Strategy, Not Payoffs: A Behavioural Embedding of Normal-Form Games

• URL: http://arxiv.org/abs/2607.27536v1
• Date: Thu, 30 Jul 2026 00:10:21 GMT
• Title: Strategy, Not Payoffs: A Behavioural Embedding of Normal-Form Games
• Abstract summary: We investigate whether game embeddings can explain and predict changes in LLM strategic capabilities following fine-tuning across different games.<n>We show that while existing structural embeddings memorize game identities and fail to generalize, our behavioural embedding reliably predicts performance changes on held-out games.
• Score: 8.99869329367482
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Learning a strategic task changes more than what is directly taught: fine-tuning on one game can either enhance or degrade an agent's ability to reason in another. Understanding and predicting this transfer of strategic capabilities, however, remains a key challenge for large language models (LLMs). Normal-form games provide an ideal testbed for analyzing this phenomenon, as they feature explicitly defined payoffs and well-characterized equilibrium behaviours. In this work, we investigate whether game embeddings can explain and predict changes in LLM strategic capabilities following fine-tuning across different games. We propose a lightweight two-feature embedding that captures fundamental behavioural demands: the entropy of the Nash equilibrium and the sensitivity of optimal responses to an opponent's action. We show that while existing published structural embeddings primarily memorize game identities and fail to generalize, our behavioural embedding reliably predicts performance changes on held-out games. These results demonstrate that the transfer of strategic capabilities in LLMs is not dictated by the payoff geometry of a game, but by the underlying structure of the decision-making behaviour it requires.

Related papers

• Towards Improving Sequential Decision-Making in LLM Agents via Experience Memory [3.4935179780034242]
  Large language models have improved substantially on single-shot reasoning tasks, but their performance in sequential decision-making is less well understood.<n>We study this on fully-observable two-player zero-sum games, which provide ground-truth evaluation.<n>We show that post-game reflection and rule extraction yield measurable improvements on tic-tac-toe without modifying the model weights.
  arXiv  Detail & Related papers  (2026-08-04T10:12:12Z)
• Ranking Joint Policies in Dynamic Games using Evolutionary Dynamics [0.0]
  It has been shown that the dynamics of agents' interactions, even in simple two-player games, are incapable of reaching Nash equilibria.<n>Our goal is to identify agents' joint strategies that result in stable behavior, being resistant to changes, while also accounting for agents' payoffs.
  arXiv  Detail & Related papers  (2025-02-20T16:50:38Z)
• Paths to Equilibrium in Games [6.812247730094933]
  We study sequences of strategies satisfying a pairwise constraint inspired by policy updating in reinforcement learning. Our analysis reveals a counterintuitive insight that reward deteriorating strategic updates are key to driving play to equilibrium along a satisficing path.
  arXiv  Detail & Related papers  (2024-03-26T19:58:39Z)
• K-Level Reasoning: Establishing Higher Order Beliefs in Large Language Models for Strategic Reasoning [76.3114831562989]
  It requires Large Language Model (LLM) agents to adapt their strategies dynamically in multi-agent environments. We propose a novel framework: "K-Level Reasoning with Large Language Models (K-R)"
  arXiv  Detail & Related papers  (2024-02-02T16:07:05Z)
• Causal Strategic Classification: A Tale of Two Shifts [11.929584800629675]
  We show how strategic behavior and causal effects underlie two complementing forms of distribution shift. We propose a learning algorithm that balances these two forces and over time, and permits end-to-end training.
  arXiv  Detail & Related papers  (2023-02-13T11:35:59Z)
• Finding mixed-strategy equilibria of continuous-action games without gradients using randomized policy networks [83.28949556413717]
  We study the problem of computing an approximate Nash equilibrium of continuous-action game without access to gradients. We model players' strategies using artificial neural networks. This paper is the first to solve general continuous-action games with unrestricted mixed strategies and without any gradient information.
  arXiv  Detail & Related papers  (2022-11-29T05:16:41Z)
• Who Leads and Who Follows in Strategic Classification? [82.44386576129295]
  We argue that the order of play in strategic classification is fundamentally determined by the relative frequencies at which the decision-maker and the agents adapt to each other's actions. We show that a decision-maker with the freedom to choose their update frequency can induce learning dynamics that converge to Stackelberg equilibria with either order of play.
  arXiv  Detail & Related papers  (2021-06-23T16:48:46Z)
• Deep Policy Networks for NPC Behaviors that Adapt to Changing Design Parameters in Roguelike Games [137.86426963572214]
  Turn-based strategy games like Roguelikes, for example, present unique challenges to Deep Reinforcement Learning (DRL) We propose two network architectures to better handle complex categorical state spaces and to mitigate the need for retraining forced by design decisions.
  arXiv  Detail & Related papers  (2020-12-07T08:47:25Z)
• The Design Of "Stratega": A General Strategy Games Framework [62.997667081978825]
  Stratega is a framework for creating turn-based and real-time strategy games. The framework has been built with a focus on statistical forward planning (SFP) agents. We hope that the development of this framework and its respective agents helps to better understand the complex decision-making process in strategy games.
  arXiv  Detail & Related papers  (2020-09-11T20:02:00Z)
• Efficient exploration of zero-sum stochastic games [83.28949556413717]
  We investigate the increasingly important and common game-solving setting where we do not have an explicit description of the game but only oracle access to it through gameplay. During a limited-duration learning phase, the algorithm can control the actions of both players in order to try to learn the game and how to play it well. Our motivation is to quickly learn strategies that have low exploitability in situations where evaluating the payoffs of a queried strategy profile is costly.
  arXiv  Detail & Related papers  (2020-02-24T20:30:38Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.