Disentangling communication across populations of neurons
Technological advances now allow us to record from large populations of neurons across multiple brain areas. These recordings may illuminate how communication between areas contributes to brain function, yet a substantial barrier remains: how do we disentangle the concurrent, bidirectional flow of signals between populations of neurons? Here we develop a dimensionality reduction framework, delayed latents across groups (DLAG), that disentangles signals relayed in each direction, identifies how these signals are represented by each population, and characterizes how they evolve within and across trials. We systematically validate DLAG in simulation, demonstrating that it performs well over a wide range of simulated conditions, including synthetic datasets similar in scale to current neurophysiological recordings. We also demonstrate its robustness to mild deviations from its model assumptions. Then we use DLAG to study bidirectional communication between neuronal populations in (1) visual areas V1 and V2, recorded simultaneously in anesthetized macaques, and (2) V1 and V4, recorded simultaneously in an awake, passively fixating macaque. In both studies, DLAG revealed signatures of bidirectional yet selective communication. To support the interpretation of DLAG models fit to these neural recordings, we develop descriptive and inferential statistics. Finally, we extend the DLAG framework to include an arbitrary number of neuronal populations (that is, three or more), and validate the extended method with simulated neural activity. This work lays a foundation for dissecting the intricate flow of signals across populations of neurons, and how this signaling contributes to cortical computation.
- Electrical and Computer Engineering
- Doctor of Philosophy (PhD)