Modular Organization of Electrical Fluctuations in the Mouse Brain

A powerful computational feature of the brain is its ability to compartmentalize different functions in a way that can be flexibly recombined. Experimental evidence for such modularity arose from cytoarchitecture, connectivity, and electroencephalograms, while magnetic resonance imaging could attest that this modularity is dynamic. More spatially precise and more widely recordable, the local field potential (LFP) has an unknown brain wide organization. Here, we developed deep electrospectroscopy, a method to assess nonlinear spectral similarity between the LFP of different areas. A brain-wide application of this technique to the mouse brain revealed an organization composed of groups of spectrally similar areas. Such communities were mostly found in the fore-brain and showed an organization that did not strictly follow the cytoarchitecture. For instance, visual parts of the cortex and the visual parts of the colliculus were found in the same community. These electrospectral communities reshaped with context, showing splitting and merging operations, growing in size around brain areas required for a task. In particular, upper and lower limb parts of the somatosensory cortex were primarily in separate communities but merged during a task that required turning a wheel. Similarly, oculomotor reflexes and associative parts of the thalamus merged during the visuo-motor task. These analyzes show that LFPs are organized in a modular fashion, offering a window onto subcortical and layer-specific contributions to the compositionality of cortical brain functions.