Research

Paper

AI LLM March 19, 2026

Evaluating Counterfactual Strategic Reasoning in Large Language Models

Authors

Dimitrios Georgousis, Maria Lymperaiou, Angeliki Dimitriou, Giorgos Filandrianos, Giorgos Stamou

Abstract

We evaluate Large Language Models (LLMs) in repeated game-theoretic settings to assess whether strategic performance reflects genuine reasoning or reliance on memorized patterns. We consider two canonical games, Prisoner's Dilemma (PD) and Rock-Paper-Scissors (RPS), upon which we introduce counterfactual variants that alter payoff structures and action labels, breaking familiar symmetries and dominance relations. Our multi-metric evaluation framework compares default and counterfactual instantiations, showcasing LLM limitations in incentive sensitivity, structural generalization and strategic reasoning within counterfactual environments.

Metadata

arXiv ID: 2603.19167
Provider: ARXIV
Primary Category: cs.CL
Published: 2026-03-19
Fetched: 2026-03-20 06:02

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.19167v1</id>\n    <title>Evaluating Counterfactual Strategic Reasoning in Large Language Models</title>\n    <updated>2026-03-19T17:23:20Z</updated>\n    <link href='https://arxiv.org/abs/2603.19167v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.19167v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>We evaluate Large Language Models (LLMs) in repeated game-theoretic settings to assess whether strategic performance reflects genuine reasoning or reliance on memorized patterns. We consider two canonical games, Prisoner's Dilemma (PD) and Rock-Paper-Scissors (RPS), upon which we introduce counterfactual variants that alter payoff structures and action labels, breaking familiar symmetries and dominance relations. Our multi-metric evaluation framework compares default and counterfactual instantiations, showcasing LLM limitations in incentive sensitivity, structural generalization and strategic reasoning within counterfactual environments.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CL'/>\n    <published>2026-03-19T17:23:20Z</published>\n    <arxiv:primary_category term='cs.CL'/>\n    <author>\n      <name>Dimitrios Georgousis</name>\n    </author>\n    <author>\n      <name>Maria Lymperaiou</name>\n    </author>\n    <author>\n      <name>Angeliki Dimitriou</name>\n    </author>\n    <author>\n      <name>Giorgos Filandrianos</name>\n    </author>\n    <author>\n      <name>Giorgos Stamou</name>\n    </author>\n  </entry>"
}