I think I needed this reminder that Gemini thinks that some brain nuclei receive projections from neurons outside the brain.

Nostalgia for neuro paper of the 90s, this is Llinas and Ribary 1992

What technique did we use? We extracted second long spectrograms and used deep learning to localize each sample. We checked that the deep nets could generalize to new experiments and new animals. We then used the confusion matrix as a graph and searched for highly interconnected groups.

Graphs highlighted resemblance between communities sometimes hinged on pairs areas. Here are examples of such pairs between communities involving the motor planning community.

The largest community outside the cerebellum and hindbrain contained parts of the basal ganglia (putamen and nucleus accumbens) and motor planning regions of cortex (ACA, GU, MOs, MOp, ORB). It split in two during visual task.

This atlas of the local field potentials showed much re-organization when comparing rest and the visual+wheel turning task. Among many interesting changes, not only did the core visual communities grew larger, a new community highlighted ressemblence of tactile areas for upper and lower limbs.

For instance, the visual layers of superior colliculus were most similar to primary cortex as well as other visual areas such as VISpm & VISa, retrosplenial and area poststriata (hippocampal). The motor parts of the superior colliculus were most similar to noncortical oculomotor areas

How are the fluctuations in electric field organized across the whole brain? We analyzed the 3.4 M samples from the international brain lab using a combination of deep learning and graph theory. We found a surprising structure made of communities and landmarks.

doi.org/10.1101/2025...