IDEAS home Printed from https://ideas.repec.org/p/arx/papers/2405.19313.html
   My bibliography  Save this paper

Language Models Trained to do Arithmetic Predict Human Risky and Intertemporal Choice

Author

Listed:
  • Jian-Qiao Zhu
  • Haijiang Yan
  • Thomas L. Griffiths

Abstract

The observed similarities in the behavior of humans and Large Language Models (LLMs) have prompted researchers to consider the potential of using LLMs as models of human cognition. However, several significant challenges must be addressed before LLMs can be legitimately regarded as cognitive models. For instance, LLMs are trained on far more data than humans typically encounter, and may have been directly trained on human data in specific cognitive tasks or aligned with human preferences. Consequently, the origins of these behavioral similarities are not well understood. In this paper, we propose a novel way to enhance the utility of LLMs as cognitive models. This approach involves (i) leveraging computationally equivalent tasks that both an LLM and a rational agent need to master for solving a cognitive problem and (ii) examining the specific task distributions required for an LLM to exhibit human-like behaviors. We apply this approach to decision-making -- specifically risky and intertemporal choice -- where the key computationally equivalent task is the arithmetic of expected value calculations. We show that an LLM pretrained on an ecologically valid arithmetic dataset, which we call Arithmetic-GPT, predicts human behavior better than many traditional cognitive models. Pretraining LLMs on ecologically valid arithmetic datasets is sufficient to produce a strong correspondence between these models and human decision-making. Our results also suggest that LLMs used as cognitive models should be carefully investigated via ablation studies of the pretraining data.

Suggested Citation

  • Jian-Qiao Zhu & Haijiang Yan & Thomas L. Griffiths, 2024. "Language Models Trained to do Arithmetic Predict Human Risky and Intertemporal Choice," Papers 2405.19313, arXiv.org.
  • Handle: RePEc:arx:papers:2405.19313
    as

    Download full text from publisher

    File URL: http://arxiv.org/pdf/2405.19313
    File Function: Latest version
    Download Restriction: no
    ---><---

    References listed on IDEAS

    as
    1. Benjamin S. Manning & Kehang Zhu & John J. Horton, 2024. "Automated Social Science: Language Models as Scientist and Subjects," Papers 2404.11794, arXiv.org, revised Apr 2024.
    2. repec:cup:judgdm:v:12:y:2017:i:1:p:19-28 is not listed on IDEAS
    3. Stephan Meier & Charles Sprenger, 2010. "Present-Biased Preferences and Credit Card Borrowing," American Economic Journal: Applied Economics, American Economic Association, vol. 2(1), pages 193-210, January.
    4. Samuel J. Gershman & Rahul Bhui, 2020. "Rationally inattentive intertemporal choice," Nature Communications, Nature, vol. 11(1), pages 1-8, December.
    5. George Wu & Richard Gonzalez, 1996. "Curvature of the Probability Weighting Function," Management Science, INFORMS, vol. 42(12), pages 1676-1690, December.
    6. Tversky, Amos & Kahneman, Daniel, 1992. "Advances in Prospect Theory: Cumulative Representation of Uncertainty," Journal of Risk and Uncertainty, Springer, vol. 5(4), pages 297-323, October.
    7. Lisa Messeri & M. J. Crockett, 2024. "Artificial intelligence and illusions of understanding in scientific research," Nature, Nature, vol. 627(8002), pages 49-58, March.
    8. Benjamin S. Manning & Kehang Zhu & John J. Horton, 2024. "Automated Social Science: Language Models as Scientist and Subjects," NBER Working Papers 32381, National Bureau of Economic Research, Inc.
    9. Taylor Webb & Keith J. Holyoak & Hongjing Lu, 2023. "Emergent analogical reasoning in large language models," Nature Human Behaviour, Nature, vol. 7(9), pages 1526-1541, September.
    10. David Laibson, 1997. "Golden Eggs and Hyperbolic Discounting," The Quarterly Journal of Economics, President and Fellows of Harvard College, vol. 112(2), pages 443-478.
    11. Brenden M. Lake & Marco Baroni, 2023. "Human-like systematic generalization through a meta-learning neural network," Nature, Nature, vol. 623(7985), pages 115-121, November.
    Full references (including those not matched with items on IDEAS)

