Reproduce “Modeling human reputation-seeking behavior in a spatio-temporally complex public good provision game” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Edward Hughes, Tina O. Zhu, Martin J. Chadwick, Raphael Koster, Antonio García Castañeda, Charles Beattie, Thore Graepel, Matthew M. Botvinick and 1 more
The paper tests whether making players identifiable and their contributions visible changes cooperation in Clean Up, a five-player game where cleaning a polluted river costs apple harvesting time. Human groups earned 1,227 versus 982 points and cleaned 396 versus 337 steps in identifiable versus anonymous conditions. A reputation-based model reproduced the direction, without establishing reputation as the human cause.
Key insights
- MARL models can serve as robust tools for computational social science, accurately simulating complex human group behaviors, including specific coordination strategies, when empirically validated.
- Identifiability and the ability to seek reputation are critical drivers for human cooperation in spatio-temporally complex social dilemmas, leading to higher contributions.
- Temporal turn-taking emerges as a predominant and effective coordination strategy for both humans and MARL agents in public good provision, mediating the relationship between identifiability and group performance.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.