Shreya Saxena

This paper introduces SSLD, a framework that learns shared neural-behavioral latent dynamics and private modality-specific variables, showing it accurately reconstructs signals and isolates behaviorally relevant information across multiple datasets.

This study shows that cooperation in rats activates specific corticolimbic pathways and leads to lasting synaptic changes and altered social behavior.

This study shows that sensorimotor cortex maintains canonical dynamics during natural prey capture but flexibly reconfigures them for online control.

This paper develops a method to model neural population dynamics as optimal feedback solutions and shows they match recorded motorcortical responses.

This study trains a multitasking RNN with a musculoskeletal arm model to perform diverse movements, revealing compositional strategies and shared low-dimensional representations for flexible movement generation.

This paper introduces Time-Varying RNNs (TV-RNNs) for early classification of neural activity sequences, showing improved accuracy and earlier detection compared to standard RNNs in mouse and human motor task data.

The paper introduces muSim, a framework that uses RNNs and reinforcement learning to model motor cortex control of movement through musculoskeletal simulations, showing it can decode neural activity and generalize to new conditions.

The paper shows that pre-movement neural activity in mice contains structured dynamics that predict goal-directed actions, but not random ones, several seconds before movement.

The paper shows that motor cortical activity covariation changes between tasks, driving distinct muscle activation patterns rather than relying on fixed flexion/extension programs or task-specific neuronal subsets.

This paper introduces a constrained subspace variational autoencoder (CS-VAE) to disentangle individual and social behavioral features in multi-subject settings, enabling analysis of latent variables for downstream tasks.

This paper introduces a cloud-based platform that provides automated, reproducible, and accessible data analysis tools for neuroscience, improving efficiency and reducing infrastructure barriers.

The paper shows that motor cortex activity during movement tasks reflects network-level strategies for generating muscle activity rather than direct muscle control.

This paper introduces LocaNMF, a method to decompose widefield calcium imaging data into interpretable spatial and temporal components for analyzing neural activity across mice.