    Most related items

    These are the items that most often cite the same works as this one and are cited by the same works as this one.
    1. Olivier Toubia & Eric Johnson & Theodoros Evgeniou & Philippe Delquié, 2013. "Dynamic Experiments for Estimating Preferences: An Adaptive Method of Eliciting Time and Risk Parameters," Management Science, INFORMS, vol. 59(3), pages 613-640, June.
    2. Groneck, Max & Ludwig, Alexander & Zimper, Alexander, 2016. "A life-cycle model with ambiguous survival beliefs," Journal of Economic Theory, Elsevier, vol. 162(C), pages 137-180.
    3. Heutel, Garth, 2019. "Prospect theory and energy efficiency," Journal of Environmental Economics and Management, Elsevier, vol. 96(C), pages 236-254.
    4. Francisco Gomes & Michael Haliassos & Tarun Ramadorai, 2021. "Household Finance," Journal of Economic Literature, American Economic Association, vol. 59(3), pages 919-1000, September.
    5. Stephen L. Cheung, 2020. "Eliciting utility curvature in time preference," Experimental Economics, Springer;Economic Science Association, vol. 23(2), pages 493-525, June.
    6. Mark Schneider, 2018. "A Dual System Model of Risk and Time Preferences," Working Papers 18-18, Chapman University, Economic Science Institute.
    7. Abdellaoui, Mohammed & Kemel, Emmanuel & Panin, Amma & Vieider, Ferdinand M., 2019. "Measuring time and risk preferences in an integrated framework," Games and Economic Behavior, Elsevier, vol. 115(C), pages 459-469.
    8. Groneck, Max & Ludwig, Alexander & Zimper, Alexander, 2024. "Who saves more, the naive or the sophisticated agent?," Journal of Economic Theory, Elsevier, vol. 219(C).
    9. Mark Schneider, 2016. "Dual Process Utility Theory: A Model of Decisions Under Risk and Over Time," Working Papers 16-23, Chapman University, Economic Science Institute.
    10. Kelli L. Johnson & Michael T. Bixter & Christian C. Luhmann, 2020. "Delay discounting and risky choice: Meta-analytic evidence regarding single-process theories," Judgment and Decision Making, Society for Judgment and Decision Making, vol. 15(3), pages 381-400, May.
    11. Neszveda, G., 2019. "Essays on behavioral finance," Other publications TiSEM 05059039-5236-42a3-be1b-3, Tilburg University, School of Economics and Management.
    12. Mograbi, Eli, 2022. "Decision-makers are more impulsive on smartphones than on computers," Journal of Behavioral and Experimental Economics (formerly The Journal of Socio-Economics), Elsevier, vol. 100(C).
    13. Tomomi Tanaka & Colin F Camerer & Quang Nguyen, 2006. "Poverty, politics, and preferences: Field Experiments and survey data from Vietnam," Levine's Bibliography 122247000000001099, UCLA Department of Economics.
    14. repec:cup:judgdm:v:15:y:2020:i:3:p:381-400 is not listed on IDEAS
    15. Jinrui Pan & Craig S. Webb & Horst Zank, 2019. "Delayed probabilistic risk attitude: a parametric approach," Theory and Decision, Springer, vol. 87(2), pages 201-232, September.
    16. Nguyen, Quang, 2016. "Linking loss aversion and present bias with overspending behavior of tourists: Insights from a lab-in-the-field experiment," Tourism Management, Elsevier, vol. 54(C), pages 152-159.
    17. Stracca, Livio, 2004. "Behavioral finance and asset prices: Where do we stand?," Journal of Economic Psychology, Elsevier, vol. 25(3), pages 373-405, June.
    18. Robert Gazzale & Julian Jamison & Alexander Karlan & Dean Karlan, 2013. "Ambiguous Solicitation: Ambiguous Prescription," Economic Inquiry, Western Economic Association International, vol. 51(1), pages 1002-1011, January.
    19. Hinnosaar, Marit, 2016. "Time inconsistency and alcohol sales restrictions," European Economic Review, Elsevier, vol. 87(C), pages 108-131.
    20. Humphreys, Brad & Ruseski, Jane & Zhou, Li, 2015. "Physical Activity, Present Bias, and Habit Formation: Theory and Evidence from Longitudinal Data," Working Papers 2015-6, University of Alberta, Department of Economics.
    21. Shi, Yun & Cui, Xiangyu & Zhou, Xunyu, 2020. "Beta and Coskewness Pricing: Perspective from Probability Weighting," SocArXiv 5rqhv, Center for Open Science.

    More about this item

    NEP fields

    This paper has been announced in the following NEP Reports:

    Statistics

    Access and download statistics

    Corrections

    All material on this site has been provided by the respective publishers and authors. You can help correct errors and omissions. When requesting a correction, please mention this item's handle: RePEc:arx:papers:2405.19313. See general information about how to correct material in RePEc.

    If you have authored this item and are not yet registered with RePEc, we encourage you to do it here. This allows to link your profile to this item. It also allows you to accept potential citations to this item that we are uncertain about.

    If CitEc recognized a bibliographic reference but did not link an item in RePEc to it, you can help with this form .

    If you know of missing items citing this one, you can help us creating those links by adding the relevant references in the same way as above, for each refering item. If you are a registered author of this item, you may also want to check the "citations" tab in your RePEc Author Service profile, as there may be some citations waiting for confirmation.

    For technical questions regarding this item, or to correct its authors, title, abstract, bibliographic or download information, contact: arXiv administrators (email available below). General contact details of provider: http://arxiv.org/ .

    Please note that corrections may take a couple of weeks to filter through the various RePEc services.

    IDEAS is a RePEc service. RePEc uses bibliographic data supplied by the respective publishers.