Can Large Language Model Agents Simulate Human Trust Behavior?
Open accessThe DOI matches its Crossref record; 17 of the 17 references it cites were found in a registry or an index.
Abstract
Large Language Model (LLM) agents have been increasingly adopted as simulation tools to model humans in social science and role-playing applications. However, one fundamental question remains: can LLM agents really simulate human behavior? In this paper, we focus on one critical and elemental behavior in human interactions, trust, and investigate whether LLM agents can simulate human trust behavior. We first find that LLM agents generally exhibit trust behavior, referred to as agent trust, under the framework of Trust Games, which are widely recognized in behavioral economics. Then, we discover that GPT-4 agents manifest high behavioral alignment with humans in terms of trust behavior, indicating the feasibility of simulating human trust behavior with LLM agents. In addition, we probe the biases of agent trust and differences in agent trust towards other LLM agents and humans. We also explore the intrinsic properties of agent trust under conditions including external manipulations and advanced reasoning strategies. Our study provides new insights into the behaviors of LLM agents and the fundamental analogy between LLMs and humans beyond value alignment. We further illustrate broader implications of our discoveries for applications where trust is paramount.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Multi-agent simulations show that LLM agents exhibit key social traits like trust and coordination54,55;”
From Understanding large language models demands distinguishing human projection from machine cognition (2026) · cites this paper as [54]
Citations
Cites
The publisher hasn’t deposited this paper’s reference list.
Cited by
Items 101–161 of 161
Will LLMs Sink or Swim? Exploring Decision-Making Under Pressure
Kim, K., Jeon, H., Ryu, J., & Suh, B. (2024) · Findings of the Association for Computational Linguistics: EMNLP 2024 · doi:10.18653/v1/2024.findings-emnlp.668 · Cited by 1
Simulating Human Decision-Making in Ultimatum Games using Large Language Models
MURASHIGE, T., & Itō, T. (2025) · ACM Collective Intelligence Conference (CI) · doi:10.1145/3715928.3737473 · Cited by 1
A Linguistic Negotiation for Consensus-Making among LLM Agents with Fuzzy Utility Functions
Yihan, D. O. N. G., & Itō, T. (2026) · IEICE Transactions on Information and Systems · doi:10.1587/transinf.2025edp7179 · Cited by 1
Computational Basis of Large Language Models’ Decision Making in Social Simulation
Ma, J. (2026) · Sociological Methodology · doi:10.1177/00811750261421220 · Cited by 1
Li, D., & Zhu, W. (2025) · International Conference on Artificial Intelligence and Computer Engineering (ICAICE) · doi:10.1109/icaice68195.2025.11382423 · Cited by 1
Automation 5.0: The Step to Systems Intelligence for a Sustainable Future
Yang, J., Dotoli, M., Wang, Y., Wang, X., Liu, Y., Tian, Y., Ge, J., Ni, Q., Carli, R., Süli, P. P., et al. (2026) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2026.3668032 · Cited by 1
“These AI-driven agents integrate the acquisition, processing, analysis, and application of knowledge into a seamless workflow centered around large-scale models [41], [42], making knowledge automation increasingly feasible in real-world applications [15].”
Can Machines Think Like Humans: A Behavioral Evaluation of LLM Agents in Dictator Games
Ma, J. (2026) · Voluntas: International Journal of Voluntary and Nonprofit Organizations · doi:10.1017/s0957876526000173 · Cited by 1
Lee, J., & Rew, J. (2026) · Computer Modeling in Engineering & Sciences · doi:10.32604/cmes.2026.076888 · Cited by 2
Rethinking UX in the age of AI: DARPA’s Heilmeier Catechism at the interface 1
Prietula, M. J., O’Leary, D. E., & Swartout, W. (2026) · Journal of Decision System · doi:10.1080/12460125.2026.2655311 · Cited by 2
Propensity to trust in Large Language Models
Plebe, A. (2026) · PLoS ONE · doi:10.1371/journal.pone.0347328 · Cited by 1
“[82], for example, study LLM behavior in the classical Trust Game and report that advanced models display patterns resembling human trust decisions, adjusting their behavior in response to perceived risk and potential reciprocity.”
