智能体
arxiv.org ·
LLM 智能体在行为博弈中表现为静态 Level-k 玩家
该 arXiv 预印本（econ.GN，v2 修订于 2026-09-16）检验 LLM 作为行为博弈人类替身的假设。研究使用 p-beauty contest 和公共品博弈，在同一模型家族内以 360 格因子设计变化 temperature、规模（0.5-32B）、量化、instruct/base 与框架，并将各设置的选择分布与已发表人类数据整体分布比较。结果显示：除量化外，各部署设置分别影响保真度的不同方面；人类选择的离散度可部分通过部署设置恢复，但其背后的策略过程无法恢复。以 level-k 认知理论看，LLM 表现为静态、按类别检索的 level-k 玩家，k 由模型规模设定，且在多轮期界中不进行博弈内信念更新或逆向归纳。公共品博弈中人类贡献衰减，LLM 在各规模下持平或上升；无限期界下 LLM 比有限期界更合作，但忽略相对轮次位置，未出现末轮背叛，提示其按期界类别检索层级而非从具体博弈设定迭代推理。
AI推荐理由 arXiv 经济学预印本（econ.GN），作者 Po Han Teo，v2 修订于 2026-09-16。核心新信息：用 p-beauty contest 与公共品博弈两个实验，在 360 格因子设计下系统变化 temperature、模型规模（0.5-32B）、量化、instruct/base 与框架，并将每个部署设置的选择分布与已发表人类数据整体分布对比。结论指出除量化外各部署设置分别控制保真度的不同侧面；LLM 表现为静态、按类别检索的 level-k 玩家，k 由模型规模决定，且在多轮中不做博弈内信念更新或逆向归纳；公共品博弈中人类贡献衰减而 LLM 持平或上升，无限期界下更合作但忽略相对轮次位置、无末轮背叛。局限：仅限同一模型家族的本地模型，结论为行为学层面的分布匹配，不能外推为通用能力结论；预印本未经同行评审。
72
正文
Economics > General Economics
arXiv:2606.27845 (econ)
[Submitted on 26 Jun 2026 (v1), last revised 16 Sep 2026 (this version, v2)]
Title:LLM Agents as Static Level-k Players in Behavioural Games
Authors:Po Han Teo
View a PDF of the paper titled LLM Agents as Static Level-k Players in Behavioural Games, by Po Han Teo
View PDF
Abstract:Large Language Models (LLMs) are increasingly used as stand-ins in behavioural games. These stand-ins rely on the assumption that the LLM's distribution of choices meaningfully matches how humans play the same game. This study tests that assumption through two games. The first is a p-beauty contest, and the second one is a public goods game. The study first investigates five local-model settings within the same model family. These settings are varied together in a 360-cell factorial, which balances temperature, scale (0.5-32B), quantisation, instruct vs base, and framing. Each cell's distribution is then compared against whole choice distributions in published human data. Each deployment setting, except for quantisation, governs a different aspect of fidelity. Mechanically, while the dispersion of human players can be somewhat recovered through deployment settings, the strategic process behind it cannot. Through the lens of the level-k cognitive theory, we find that LLMs act as static, category-retrieved level-k players, where k is set by the model scale. The models also do not run within-game belief-updating or backward induction throughout multiple-round horizon settings. While human contributions decayed in the public goods game, LLMs stayed flat or rose at every scale. When the horizon test was administered, LLMs were more cooperative under an indefinite horizon compared to a finite one. However, LLMs ignore their relative round position, so no last-round defection was displayed. This implies that LLMs retrieved levels relative to the horizon category rather than working out iteratively from the specific game setting.
Comments:
Subjects:
General Economics (econ.GN); Theoretical Economics (econ.TH)
Cite as:
arXiv:2606.27845 [econ.GN]
(or
arXiv:2606.27845v2 [econ.GN] for this version)
https://doi.org/10.48550/arXiv.2606.27845
Focus to learn more
arXiv-issued DOI via DataCite
Submission history
From: Po Han Teo [view email]
[v1]
Fri, 26 Jun 2026 08:34:12 UTC (188 KB)
[v2]
Wed, 16 Sep 2026 13:53:46 UTC (193 KB)
Full-text links:
Access Paper:
View a PDF of the paper titled LLM Agents as Static Level-k Players in Behavioural Games, by Po Han Teo
View PDF
view license
Current browse context:
econ.GN
new
|
recent
| 2026-06
Change to browse by:
econ
econ.TH
q-fin
q-fin.EC
References & Citations
NASA ADS
Google Scholar
Semantic Scholar
export BibTeX citation
Loading...
BibTeX formatted citation
×
loading...
Data provided by:
Bookmark
Bibliographic Tools
Bibliographic and Citation Tools
Bibliographic Explorer Toggle
Bibliographic Explorer (What is the Explorer?)
Connected Papers Toggle
Connected Papers (What is Connected Papers?)
Litmaps Toggle
Litmaps (What is Litmaps?)
scite.ai Toggle
scite Smart Citations (What are Smart Citations?)
Code, Data, Media
Code, Data and Media Associated with this Article
alphaXiv Toggle
alphaXiv (What is alphaXiv?)
Links to Code Toggle
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub Toggle
DagsHub (What is DagsHub?)
GotitPub Toggle
Gotit.pub (What is GotitPub?)
Huggingface Toggle
Hugging Face (What is Huggingface?)
ScienceCast Toggle
ScienceCast (What is ScienceCast?)
Demos
Demos
Replicate Toggle
Replicate (What is Replicate?)
Spaces Toggle
Hugging Face Spaces (What is Spaces?)
Spaces Toggle
TXYZ.AI (What is TXYZ.AI?)
Related Papers
Recommenders and Search Tools
Link to Influence Flower
Influence Flower (What are Influence Flowers?)
Core recommender toggle
CORE Recommender (What is CORE?)
Author
Venue
Institution
Topic
About arXivLabs
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Which authors of this paper are endorsers? |
Disable MathJax (What is MathJax?)