Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat
Sign Up
Cited by 8

The DOI matches its Crossref record; 40 of the 40 references it cites were found in a registry or an index.

The machine psychology of cooperation: can GPT models operationalize prompts for altruism, cooperation, competitiveness, and selfishness in economic games?

Phelps, Steve, Russell, Yvan I·2025·Journal of Physics: Complexity
DOI 10.1088/2632-072x/ada711
Open this paper in CoChat — free account

Abstract

Large language models (LLMs) are capable of playing the ‘human’ role as participants in economic games. We investigated the capability of GPT-3.5 to play the one-shot dictator game (DG) and the repeated Prisoner’s Dilemma game (PDG), the latter of which introduced tit-for-tat scenarios. In particular, we investigated whether the LLMs could be prompted to play in accordance to five roles (‘personalities’) assigned prior to game play: the five ‘simulacra’ were: (1) cooperative, (2) competitive, (3) altruistic, (4) selfish, and (5) control, all of which were natural language descriptions (‘ruthless equities trader … ’, ‘selfless philanthropist … ’, etc). We predicted that the LLM-participant would play in accordance to the semantic content of the prompt (ruthless would play ruthlessly, etc). Across five simulacra (roles), we tested the AI equivalent of 450 human participants (32 400 observations in total, qua counterbalancing and re-testability). Using a general linear mixed model for the PDG, and a cumulative link mixed model for the DG, we found that level of cooperation/donation followed the general pattern of altruistic ⩾ cooperative > control > selfish ⩾ competitive. We proposed ten hypotheses, three of which were convincingly supported: cooperative/altruistic did cooperate more than competitive/selfish; cooperation was higher in repeated games (PDG); cooperative/altruistic were sensitive to the opponent’s behavior in repeated games. We also found some variation among the three versions of GPT-3.5 we used. Our study demonstrates the potential of using prompt engineering for LLM-chatbots to study the mechanisms of cooperation in both real and artificial worlds.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1088/2632-072x/ada711

