Research map / Multi-agent systems and games / Games and equilibria
AGiR: Mitigating Gift Over-Reliance in Mixed-Motive Games
Woohyeon Byeon, Seongmin Kim, Jiwon Jeon, Woojun Kim, Youngchul Sung
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Identifies gift over-reliance in gifting methods for mixed-motive games and proposes AGiR, a refusal-augmented gifting framework to mitigate it.
Problem
In mixed-motive games, existing gifting methods lead agents to become overly dependent on incoming gifts and fail to learn reward-yielding behaviors (gift over-reliance).
Approach
A refusal-augmented gifting framework (AGiR) that lets agents refuse gifts so they must learn reward-yielding behaviors instead of depending on incoming gifts; details beyond the abstract are not available.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add adaptive gift refusal (AGiR) to LASE | LASE · Two-agent Cleanup, spawn gap | river-side agent extrinsic return: 246 -> 276 | Improved |
| add AGiR to LASE | LASE · Two-agent Cleanup, mismatch | talented harvester total return: 128 -> 176 | Improved |
| add AGiR to LASE | LASE · Two-agent Cleanup, mismatch | talented cleaner total return: 8 -> 81 | Improved |
| add AGiR to LASE when there is no spawn or competence gap | LASE · Two-agent Cleanup, no gap | outcome | No clear effect |
| add AGiR to LASE, Unfair Cleanup-Hard | LASE, RX · Unfair Cleanup-Hard | Min, GM, GRR | Improved |
| add AGiR to base gifting methods in Unfair Coin Game | LASE, RX, LIO · Unfair Coin Game | Min | Improved |
| add naive refund of refused gifts to the donor (ablation) | AGiR · Unfair environments | average over metrics | Hurt |
| remove refusal weight w from the policy input (ablation) | AGiR · Unfair environments | fairness metrics | Hurt |
| use adaptive baseline instead of fixed baseline pi/4 (ablation) | AGiR · Unfair Cleanup-Hard | Min, GM | Improved |
| replace adaptive refusal with fixed acceptance weight, gift clipping or linear annealing (ablation) | AGiR · Unfair Cleanup-Hard | GM, Min | Hurt |
| vary window size H (ablation) | AGiR · Unfair Cleanup-Hard | Min, GM | No clear effect |
Surprising findings
- Existing gifting methods can cause agents to become overly dependent on incoming gifts and fail to learn reward-yielding behaviors (gift over-reliance).
Experimental setup
Builds on: gifting methods
Closest papers by meaning
- Learning to Balance Altruism and Self-interest Based on Empathy in Mixed-Motive Games (NeurIPS 2024 · similarity 0.88)
- Fair Cooperation in Mixed-Motive Games via Conflict-Aware Gradient Adjustment (NeurIPS 2025 · similarity 0.88)
- Beyond Mandatory Federations: Balancing Egoism, Utilitarianism and Egalitarianism in Mixed-Motive Games (AAAI 2025 · similarity 0.86)
- Aligning Individual and Collective Objectives in Multi-Agent Cooperation (NeurIPS 2024 · similarity 0.86)
- Scalable Neural Incentive Design with Parameterized Mean-Field Approximation (NeurIPS 2025 · similarity 0.85)
- Mitigating Retaliatory Algorithmic Collusion in Repeated Games (NeurIPS 2026 · similarity 0.84)
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.
