# Prism: Policy Reuse via Interpretable Strategy Mapping in Reinforcement Learning

arXiv: https://arxiv.org/abs/2604.02353 · HTML: https://mlsanity.com/paper/2604.02353/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Interpretable and logic-guided RL, one of 15 directions in Deep reinforcement learning methods.
- Builds on 13 of its 20 references that are in the map.
- First posted on arXiv 4 Mar 2026.

## Placed in

- Deep reinforcement learning methods / [Interpretable and logic-guided RL](https://mlsanity.com/deeprl/interpretable-and-logic-guided-rl/)

## Builds on (most cited first, 13 of 13)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks](https://mlsanity.com/paper/1703.03400/) (2017)
- [Progressive Neural Networks](https://mlsanity.com/paper/1606.04671/) (2016)
- [Concept Bottleneck Models](https://mlsanity.com/paper/2007.04612/) (2020)
- [Mastering Atari with Discrete World Models](https://mlsanity.com/paper/2010.02193/) (2020)
- [Diversity is All You Need: Learning Skills without a Reward Function](https://mlsanity.com/paper/1802.06070/) (2018)
- [Data-Efficient Hierarchical Reinforcement Learning](https://mlsanity.com/paper/1805.08296/) (2018)
- [Policy Distillation](https://mlsanity.com/paper/1511.06295/) (2015)
- [Causal Abstractions of Neural Networks](https://mlsanity.com/paper/2106.02997/) (2021)
- [Successor Features for Transfer in Reinforcement Learning](https://mlsanity.com/paper/1606.05312/) (2016)
- [Probabilistic Concept Bottleneck Models](https://mlsanity.com/paper/2306.01574/) (2023)
- [Stochastic Concept Bottleneck Models](https://mlsanity.com/paper/2406.19272/) (2024)
- [Interpretable Concept Bottlenecks to Align Reinforcement Learning Agents](https://mlsanity.com/paper/2401.05821/) (2024)

## Abstract (arXiv)

We present PRISM (Policy Reuse via Interpretable Strategy Mapping), a framework that grounds reinforcement learning agents' decisions in discrete, causally validated concepts and uses those concepts as a zero-shot transfer interface between agents trained with different algorithms. PRISM clusters each agent's encoder features into K concepts via K-means. Causal intervention establishes that these concepts directly drive - not merely correlate with - agent behavior: overriding concept assignments changes the selected action in 69.4% of interventions (p = 8.6 × 10-86, 2500 interventions). Concept importance and usage frequency are dissociated: the most-used concept (C47, 33.0% frequency) causes only a 9.4% win-rate drop when ablated, while ablating C16 (15.4% frequency) collapses win rate from 100% to 51.8%. Because concepts causally encode strategy, aligning them via optimal bipartite matching transfers strategic knowledge zero-shot. On Go~7×7 with three independently trained agents, concept transfer achieves 69.5%±3.2% and 76.4%±3.4% win rate against a standard engine across the two successful transfer pairs (10 seeds), compared to 3.5% for a random agent and 9.2% without alignment. Transfer succeeds when the source policy is strong; geometric alignment quality predicts nothing (R2 ≈ 0). The framework is scoped to domains where strategic state is naturally discrete: the identical pipeline on Atari Breakout yields bottleneck policies at random-agent performance, confirming that the Go results reflect a structural property of the domain.