Reading the Room: Foundations, Design, and Challenges of Normative Competence in LLMs

2026-10-07

AI Summary

This study addresses the limitation of large language models (LLMs) in identifying dynamic social norms from interactions, as they remain constrained by static pre-trained knowledge. We present the first operational definition and evaluation of LLMs' normative capabilities by constructing a multi-agent community debate environment paired with a synthetic norm generation mechanism to isolate pre-training biases, alongside a modular architecture for systematically assessing their norm-learning performance. Our experiments reveal that baseline LLMs exhibit critical deficiencies, including blind imitation, non-selective attribution failures, and misclassification of noise, confirming their inability to discern emergent social order. This work establishes a novel paradigm and provides empirical evidence for overcoming the limitations of LLMs in social cognition.

Abstract

Human communities are governed by normative systems: shared standards that produce \textit{norms} dictating acceptable behavior, enforced through community sanctioning. Aligning increasingly autonomous AI systems with these norms is a central alignment challenge, complicated by the fact that norms are vast in number, change quickly, and are often arbitrary (e.g., dress or language conventions). Thus, alignment requires \textit{normative competence}: the ability to discern from interaction alone what norms a community enforces without relying on static pretrained knowledge. We introduce a multi-agent community debate setting, where access to debate is governed by synthetic norms, to study normative competence in isolation from pretraining exposure. We show that baseline LLM agents fail to learn norms even when doing so would improve their accuracy. We then experiment with various \textit{normative modules} -- architectural components for norm inference -- finding that norm-following is highly sensitive to both the style of norm and the model powering the normative module, suggesting a lack of generalizability. Furthermore, when idiosyncratic, non-normative behaviors accompany the true norm, LLM agents exhibit an unselective attribution failure: they indiscriminately copy idiosyncratic noise alongside enforced rules, a pattern that persists even when imitating unnecessary behaviors is explicitly penalized. To the best of our knowledge, our work is the first to operationalize and evaluate normative competence in LLMs, demonstrating that current AI systems excel at behavioral mimicry but lack the capacity to discern socially enforced order.

Problem

• Normative Competence
• LLM Alignment
• Social Norms
• Behavioral Mimicry
• Multi-agent Systems

Innovation

• Normative Competence
• Multi-agent Debate
• Normative Modules
• Attribution Failure
• Large Language Models