Using Cognitive Models to Improve Language Model Simulation of Human Persuasion Games

2026-06-16

AI Summary

This study addresses the challenge that large language models (LLMs) struggle to accurately simulate the diverse decision-making behaviors exhibited by humans in strategic interactions—such as persuasion games—particularly when these behaviors involve cognitive biases like motivated reasoning and Bayesian updating. To overcome this limitation, the authors propose Equation-to-Behavior Prompting, a novel approach that integrates formal cognitive models (e.g., Grether’s α-β model and affine distortion functions) into LLMs via prompting or rule-based reinforcement learning, enabling controllable simulation of multiple human decision mechanisms. Experimental results demonstrate that large models can approximate various cognitive models through prompting alone, while smaller models trained with Equation-to-Behavior reinforcement learning reduce belief prediction error by 26.5% under out-of-distribution parameters. Furthermore, training across diverse environments yields additional improvements of 2.5%–12% in modeling belief change.

Abstract

People make decisions differently in strategic interactions. Some update beliefs like a Bayesian; others exhibit biases like motivated reasoning. Although creators of large language models use simulated humans for safety evaluations and training, they often fail to cover this breadth of human behavior. We argue that cognitive science and economics provide a convenient tool for doing so, making use of mathematical models of human decision-making. We propose an approach that we call Equation-to-Behavior Prompting for guiding large language models to match cognitive models, and evaluate this approach on persuasion games based on legal decision-making. We find that large models can approximate equation-based specifications -- Bayesian updating, affine distortion, motivated updating, and Grether's $α$-$β$ model -- using prompting, but small models fail to do so. However, training small models with reinforcement learning to adhere to mathematical rules, Equation-to-Behavior RL, reduces belief error by 26.5% in out-of-distribution parameterizations. We show that these simulations can help create diverse training environments; training small models to consider different kinds of decision-makers improves average belief change by 2.5%--12% over Bayesian-only training, even when persuading GPT-5-mini. Our work could improve human simulations for training and evaluation in increasingly realistic settings, and could also enable novel research into more complicated mathematical models of human decision-making.

Problem

• human persuasion games
• cognitive models
• language model simulation
• decision-making biases
• belief updating

Innovation

• Equation-to-Behavior Prompting
• cognitive modeling
• language model simulation
• reinforcement learning
• human decision-making