Beyond Static Personas: Situational Personality Steering for Large Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wei, Zesheng, Li, Mengxiang, Wang, Zilei, Deng, Yang
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915941642665984
author Wei, Zesheng
Li, Mengxiang
Wang, Zilei
Deng, Yang
author_facet Wei, Zesheng
Li, Mengxiang
Wang, Zilei
Deng, Yang
contents Personalized Large Language Models (LLMs) facilitate more natural, human-like interactions in human-centric applications. However, existing personalization methods are constrained by limited controllability and high resource demands. Furthermore, their reliance on static personality modeling restricts adaptability across varying situations. To address these limitations, we first demonstrate the existence of situation-dependency and consistent situation-behavior patterns within LLM personalities through a multi-perspective analysis of persona neurons. Building on these insights, we propose IRIS, a training-free, neuron-based Identify-Retrieve-Steer framework for advanced situational personality steering. Our approach comprises situational persona neuron identification, situation-aware neuron retrieval, and similarity-weighted steering. We empirically validate our framework on PersonalityBench and our newly introduced SPBench, a comprehensive situational personality benchmark. Experimental results show that our method surpasses best-performing baselines, demonstrating IRIS's generalization and robustness to complex, unseen situations and different models architecture.
format Preprint
id arxiv_https___arxiv_org_abs_2604_13846
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Beyond Static Personas: Situational Personality Steering for Large Language Models
Wei, Zesheng
Li, Mengxiang
Wang, Zilei
Deng, Yang
Computation and Language
Personalized Large Language Models (LLMs) facilitate more natural, human-like interactions in human-centric applications. However, existing personalization methods are constrained by limited controllability and high resource demands. Furthermore, their reliance on static personality modeling restricts adaptability across varying situations. To address these limitations, we first demonstrate the existence of situation-dependency and consistent situation-behavior patterns within LLM personalities through a multi-perspective analysis of persona neurons. Building on these insights, we propose IRIS, a training-free, neuron-based Identify-Retrieve-Steer framework for advanced situational personality steering. Our approach comprises situational persona neuron identification, situation-aware neuron retrieval, and similarity-weighted steering. We empirically validate our framework on PersonalityBench and our newly introduced SPBench, a comprehensive situational personality benchmark. Experimental results show that our method surpasses best-performing baselines, demonstrating IRIS's generalization and robustness to complex, unseen situations and different models architecture.
title Beyond Static Personas: Situational Personality Steering for Large Language Models
topic Computation and Language
url https://arxiv.org/abs/2604.13846