YSocial: An artificial intelligence powered social media virtual twin
Rossetti, G., Stella, M., Cazabet, R., Abramski, K., Citraro, S., Cau, E., Failla, A., Morini, V., & Pansanella, V. (2026) · Big Data & Society · doi:10.1177/20539517261431576 · Cited by 4
Large language models for computational social science
Xu, F., Shao, C., Chen, L., Zeng, Q., Chen, Z., Sukiennik, N., Wang, J., Gao, C., Wang, H., Lian, J., et al. (2026) · Nature Human Behaviour · doi:10.1038/s41562-026-02539-9 · Cited by 1
Self-citationLiao, J., Feng, Y. Z., Zheng, Y., Zhao, J., Wang, S., & Zheng, J. (2025) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2025.acl-long.787 · Cited by 1
Xiao, X., Chen, P., Jia, Q., Tong, J., Liang, J., & Wang, H. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2502.00022 · Cited by 1
Computational Basis of LLM's Decision Making in Social SimulationPreprint
Ma, J. (2025) · RePEc: Research Papers in Economics · doi:10.17605/osf.io/j9rqk · Cited by 1
Zhang, J. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2512.24415 · Cited by 1
Wang, J., Zhang, Y., Borhi, M., Wei, W., Li, H., Hémono, R., Giorgio, J., Su, J., Ashe, K., Zheng, Z., et al. (2026) · Research Square · doi:10.21203/rs.3.rs-8823969/v1 · Cited by 1
Proceedings of the 2nd Workshop on Advancing Artificial Intelligence through Theory of MindPreprint
Alon, N., Barnby, J. M., Mirsky, R., & Sarkadi, Ş. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2603.18786 · Cited by 1
Collective Opinion Dynamics in Structured LLM PopulationsPreprint
Cau, E., Failla, A., & Rossetti, G. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.11312 · Cited by 1
AgentSchool: An LLM-Powered Multi-Agent Simulation for EducationPreprint
Ye, Y., Li, W., Wen, Z., Huang, Y., Hu, Y., Wei, Z., Wang, Y., Xie, X., Yang, H., Huang, Y., et al. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.30144 · Cited by 1
“It has been widely used in education (Gu and Blackmore, 2015; Bodine et al., 2020), medicine (Li et al., 2025; Hicke et al., 2025), economics (Xie et al., 2024), and sociology (Mi et al., 2025).”
InfluentialUnderstanding large language models demands distinguishing human projection from machine cognition
Li, L., Teng, Y., Wang, Y., & Hu, X. (2026) · Communications Psychology · doi:10.1038/s44271-026-00508-6
“Recent critiques cautioned that treating model outputs as interchangeable with human behavioral data without verifying the underlying cognitive equivalence 44,54,76 may compromise the reproducibility of such results in human participants and their overall reliability 77,78 .”
What does AI consider praiseworthy?
Peterson, A. (2025) · AI and Ethics · doi:10.1007/s43681-025-00682-z · Cited by 1
Unsupervised and Interpretable Detection of User Personalities in Online Social Networks
Cascione, A., Pollacci, L., & Guidotti, R. (2025) · Communications in computer and information science · doi:10.1007/978-3-032-08327-2_8
Epitome: Pioneering an Experimental Platform for AI-Social Science IntegrationPreprint
Qu, J., Hu, K., Zhu, J., Ye, Y., Li, W., Wang, T., Chen, Z., Lu, C., Zhou, A., Wang, X., et al. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2507.01061
“By integrating AI-driven complexity management, Epitome significantly improves research efficiency and methodological rigor in social science studies.”
Self-citationOutraged AI: Large language models prioritise emotion over cost in fairness enforcementPreprint
Wu, Z., Liu, H., Dai, Y., Tan, H., Lei, Y., & Zhou, Y. (2025) · Research Square · doi:10.21203/rs.3.rs-7830884/v1
Exploration on Demand: From Algorithmic Control to User EmpowermentPreprint
Bianchi, E. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2507.21884
“These approaches build on broader research demonstrating that LLMs can effectively approximate human behavioral patterns and decision-making processes across various domains [13], [14].”
Bunjamin, C. (2025) · American journal of student research. · doi:10.70251/hyjr2348.36781790
Psychometric Comparability of LLM-Based Digital TwinsPreprint
Zhang, Y., & Ma, Z. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.5967255
Language Models Exhibit Inconsistent Biases Towards Algorithmic Agents and Human Experts
Bo, J. Y., Mok, L., & Anderson, A. (2026) · Proceedings of IASEAI Conference · doi:10.1609/iaseai.v2i1.43015
Jiang, G., Li, J., Ma, X., & Luo, Q. (2026) · ACM International Conference on Intelligent User Interfaces (IUI) · doi:10.1145/3742413.3789094
PULSE – Polling Using LLM-based Sentiment Extraction
Karanikolopoulos, C., Papadakos, P., & Tsaparas, P. (2025) · IEEE ... International Conference on Data Mining workshops · doi:10.1109/icdmw69685.2025.00336
“…to produce human-like text has attracted significant research interest in building agents that can seamlessly interact with humans [1], [2], as well as in simulating human behavior across different settings, including games, cognitive tests, opinion surveys and social network interactions [3]–[8].”
