Want to read

Mental World Modeling

Artificial Intelligence Computer Science

This paper introduces Mental World Modeling, a framework integrating physical and mental state variables to predict human actions. The authors propose MENTIS, a training-free baseline that parses scenes, generates target-specific observations, and simulates coupled transitions. Experiments across eight large language models demonstrate that explicitly tracking mental states is essential for accurate decision prediction, highlighting current bottlenecks in existing world modeling approaches.

Read the paper

Key takeaways

Mental World Modeling treats mental variables as core components rather than post-hoc rationales.
The framework maintains a coupled physical-mental world state for joint simulation.
A target-specific partial observation is rendered based on the agent's perspective.
Candidate actions are decomposed to update both physical and mental components.
The MENTIS baseline is training-free and fully inspectable across all stages.
Explicit mental state modeling is essential for predicting human decisions accurately.

Reflections

No notes on this one yet. I add reflections as I finish or revisit a paper.