Research · since 2021
Hippocampal replay for continual RL
If the hippocampus replays to cortex to build a world model rather than a policy, does that make continual RL more sample-efficient and less forgetful?
Working on it now
Aziz Alyahya
PhD · Monash · 2026 to present
world models and replay (ARROW)
How we got here
- 2023Luke Yang, Honours · Monash replay for continual RLAbout Luke’s project →World Model + Replay for model-based continual RL (WMAR)
- 2026
Project notes
Hippocampal-style Replay for Continual RL (2021-2023)
This project takes a different approach to RL, inspired by evidence that Hippocampus replays to the frontal cortex directly. It is likely used for model building, as opposed to the mainstream view in cognitive science and ML – where ‘experience replay’ ultimately improves policy. We hypothesise that this approach would be more sample efficient, better able to generalise to new tasks and have higher plasticity whilst maintaining stability.
We tackled this back in 2021 based on the World Models architecture by Ha and Shmidhuber 2018. Recently Luke completed his it as an Honours project based on the Dreamer architecture. See the preprint here.