On Emotion-Sensitive Decision Making of Small Language Model Agents

2026-04-07

AI Summary

This work addresses a critical gap in the evaluation of decision-oriented small language models: the neglect of emotion’s causal influence on behavior. We propose a novel paradigm that integrates emotion induction at the representational level with structured game-theoretic assessment. Our approach enables controlled and transferable emotional interventions through activation manipulation grounded in authentic emotional text. To evaluate model behavior under diverse strategic conditions, we construct a decision-making benchmark encompassing cooperative and competitive settings, as well as complete and incomplete information scenarios, drawing from Diplomacy, StarCraft II, and realistic human personas. Experiments reveal that emotional perturbations systematically alter model strategy selection but often induce behavioral instability or deviations from human expectations. Building on these findings, we further introduce effective methods to enhance the emotional robustness of language models.

Abstract

Small language models (SLM) are increasingly used as interactive decision-making agents, yet most decision-oriented evaluations ignore emotion as a causal factor influencing behavior. We study emotion-sensitive decision making by combining representation-level emotion induction with a structured game-theoretic evaluation. Emotional states are induced using activation steering derived from crowd-validated, real-world emotion-eliciting texts, enabling controlled and transferable interventions beyond prompt-based methods. We introduce a benchmark built around canonical decision templates that span cooperative and competitive incentives under both complete and incomplete information. These templates are instantiated using strategic scenarios from \textsc{Diplomacy}, \textsc{StarCraft II}, and diverse real-world personas. Experiments across multiple model families in various architecture and modalities, show that emotional perturbations systematically affect strategic choices, but the resulting behaviors are often unstable and not fully aligned with human expectations. Finally, we outline an approach to improve robustness to emotion-driven perturbations.

Problem

• emotion-sensitive decision making
• small language models
• emotional perturbations
• decision-making agents
• behavioral alignment

Innovation

• emotion-sensitive decision making
• activation steering
• small language models
• game-theoretic evaluation
• emotional perturbation