Reframes plasticity loss in Reinforcement Learning as an optimization problem where networks get trapped in local optima of previous tasks.
March 24, 2026
Original Paper
Rethinking Plasticity in Deep Reinforcement Learning
arXiv · 2603.21173
The Takeaway
It moves past descriptive metrics like 'dormant neurons' to explain *why* networks stop learning in non-stationary environments. This insight provides a rigorous framework for developing methods to restore network plasticity in long-term RL agents.
From the abstract
This paper investigates the fundamental mechanisms driving plasticity loss in deep reinforcement learning (RL), a critical challenge where neural networks lose their ability to adapt to non-stationary environments. While existing research often relies on descriptive metrics like dormant neurons or effective rank, these summaries fail to explain the underlying optimization dynamics. We propose the Optimization-Centric Plasticity (OCP) hypothesis, which posits that plasticity loss arises because o