Cites

40 of 40 references verified0 citation errors · 0 not checkable

Statistics

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 40 references

Research works printed but not linked: 101

  1. Dhar, 2017, A perspective on natural language understanding capability: an interview with Sam Bowman, Big Data, 5, 5, 10.1089/big.2017.29016.sbo · DOI
  2. Magnini, 2022, Understanding dialogue for human communication, 1159, 10.1007/978-3-031-03945-4_20 · DOI
  3. Paaß, 2023, 10.1007/978-3-031-23190-2 · DOI
  4. Burgos, 2022, Neural nets, 4634, 10.1007/978-3-319-55065-7_775 · DOI
  5. Khan, 2023, Exploring the frontiers of deep learning and natural language processing: a comprehensive overview of key challenges and emerging trends, Nat. Lang. Process. J., 4, 10.1016/j.nlp.2023.100026 · DOI
  6. Goldberg, 2016, A primer on neural network models for natural language processing, J. Artif. Intell. Res., 57, 345, 10.1613/jair.4992 · DOI · Paper page
  7. Wang, 2023, Pre-trained language models and their applications, Engineering, 25, 51, 10.1016/j.eng.2022.04.024 · DOI
  8. Gill, 2022, ChatGPT: vision and challenges, Int. Things Cyber-Phys. Syst., 3, 262, 10.1016/j.iotcps.2023.05.004 · DOI
  9. Dodge, 2008, Stochastic process, 521, 10.1007/978-0-387-32833-1_390 · DOI
  10. Lorè, 2024, Strategic behavior of large language models and the role of game structure versus contextual framing, Sci. Rep., 14, 10.1038/s41598-024-69032-z · DOI
  11. Koubaa, 2023, Exploring ChatGPT capabilities and limitations: a survey, IEEE Access, 11, 118698, 10.1109/ACCESS.2023.3326474 · DOI
  12. Lloyd, 2024, What is it like to be a bot? The world according to GPT-4, Front. Psychol., 15, 1, 10.3389/fpsyg.2024.1292675 · DOI
  13. Hacker, 2002, Is there anything it is like to be a bat?, Philosophy, 77, 157, 10.1017/S0031819102000220 · DOI
  14. Guthrie, 2013, Anthropomorphism, 111, 10.1007/978-1-4020-8265-8_1495 · DOI
  15. Warwick, 2022, Turing test, 7087, 10.1007/978-3-319-55065-7_477) · DOI
  16. Hoffman, 2022, Is AI intelligent? An assessment of artificial intelligence, 70 years after Turing, Technol. Soc., 68, 10.1016/j.techsoc.2022.101893 · DOI
  17. Skjuve, 2023, The user experience of ChatGPT: findings from a questionnaire study of early users, 1, 10.1145/3571884.3597144 · DOI
  18. Butlin, 2023, Sharing our concepts with machines, Erkenntnis, 88, 3079, 10.1007/s10670-021-00491-w · DOI
  19. Dung, 2023, Current cases of AI misalignment and their implications for future risks, Synthese, 202, 138, 10.1007/s11229-023-04367-0 · DOI
  20. Golden, 2023, ChatGPT: a trusted source?, Irish J. Technol. Enhanced Learn., 7, 113, 10.22554/ijtel.v7i2.129 · DOI
  21. Gabriel, 2024, Advancing equality: harnessing generative AI to combat systemic racism, 10.21428/e4baedd9.7dc53bbf · DOI
  22. Aftanas, 2018, Historical traces of a general measurement theory in psychology, Rev. Gen. Psychol., 22, 278, 10.1037/gpr0000143 · DOI
  23. Zamfirescu-Pereira, 2023, Why Johnny can’t prompt: how non-AI Experts try (and fail) to design LLM prompts, 1, 10.1145/3544548.3581388 · DOI · Paper page
  24. Rosenblueth, 1943, Behavior, purpose and teleology, Phil. Sci., 10, 18, 10.1086/286788 · DOI · Paper page
  25. Speed, 2024, Assessing the nature of large language models: a caution against anthropocentrism, 1, 10.2172/2429946 · DOI
  26. Taylor, 2021, Artificial cognition: how experimental psychology can help generate explainable artificial intelligence, Psychon. Bull. Rev., 28, 454, 10.3758/s13423-020-01825-5 · DOI
  27. Siemens, 2022, Human and artificial cognition, Comput. Educ. Artif. Intell., 3, 10.1016/j.caeai.2022.100107 · DOI · Paper page
  28. DiBlasi, 2022, Behaviorism, 752, 10.1007/978-3-319-55065-7_1230 · DOI
  29. Ma, 2023, Is ChatGPT humanly irrational?, Res. Sq., 1, 10.21203/rs.3.rs-3220513/v1 · DOI
  30. Azaria, 2023, ChatGPT: more human-like than computer-like, but not necessarily in a good way, 468, 10.1109/ICTAI59109.2023.00074 · DOI
  31. Trott, 2023, Do large language models know what humans know?, Cogn. Sci., 47, 10.1111/cogs.13309 · DOI · Paper page
  32. Engel, 2011, Dictator games: a meta study, Exper. Econ., 14, 583, 10.1007/s10683-011-9283-7 · DOI
  33. Vonk, 2022, Dictator game, 2016, 10.1007/978-3-319-55065-7_1360 · DOI
  34. Buchholz, 2021, Prisoner’s dilemma, 1, 10.1007/978-1-4614-7883-6_557-2 · DOI
  35. Peterson, 2015, 10.1017/CBO9781107360174 · DOI
  36. Traulsen, 2010, Human strategy updating in evolutionary games, Proc. Natl Acad. Sci. USA, 107, 2962, 10.1073/pnas.0912515107 · DOI
