English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

In motor neuroscience, artificial recurrent neural networks models oftencomplement animal studies. However, most modeling efforts are limited todata-fitting, and the few that examine virtual embodied agents in areinforcement learning context, do not draw direct comparisons to theirbiological counterparts. Our study addressing this gap, by uncoveringstructured neural activity of a virtual robot performing legged locomotion thatdirectly support experimental findings of primate walking and cycling. We findthat embodied agents trained to walk exhibit smooth dynamics that avoidtangling -- or opposing neural trajectories in neighboring neural space -- acore principle in computational neuroscience. Specifically, across a wide suiteof gaits, the agent displays neural trajectories in the recurrent layers areless tangled than those in the input-driven actuation layers. To betterinterpret the neural separation of these elliptical-shaped trajectories, weidentify speed axes that maximizes variance of mean activity across differentforward, lateral, and rotational speed conditions.

From Data-Fitting to Discovery: Interpreting the Neural Dynamics of Motor Control through Reinforcement Learning