Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper addresses the challenge of training world models in multi-agent reinforcement learning, where distinguishing between statistical correlations and causal mechanisms is critical. The authors introduce Implicit Causal World Models, which leverage offline demonstrations to uncover environmental dynamics without needing predefined causal structures. Their evaluations show that these models yield interpretable causal representations and improve accuracy in coordination tasks as interventional strength increases.
Implicit Causal World Models can accurately discern causal dynamics in multi-agent systems, enhancing interpretability and performance in complex tasks.
In model-based reinforcement learning, world models exist as internal simulators, but their training often conflates statistical correlations with causal mechanisms. This problem is exacerbated in multi-agent systems where physical transitions are intertwined with strategic agent intents, causing world models to fail under distribution shift. We introduce Implicit Causal World Models to recover environmental dynamics from offline demonstrations without requiring pre-defined causal graphs. By incorporating policy variance, we render world models discoverable via the sequential backdoor condition. Evaluations across coordination tasks (Two-Door, Navigation, and Giveway) demonstrate that these models provide interpretable causal representations under both full and partial observability, with model accuracy scaling directly with interventional strength.