  37. Roth, 1995, Introduction to experimental economics, 3, 10.2307/j.ctvzsmff5.5 · DOI · Paper page
  38. Bal, 2020, Fairness, 1549, 10.1007/978-3-319-24612-3_2096 · DOI
  39. Apicella, 2019, The evolution of human cooperation, Curr. Biol., 29, R447, 10.1016/j.cub.2019.03.036 · DOI
  40. Fehr, 2004, Third-party punishment and social norms, Evol. Hum. Behav., 25, 63, 10.1016/S1090-5138(04)00005-4 · DOI · Paper page
  41. Young, 2018, Social norms, 12591, 10.1057/978-1-349-95189-5_2338 · DOI
  42. Rossetti, 2022, (Mis)perceiving cooperativeness, Curr. Opin. Psychol., 43, 151, 10.1016/j.copsyc.2021.06.020 · DOI
  43. Abrams, 2024, Behavioral economics and ChatGPT: from William Shakespeare to Elena Ferrante, Dig. Scholarsh. Human., 1, 10.1093/llc/fqae079 · DOI
  44. Johnson, 2024, Measuring an artificial intelligence language model’s trust in humans using machine incentives, J. Phys.: Complex., 5, 10.1088/2632-072X/ad1c69 · DOI
  45. Aher, 2023, Using large language models to simulate multiple humans and replicate human subject studies, Proc. Mach. Learn. Res., 202, 337
  46. Brookins, 2024, Playing games with GPT: what can we learn about a large language model from canonical strategic games?, Econ. Bull., 44, 25
  47. Masikisiki, 2023, Investigating the efficacy of large language models in reflective assessment methods through chain of thoughts prompting, 44, 10.1145/3628096.3628747 · DOI · Paper page
  48. Keister, 1996, A prisoner’s dilemma experiment on cooperation with people and human-like computers, J. Personal. Soc. Psychol., 70, 47, 10.1037/0022-3514.70.1.47 · DOI
  49. Bolker, 2009, Generalized linear mixed models: a practical guide for ecology and evolution, Trends Ecol. Evol., 24, 127, 10.1016/j.tree.2008.10.008 · DOI · Paper page
  50. Brooks, 2017, glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling, R J., 9, 378, 10.32614/RJ-2017-066 · DOI · Paper page
  51. Tutz, 1996, Random effects in ordinal regression models, Comput. Stat. Data Anal., 22, 537, 10.1016/0167-9473(96)00004-7 · DOI
  52. Hartig, 2022, DHARMa: residual diagnostics for hierarchical (multi-level / mixed) regression models, 10.32614/CRAN.package.DHARMa · DOI · Paper page
  53. Christensen, 2024, ordinal: regression models for ordinal data, 10.32614/CRAN.package.ordinal · DOI · Paper page
  54. Bates, 2015, Fitting linear mixed-effects models using lme4, J. Stat. Softw., 67, 1, 10.18637/jss.v067.i01 · DOI · Paper page
  55. Kassambara, 2023, rstatix: pipe-friendly framework for basic statistical tests, 10.32614/CRAN.package.rstatix · DOI
  56. Arel-Bundock, 2022, modelsummary: data and model summaries in R, J. Stat. Softw., 103, 1, 10.18637/jss.v103.i01 · DOI
  57. Hlaváč, 2022, stargazer: well-formatted regression and summary statistics tables, R package version 5.2.3, 10.32614/CRAN.package.stargazer · DOI · Paper page
  58. Angerer, 2023, repr: serializable representations, 10.32614/CRAN.package.repr · DOI
  59. Wickham, 2023, dplyr: a grammar of data manipulation, 10.32614/CRAN.package.dplyr · DOI · Paper page
  60. Dahl, 2019, xtable: export tables to LaTeX or HTML, R package version 1.8-4, 10.32614/CRAN.package.xtable · DOI
  61. Leifeld, 2013, texreg: conversion of statistical model output in R to LaTeX and HTML tables, J. Stat. Softw., 55, 1, 10.18637/jss.v055.i08 · DOI
  62. Højsgaard, 2006, The R package geepack for generalized estimating equations, J. Stat. Softw., 15, 1, 10.18637/jss.v015.i02 · DOI · Paper page
  63. Yan, 2004, Estimating equations for association structures, Stat. Med., 23, 859, 10.1002/sim.1650 · DOI
  64. Yan, 2002, geepack: yet another package for generalized estimating equations, R News, 2, 12
  65. Hadfield, 2010, MCMC methods for multi-response generalized linear mixed models: the MCMCglmm R package, J. Stat. Softw., 33, 1, 10.18637/jss.v033.i02 · DOI · Paper page
  66. Plummer, 2006, CODA: convergence diagnosis and output analysis for MCMC, R News, 6, 7
  67. Fox, 2018, Visualizing fit and lack of fit in complex regression models with predictor effect plots and partial residuals, J. Stat. Softw., 87, 1, 10.18637/jss.v087.i09 · DOI · Paper page
  68. Fox, 2003, Effect displays in R for generalised linear models, J. Stat. Softw., 8, 1, 10.18637/jss.v008.i15 · DOI · Paper page
  69. Fox, 2010, Effect displays in R for multinomial and proportional-odds logit models: extensions to the effects package, J. Stat. Softw., 32, 1, 10.18637/jss.v032.i01 · DOI · Paper page
