|
Revealing a reservoir of latent variables for flexible behavior| title | Revealing a reservoir of latent variables for flexible behavior |
|---|
| start_date | 2023/11/20 |
|---|
| schedule | 11h |
|---|
| online | no |
|---|
| location_info | Salle de conférence |
|---|
| summary | Decisions involve selecting actions based on evidence, but in many real-world situations, there are multiple potential solutions, which may call for a higher-level decision about « how to make the decision ». Yet, how the brain determines which solution to use and when remains poorly understood. Addressing this question is challenging because it requires unraveling the latent processes employed by individual brains during decision-making. To move toward this goal, we developed a behavioral task for mice that allowed us to quantitatively probe a class of strategies employed by individual mice to solve the task. Surprisingly, we found that regardless of which strategy best explained the current behavior, neural activity in the frontal cortex concurrently encoded a full basis set of computations defining a reservoir of latent variables appropriate for alternative tasks. We discovered that these latent variables could be accessed not only through neural recordings but by decoding mice’s facial expressions. We provided evidence from both correlational and causal experiments that facial expressions of latent computations partially originate from neural activity in the secondary motor cortex (M2). These findings show that characterization of multidimensional neural activity and movements can offer a window into the computations used to form decisions and expose otherwise latent internal processes. |
|---|
| responsibles | Burban |
|---|
Workflow history| from state (1) | to state | comment | date |
|---|
| submitted | published | | 2023/11/14 13:58 UTC |
| |
|