Jing, G., Pei, W., Yinzhe, M., Luxi, C., Ke, G., Yanxi, H., & He, L. (2026) · Advances in Psychological Science · doi:10.3724/sp.j.1042.2026.1072
Cohen, M. C., Zheng, M., Bhandari, N., Kao, H.-T., Zhou, X., Nguyen, D., Cassani, L., Sap, M., & Volkova, S. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.15607
“However, Xie et al. (2024) examined whether LLMs replicate both actions and underlying reasoning, finding high alignment in GPT-4’s emulation of trust behaviors present in imperfectly cooperative social norms.”
From Per-Timestep Deciders to Holistic Strategy Generators: Evolving Strategic Complexity In LLMS
Li, S., & Wang, X. (2026) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp55912.2026.11460324
Pataranutaporn, P., Albrecht, C., Archiwaranguprok, C., Karny, S., & Baez, A. (2026) · Augmented Humans International Conference (AHs) · doi:10.1145/3795011.3795070
Distorted Perspectives of LLM-Simulated Preferences: Can AI Mislead Design?Preprint
Kuric, E., Demcak, P., & Krajcovic, M. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.18311
“Simulation literature refers to this type of accuracy using the term algorithmic fidelity (Jia et al., 2024).”
Takenami, Y., Huang, Y. J., Murawaki, Y., Takeuchi, K., & Chu, C. (2026) · Journal of Natural Language Processing · doi:10.5715/jnlp.33.473
Synthetic Audiences as Proxies for Human Users: A Systematic Literature Review
Lappas, T., & Filippas, A. (2026) · IEEE Access · doi:10.1109/access.2026.3703706
“5 reproduces only 37.5% of Many Labs 2 effects [53], whereas GPT-4 successfully replicates canonical results from economic and social psychology experiments [5] and closely matches human behavior in trust games [54].”
InfluentialYang, H., Zhang, Y., Ramani, V., Ionescu, S., & Sonta, A. J. (2026) · ACM Sustainability Week · doi:10.1145/3765611.3815494
Institutions as cached computation for resource-rational negotiation
Huang, M., & Evans, J. A. (2026) · Behavioral and Brain Sciences · doi:10.1017/s0140525x2510174x
Self-citationСравнительный анализ поведения больших языковых моделей и людей в игре «Диктатор»
Андреевна, Ш. Е., Parshakov, P. A., & Маткин, Н. А. (2026) · Journal of the New Economic Association · doi:10.31737/22212264_2026_2_177-203
Chinese Court Simulation with LLM-Based Agents System
Zhang, K., Li, J., Wu, Y., Li, H., Luo, C., Zou, S., Zhou, Y., Su, W., Liu, Y., & Ai, Q. (2026) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2026.findings-acl.411 · Cited by 2
Cohen, M. C., Zheng, M., Bhandari, N., Kao, H.-T., Zhou, X., Nguyen, D., Cassani, L., Sap, M., & Volkova, S. (2026) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2026.findings-acl.1340 · Cited by 1
AI Agents for the Science of Science: A Survey of Tasks, Architectures, Evaluations, and Challenges
Liu, Y., & Zhang, Y. (2026) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2026.findings-acl.1804
“Consequently, AI agents can model micro-level human-like decision processes at scale, enabling more flexible analyses of macro-level scientific structures (Xie et al., 2024).”
Cakar, M., & Fischer, M. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-31048-4_25
Chen, K. (2026) · Contemporary systems thinking · doi:10.1007/978-3-032-18066-7_1
The self-positivity bias of large language models
Peng, Z. (2026) · Scientific Reports · doi:10.1038/s41598-026-63768-6 · Cited by 1
Exploration on Demand: From Algorithmic Control to User Empowerment
Bianchi, E. (2026) · International Conference on Control, Decision and Information Technologies (CoDIT) · doi:10.1109/codit70676.2026.11631067
“These approaches build on broader research demonstrating that LLMs can effectively approximate human behavioral patterns and decision-making processes across various domains [13], [14].”
