Research map / Multi-agent systems and games / Cooperative multi-agent RL
Social Influence as Intrinsic Motivation for Multi-Agent Deep Reinforcement Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add a causal social influence reward | A3C agents in the basic influence setting · Cleanup | final collective reward: 89 -> 190 | Helped |
| Add a causal social influence reward | A3C agents in the basic influence setting · Harvest | final collective reward: 485 -> 1073 | Helped |
| Add social influence to communication-policy training (ablation) | Agents with an explicit communication channel · Cleanup | collective reward | Helped |
| Add social influence to communication-policy training (ablation) | Agents with an explicit communication channel · Harvest | collective reward | Helped |
| Train the communication policy only with environmental reward instead of social influence (ablation) | Agents with an explicit communication channel · Cleanup and Harvest | collective reward | Hurt |
| Add an MOA-based influence reward rather than using the MOA only as an auxiliary prediction task (ablation) | Agents with a Model of Other Agents · Cleanup and Harvest | collective reward | Helped |
| Use the MOA as an auxiliary task without using its output to compute an influence reward (ablation) | Agents with a Model of Other Agents · Cleanup and Harvest | collective reward | Helped |
| Add an internal model of other agents and use it to compute influence | A3C agents in the MOA setting · Cleanup | final collective reward: 89 -> 392 | Helped |
| Add an internal model of other agents and use it to compute influence | A3C agents in the MOA setting · Harvest | final collective reward: 485 -> 588 | Helped |
| Add the social influence reward | A3C agents in the Box trapped environment · Box trapped | trajectories in which the open box action occurs: 0% -> 88% | Helped |
| Allow agents to condition their policy on other agents' visible actions without adding an influence reward (ablation) | A3C agents in the visible actions baseline · Cleanup and Harvest | collective reward | Helped |
| Set the extrinsic reward weight to zero for the communication head | Influence communication agents · Cleanup | collective reward | Helped |
Closest papers by meaning
- Situation-Dependent Causal Influence-Based Cooperative Multi-Agent Reinforcement Learning (AAAI 2024 · similarity 0.89)
- Bridging Training and Execution via Dynamic Directed Graph-Based Communication in Cooperative Multi-Agent Systems (AAAI 2025 · similarity 0.88)
- A Principle of Targeted Intervention for Multi-Agent Reinforcement Learning (NeurIPS 2025 · similarity 0.88)
- Detecting Influence Structures in Multi-Agent Reinforcement Learning (ICML 2024 · similarity 0.88)
- Deep Reinforcement Learning for Communication Networks (AAAI 2024 · similarity 0.87)
- Multi-Agent Reinforcement Learning with Communication-Constrained Priors (NeurIPS 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.