouse datasets it identifies stimulus and behavioral-state modulation that reconfigures interaction geometry and prefrontal projection patterns.

AI summary of the abstract · may contain errors · not medical advice · read the paper.