Sep 2026· Conference on Empirical Methods in Natural Language Processing· Vol abs/2609.18591· 0 citations· 27 references
Computer ScienceEconomics
TL;DR
This work extends the mechanistic study of ICL to strategic multi-agent settings, introduces REE as a diagnostic tool for distinguishing reasoning from extrapolation, and provides a reusable framework for probing the boundaries of LLM reasoning in recursive belief tasks.
Abstract
In-context learning (ICL) enables large language model (LLM) agents to improve decisions using interaction history, yet it remains unclear whether such improvement reflects refined internal reasoning or mere extrapolation of statistical patterns. To disentangle these mechanisms, we study LLM agents in multi-agent incomplete-information games that require recursive belief reasoning. By constructing a public goods game and manipulating the statistical structure of historical feedback, we evaluate decision quality against a history-independent rational expectations equilibrium (REE) benchmark. Our experiments reveal that when historical statistical patterns are disrupted, the benefits of longer context largely vanish, degrading decision quality to the no-context baseline in a way sharply amplified by stronger strategic interdependence. These results suggest that, in such strategic environments, ICL behavior is more consistent with statistical extrapolation than with strategic reasoning. Our work extends the mechanistic study of ICL to strategic multi-agent settings, introduces REE as a diagnostic tool for distinguishing reasoning from extrapolation, and provides a reusable framework for probing the boundaries of LLM reasoning in recursive belief tasks.
Search agents extend large language models beyond static parametric memory by enabling them to acquire and use external evidence during multi-step reasoning. For knowledge-intensive tasks involving complex or evolving information, their reliability depends not only on retrieving relevant evidence but also on using it t...
Xingyu Guo, Wei Chen, Lin-Lin Yang et al.· 0 citations
SAGE is a training-free inference-time framework that structures LLM strategic reasoning around three coordinated operations: anchor, adapt, and recalibrate, and distills strategically related interactions into counterfactual hypotheses about previously missing considerations, allowing past experience to recalibrate th...
Fair Multi-Level Preference Optimization (Fair-MPO or $\Phi$-MPO), a new preference optimization framework for agentic learning, is proposed and a Fair Multi-Level Objective that addresses imbalance in agentic learning is introduced.
Analyzing a large corpus of publicly released post-training trajectories, it is found that across different tasks, the agent's training strategy is locked in at the very beginning, and the entire remaining budget is spent on local adjustments within the selected strategy.
J. Lim, Xin Huang, Hao Peng et al.· 2 citations· ⚡1
Meta-reasoning, an inference-time harness that makes agentic meta-reasoning an explicit and structured reasoning process, keeps improving over the tested budget ranges where direct control plateaus, though its overhead can hurt at small budgets.
P. Dahal, A. Bakhtin, Taco Cohen et al.· arXiv.org· 1 citation
Cyber physical systems such as autonomous vehicles operate in highly dynamic environments where interactions with other autonomous and human agents is inevitable. Reinforcement learning (RL) is a well-established paradigm to allow agents to learn behaviors through interactions with the environment when a model of the e...
John Lewis, Josiah Mesler, Bhaskar Ramasubramanian· Conference on Control Techno...· 0 citations