  70. Lüdecke, 2018, ggeffects: tidy data frames of marginal effects from regression models, J. Open Source Softw., 3, 772, 10.21105/joss.00772 · DOI · Paper page
  71. Auguie, 2017, gridExtra: miscellaneous functions for ‘grid’ graphics, 10.32614/CRAN.package.gridExtra · DOI · Paper page
  72. Wickham, 2023, purrr: functional programming tools, 10.32614/CRAN.package.purrr · DOI · Paper page
  73. Wickham, 2021, memoise: ‘memoisation’ of functions, 10.32614/CRAN.package.memoise · DOI
  74. Bengtsson, 2022, R.cache: fast and light-weight caching (memoization) of objects and results to speed up computations, R package version 0.16.0, 10.32614/CRAN.package.R.cache · DOI
  75. Pedersen, 2023, patchwork: the composer of plots, 10.32614/CRAN.package.patchwork · DOI · Paper page
  76. Kassambara, 2023, ggpubr: ‘ggplot2’ based publication ready plots, 10.32614/CRAN.package.ggpubr · DOI · Paper page
  77. Kalinowski, 2023, reticulate: interface to ‘Python’, 10.32614/CRAN.package.reticulate · DOI
  78. Carruthers, 2008, Generalized linear models: model selection, diagnostics, and overdispersion, CiteSeerX, 1
  79. Kim, 2015, How to choose the level of significance: a pedagogical note, Munich Pers. RePEc Arch., 69992, 1
  80. Kim, 2021, Choosing the level of significance: a decision-theoretic approach, Abacus, 57, 27, 10.1111/abac.12172 · DOI
  81. Balluerka, 2005, The controversy over null hypothesis significance testing revisited, Methodol., 1, 55, 10.1027/1614-1881.1.2.55 · DOI
  82. Grühn, 2016, An English word database of EMOtional TErms (EMOTE), Psychol. Rep., 119, 290, 10.1177/0033294116658474 · DOI
  83. Smith, 1976, Experimental economics: induced value theory, Am. Econ. Rev., 66, 274
  84. Read, 2005, Monetary incentives, what are they good for?, J. Econ. Methodol., 12, 265, 10.1080/13501780500086180 · DOI
  85. Bonfrisco, 2025, Averting depletion in a two-player common pool resource game: being seen, the expectation of future encounters and biophilia play a role in cooperation, Dyn. Games Appl., 15, 1, 10.1007/s13235-024-00557-8 · DOI
  86. Awan, 2020, Talking the talk, but not walking the walk: a comparison of self-reported and observed prosocial behaviour, Publ. Adm., 98, 995, 10.1111/padm.12664 · DOI
  87. Roszczyńska-Kurasińska, 2013, The dynamics of trust from the perspective of a trust game, 191, 10.1007/978-3-642-31436-0_11 · DOI
  88. Tan, 2010, When happiness makes us selfish, but sadness makes us fair: affective influences on interpersonal strategies in the dictator game, J. Exper. Soc. Psychol., 46, 571, 10.1016/j.jesp.2010.01.007 · DOI
  89. Dal Bó, 2018, On the determinants of cooperation in infinitely repeated games: a survey, J. Econ. Lit., 56, 60, 10.1257/jel.20160980 · DOI · Paper page
  90. Woo, 2022, Human morality is based on an early-emerging moral core, Annu. Rev. Developmental Psychol., 4, 41, 10.1146/annurev-devpsych-121020-023312 · DOI
  91. Conitzer, 2024, Foundations of cooperative AI, 15359, 10.1609/aaai.v37i13.26791 · DOI · Paper page
  92. Gómez, 2022, Comparative psychology, 1569, 10.1007/978-3-319-55065-7_1804 · DOI
  93. Panic, 2024, 10.1201/9781003359982 · DOI
  94. Hancock, 2020, The humanity of humanless systems, Ergon. Des., 28, 4, 10.1177/1064804619880047 · DOI
  95. Lafleche, 2011, Anterograde amnesia, 191, 10.1007/978-0-387-79948-3_1106 · DOI
  96. Chang, 2023, Examining GPT-4’s capabilities and enhancement with SocraSynth, 7, 10.1109/CSCI62032.2023.00009 · DOI
  97. Fjelland, 2024, Computers will not acquire general intelligence, but may still rule the world, Cosmos+Taxis, 12, 58
  98. Gracia-Lázaro, 2012, Hetereogeneous networks do not promote cooperation when humans play a prisoner’s dilemma, Proc. Natl Acad. Sci. USA, 109, 12922, 10.1073/pnas.1206681109 · DOI · Paper page
  99. Russell, 2020, Cooperation through image scoring: a replication, Games, 11, 58, 10.3390/g11040058 · DOI
  100. Hancock, 2022, Avoiding adverse autonomous agent actions, Human-Comput. Interaction, 37, 211, 10.1080/07370024.2021.1970556 · DOI
  101. Chincarini, 2003, Experimental evidence of trigger strategies in repeated games, SSRN Electron. J., 417540, 1, 10.2139/ssrn.417540 · DOI

Cited by

Metadata

Authors
Phelps, S., & Russell, Y. I.
Venue
Journal of Physics: Complexity
Publisher
IOP Publishing
Published
2025-03-01
Volume
6
Issue
1
Pages
015018
ISSN
2632-072X
DOI
10.1088/2632-072x/ada711
Cited by
8
References
40
Licence

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.