How a shared state is described determines whether AI agents synchronizePreprint
Ezaki, T., Imura, N., & Nishinari, K. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2608.06968
Kim, M., Wang, H., & Kang, Y. (2026) · ASCE Construction Research Congress (CRC) · doi:10.1061/9780784486986.084
CompanionSim: Synthetic Data for Evaluating Anthropomorphism in Human-AI Relationships
Anthis, J. R., Díaz, M., & Shelby, R. (2026) · Proceedings of the AAAI/ACM Conference on AI Ethics and Society · doi:10.1609/aies.v9i1.43135
“…design choices can foster one-sided attachments and psychological dependence, particularly when chatbots present themselves as stable friends, mentors, or romantic partners and respond to user distress with highly emotive language (Xie et al. 2024; Maeda and Quan-Haase 2024; Zhang et al. 2025).”
Shen, A., Xu, Z., Ma, Z., Yin, M., Lian, B., Chen, Y., Fan, H., & Chai, C. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7478078
PsyAgentRec: Psychology-Grounded Agents for Adaptive Recommendation
Zhong, H., & Zhu, S. (2026) · International Conference on Intelligent Systems, Metaheuristics & Swarm Intelligence (ISMSI) · doi:10.1145/3820820.3820821
PowerBench: Measuring Language Model Bias in Power-shifting RequestsPreprint
Martorell, N., Brau, W., Heredia, G. A., Korenblit, T. P., Labastié, G., & Molina, T. G. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2610.02303
LLMA-UML: Extending the UML for Modeling LLM-Agent Systems
Walter, L., Rother, K. M., & Henkler, S. (2026) · ACM/IEEE International Conference on Model Driven Engineering Languages and Systems (MODELS) · doi:10.1145/3837062.3838894
Benchmarking Psychological Dynamics in Generative AgentsPreprint
Vaid, S. S., & Whillans, A. V. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2610.04246
“Although constrained repeated-game scenarios show initial promise in following path-dependent variables like trust (Xie et al., 2024), discrete prompt engineering can stabilize specific value alignments (Rozen et al., 2025), and multi-turn reinforcement learning can reduce persona inconsistency…”
Peng, S., Jiang, K., Zheng, J., Yang, Q., & Chen, J. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2610.08250
“A complementary line of work develops benchmarks and protocols for evaluating role-playing consistency (Liu et al. 2025; Xiao et al. 2025; Xie et al. 2024).”
Social Priors: Descriptions of AI Users Shape How Language Models Trust and CooperatePreprint
Kumar, H., Eilender, G., & Anderson, A. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2610.16946
The Long Con on the Web: Earn-then-Exploit Fraud Attacks against Coding AgentsPreprint
Xu, X., Jiang, Z., Zhi, Z., Dong, F., Guo, L., Dong, W. F., Ma, C., & Jiang, J. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2611.00725
LLM Agents for Simulating Human Stated-Choice Behavior
Wang, Z., Kim, M.-K., Wang, R., & Cheng, L. (2026) · International Conference on Machine Learning and Applications · doi:10.1109/icmla69890.2026.00093
Integrated Guardrails for Unbiased and Adaptive Neural Network Architectures
Kathiravelu, P., & Grbac, T. G. (2026) · IEEE International Symposium on Systems Engineering · doi:10.1109/isse70179.2026.11762779
Metadata
- Authors
- Xie, C., Chen, C., Jia, F., Ye, Z., Lai, S., Shu, K., Gu, J., Bibi, A., Hu, Z., Jurgens, D., Evans, J., Torr, P., Ghanem, B., & Li, G.
- Venue
- Advances in Neural Information Processing Systems 37
- Published
- 2024
- Pages
- 15674-15729
- ISSN
- 1049-5258 (ISSN-L)
- OpenAlex
- W4415798544
- Semantic Scholar
- 8b1bed871bdddbd2b2df89e962dbe4d7cf58d62b
- Cited by
- 31
- References
- 0
- Citations by year
- 2026: 23 · 2025: 5 · 2024: 3
- FWCI
- 20.83
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- cc-by
- Locations
- neural information processing systems (publishedVersion)
- arXiv (Cornell University) (submittedVersion, cc-by)
- arXiv (Cornell University)
- Funders
- Emory University
- California Institute of Technology
- University of Chicago
- Pennsylvania State University
- UK Research and Innovation (EP/W002981/1)
- Royal Academy of Engineering
- University of Oxford
- King Abdullah University of Science and Technology
- Multidisciplinary University Research Initiative (EP/N019474/1)
- Engineering and Physical Sciences Research Council (EP/W002981/1, EP/N019474/1)
- Topics
- Social Sciences › Psychology › Social Psychology › Social Robot Interaction and HRI
- Physical Sciences › Computer Science › Artificial Intelligence › Multi-Agent Systems and Negotiation
- Physical Sciences › Computer Science › Computer Networks and Communications › Access Control and Trust
- Keywords
- LLM agents, trust behavior, human-agent interaction, trustworthiness
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.