🎯 With training, RNNs implemented chronological memory subspaces allowing them to track object information using rotational dynamics—supporting resource-based models of working memory.

📐 Surprisingly, object features are less orthogonalized in RNN representations compared to perceptual space.

🧠 We found that multi-task RNNs (unlike single-task ones) retain both task-relevant & irrelevant info but reusable representations only emerged in simple gateless architectures.

🖥️ To answer this question, we trained multi-task RNNs (vanilla, GRU, LSTM) on 9 N-back tasks using naturalistic 3D object stimuli to study encoding, retention, & retrieval dynamics.

🌟 New Research Alert! 🌟
Excited to share our latest work (accepted to NeurIPS2024) on understanding working memory in multi-task RNN models using naturalistic stimuli!: with @takuito.bsky.social and @bashivan.bsky.social
#tweeprint below: