Back

Mapping neural subspace dynamics onto the structure of the mouse descending motor system

Hasnain, M. A.; Han, Y.; Economo, M. N.

2025-10-17 neuroscience
10.1101/2025.10.17.682917 bioRxiv
Show abstract

The motor cortex supports various cognitive and motor functions. To prevent interference between these processes, the associated neural dynamics may be organized into orthogonal subspaces. In this subspace model, activity in a movement-null subspace encodes internal processes, while activity in a movement-potent subspace relates to movements. The biological implementation of this model - how activity in different subspaces map onto neural circuits - remains unclear. Particularly, it is unknown whether different cell types, with specific connectivity patterns, preferentially contribute to specific subspaces. Here, we test whether the cell type that directly links motor cortex to motor centers in the medulla and spinal cord - lower layer 5b extratelencephalic neurons (L-ETN) - preferentially encodes activity contained in the movement-potent subspace. We performed cell-type-specific recordings in the motor cortex while mice performed a delayed-response licking task, and decomposed population activity into movement-null and movement-potent subspaces. We find that L-ETN activity spans both movement-null and movement-potent subspaces in a manner that could not be distinguished from the broader motor cortex population. Notably, downstream medullary circuits retain a subset of the movement-null dynamics contained in motor cortex, with select movement-null signals specifically filtered out. These results indicate that a distinct cortical output alone does not mediate the cancelation of null dynamics; movement-null computations are instead distributed across the descending motor hierarchy.

Matching journals

The top 3 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.