# How RL Agents Behave When Their Actions Are Modified

arXiv: https://arxiv.org/abs/2102.07716 · HTML: https://mlsanity.com/paper/2102.07716/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- Builds on 8 of its 33 references that are in the map.
- First posted on arXiv 15 Feb 2021.

## Builds on (most cited first, 8 of 8)

- [Evolution Strategies as a Scalable Alternative to Reinforcement Learning](https://mlsanity.com/paper/1703.03864/) (2017)
- [Safe Model-based Reinforcement Learning with Stability Guarantees](https://mlsanity.com/paper/1705.08551/) (2017)
- [Deep Reinforcement Learning for Robotic Manipulation with Asynchronous Off-Policy Updates](https://mlsanity.com/paper/1610.00633/) (2016)
- [Safe Exploration in Continuous Action Spaces](https://mlsanity.com/paper/1801.08757/) (2018)
- [Fixed Point Quantization of Deep Convolutional Networks](https://mlsanity.com/paper/1511.06393/) (2015)
- [AI Safety Gridworlds](https://mlsanity.com/paper/1711.09883/) (2017)
- [Trial without Error: Towards Safe Reinforcement Learning via Human Intervention](https://mlsanity.com/paper/1707.05173/) (2017)
- [Agent-Agnostic Human-in-the-Loop Reinforcement Learning](https://mlsanity.com/paper/1701.04079/) (2017)

## Built on by (most cited first, 10 of 10)

- [Alignment of Language Agents](https://mlsanity.com/paper/2103.14659/) (2021)
- [Towards Safe Reinforcement Learning with a Safety Editor Policy](https://mlsanity.com/paper/2201.12427/) (2022)
- [Discovering Agents](https://mlsanity.com/paper/2208.08345/) (2022)
- [Reasoning about Causality in Games](https://mlsanity.com/paper/2301.02324/) (2023)
- [Collaborate, Deliberate, Evaluate: How LLM Alignment Affects Coordinated Multi-Agent Outcomes](https://mlsanity.com/paper/2509.05882/) (2025)
- [A Complete Criterion for Value of Information in Soluble Influence Diagrams](https://mlsanity.com/paper/2202.11629/) (2022)
- [Quantifying stability of non-power-seeking in artificial agents](https://mlsanity.com/paper/2401.03529/) (2024)
- [Learning "Partner-Aware" Collaborators in Multi-Party Collaboration](https://mlsanity.com/paper/2510.22462/) (2025)
- [Governable AI: Provable Safety Under Extreme Threat Models](https://mlsanity.com/paper/2508.20411/) (2025)
- [Flout at Your Own Risk: LLMs Struggle with Pragmatic Cooperativity Under Epistemic Asymmetry](https://mlsanity.com/paper/2607.11053/) (2026)

## Abstract (arXiv)

Reinforcement learning in complex environments may require supervision to prevent the agent from attempting dangerous actions. As a result of supervisor intervention, the executed action may differ from the action specified by the policy. How does this affect learning? We present the Modified-Action Markov Decision Process, an extension of the MDP model that allows actions to differ from the policy. We analyze the asymptotic behaviours of common reinforcement learning algorithms in this setting and show that they adapt in different ways: some completely ignore modifications while others go to various lengths in trying to avoid action modifications that decrease reward. By choosing the right algorithm, developers can prevent their agents from learning to circumvent interruptions or constraints, and better control agent responses to other kinds of action modification, like self-damage.