Discovering Differences in Strategic Behavior Between Humans and LLMs

As Large Language Models (LLMs) are increasingly deployed in social and strategic scenarios, it becomes critical to understand where and why their behavior diverges from that of humans. While behavioral game theory (BGT) provides a framework for analyzing behavior, existing models do not fully capture the idiosyncratic behavior of humans or black-box, non-human agents like LLMs. We employ AlphaEvolve, a cutting-edge program discovery tool, to directly discover interpretable models of human and LLM behavior from data, thereby enabling open-ended discovery of structural factors driving human and LLM behavior. Our analysis on iterated rock-paper-scissors reveals that frontier LLMs can be capable of deeper strategic behavior than humans. These results provide a foundation for understanding structural differences driving differences in human and LLM behavior in strategic interactions.

• Publication year

  2026

• Venue

  Unknown venue

• Publication date

  2026-02-10

• Fields of study

  Computer Science, Psychology

• Identifiers
  arXiv 2602.10324
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• To Mask or to Mirror: Human-AI Alignment in Collective Reasoning

  2025influential reference
• Strategic Tradeoffs Between Humans and AI in Multi-Agent Bargaining

  2025influential reference
• Evaluating the ability of large Language models to predict human social decisions

  2025cited by this paper
• AI Testing Should Account for Sophisticated Strategic Behaviour

  2025cited by this paper
• Behavioral Economics

  2025cited by this paper
• Game Theory Meets Large Language Models: A Systematic Survey with Taxonomy and New Frontiers

  2025cited by this paper
• Game Reasoning Arena: A Framework and Benchmark for Assessing Reasoning Capabilities of Large Language Models via Game Play

  2025cited by this paper
• Strategic Intelligence in Large Language Models: Evidence from evolutionary Game Theory

  2025cited by this paper
• Reasoning Models Don't Always Say What They Think

  2025cited by this paper
• Language Agents as Digital Representatives in Collective Decision-Making

  2025cited by this paper
• Discovering Symbolic Cognitive Models from Human and Animal Behavior

  2025influential reference
• Generating Computational Cognitive Models using Large Language Models

  2025cited by this paper
• ALYMPICS: LLM Agents Meet Game Theory

  2025cited by this paper
• Neural decoding of competitive decision-making in Rock–Paper–Scissors

  2025cited by this paper
• Take caution in using LLMs as human surrogates

  2025cited by this paper
• Can Large Language Model Agents Simulate Human Trust Behavior?

  2024cited by this paper
• Repeated rock, paper, scissors play reveals limits in adaptive sequential behavior.

  2024cited by this paper
• Generative Agent Simulations of 1,000 People

  2024cited by this paper
• Centaur: a foundation model of human cognition

  2024cited by this paper
• CogBench: a large language model walks into a psychology lab

  2024cited by this paper
• An Experimental Study of Competitive Market Behavior Through LLMs

  2024influential reference
• Large Language Models and Cognitive Science: A Comprehensive Review of Similarities, Differences, and Challenges

  2024cited by this paper
• Capturing the complexity of human strategic decision-making with machine learning

  2024cited by this paper
• Learning to be Homo Economicus: Can an LLM Learn Preferences from Choice

  2024cited by this paper
• How Well Can LLMs Negotiate? NegotiationArena Platform and Analysis

  2024cited by this paper
• Hypothetical Minds: Scaffolding Theory of Mind for Multi-Agent Tasks with Large Language Models

  2024cited by this paper
• Using games to understand the mind

  2024influential reference
• Nicer Than Humans: How do Large Language Models Behave in the Prisoner's Dilemma?

  2024cited by this paper
• Decision-Making Behavior Evaluation Framework for LLMs under Uncertain Context

  2024cited by this paper
• Do Large Language Models Learn Human-Like Strategic Preferences?

  2024cited by this paper
• Testing theory of mind in large language models and humans

  2024influential reference
• Assessing Large Language Models' ability to predict how humans balance self-interest and the interest of others

  2023cited by this paper
• Large Language Models as Simulated Economic Agents: What Can We Learn from Homo Silicus?

  2023cited by this paper
• Population-based Evaluation in Repeated Rock-Paper-Scissors as a Benchmark for Multiagent Reinforcement Learning

  2023cited by this paper
• Inverse Generative Social Science: Backward to the Future

  2023cited by this paper
• Playing repeated games with large language models

  2023cited by this paper
• Cognitive Model Discovery via Disentangled RNNs

  2023cited by this paper
• EconAgent: Large Language Model-Empowered Agents for Simulating Macroeconomic Activities

  2023cited by this paper
• Can Large Language Models Serve as Rational Players in Game Theory? A Systematic Analysis

  2023influential reference
• Mathematical discoveries from program search with large language models

  2023cited by this paper
• Using Large Language Models to Simulate Multiple Humans and Replicate Human Subject Studies

  2022cited by this paper
• Why Do We Turn to Virtual Companions? A Text Mining Analysis of Replika Reviews

  2022cited by this paper
• Neural Theory-of-Mind? On the Limits of Social Intelligence in Large LMs

  2022cited by this paper
• Social Simulacra: Creating Populated Prototypes for Social Computing Systems

  2022cited by this paper
• Theory of Games and Economic Behavior: 60th Anniversary Commemorative Edition

  2020cited by this paper
• Level-0 Models for Predicting Human Behavior in Games

  2019cited by this paper
• Behavior in Strategic Settings: Evidence from a Million Rock-Paper-Scissors Games

  2019cited by this paper
• Mastering the game of Go with deep neural networks and tree search

  2016cited by this paper
• Negative outcomes evoke cyclic irrational decisions in Rock, Paper, Scissors

  2016cited by this paper
• Deep Learning for Predicting Human Strategic Behavior

  2016cited by this paper
• Human-level control through deep reinforcement learning

  2015cited by this paper
• Program

  2014cited by this paper
• Social cycling and conditional responses in the Rock-Paper-Scissors game

  2014cited by this paper
• On the Properties of Neural Machine Translation: Encoder–Decoder Approaches

  2014cited by this paper
• Cycle frequency in standard Rock-Paper-Scissors games: Evidence from experimental economics

  2013cited by this paper
• A widely applicable Bayesian information criterion

  2012cited by this paper
• Trial-by-trial data analysis using computational models

  2011cited by this paper
• Winning at Rock-Paper-Scissors

  2011influential reference
• Neural mechanisms of the cognitive model of depression

  2011cited by this paper
• Understanding Neural Coding through the Model-Based Analysis of Decision Making

  2007cited by this paper
• A Cognitive Hierarchy Model of Games

  2004cited by this paper
• Behavioral Game Theory: Experiments in Strategic Interaction

  2003cited by this paper
• Sophisticated Experience-Weighted Attraction Learning and Strategic Teaching in Repeated Games

  2002influential reference
• Experience‐weighted Attraction Learning in Normal Form Games

  1999influential reference
• Quantal Response Equilibria for Normal Form Games

  1995cited by this paper
• On Players' Models of Other Players: Theory and Experimental Evidence

  1995cited by this paper
• Does the chimpanzee have a theory of mind?

  1978cited by this paper
• Depression: Clinical, Experimental, and Theoretical Aspects

  1968cited by this paper

• No citing papers are available for this paper.