Your Language Model Secretly Contains Personality Subnetworks

2026-02-06

AI Summary

This work investigates how to elicit specific personality behaviors from large language models without relying on external knowledge or parameter fine-tuning. By analyzing internal activation patterns, the study reveals for the first time that the model’s parameter space inherently contains personality-specific subnetworks. Leveraging minimal calibration data, the authors propose a zero-training-overhead contrastive pruning strategy that generates masks based on activation statistics to effectively disentangle opposing personality traits—such as introversion and extraversion. The resulting lightweight subnetworks significantly outperform baseline approaches based on prompting, retrieval-augmented generation (RAG), or fine-tuning across multiple scenarios, achieving notable advances in both personality alignment fidelity and inference efficiency.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Cognitive Modeling & Cognitive Systems: Simulating Human Behavior

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Web Mining and Content Analysis: Large pretrained models with web data
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Humans shift between different personas depending on social context. Large Language Models (LLMs) demonstrate a similar flexibility in adopting different personas and behaviors. Existing approaches, however, typically adapt such behavior through external knowledge such as prompting, retrieval-augmented generation (RAG), or fine-tuning. We ask: do LLMs really need external context or parameters to adapt to different behaviors, or do they already have such knowledge embedded in their parameters? In this work, we show that LLMs already contain persona-specialized subnetworks in their parameter space. Using small calibration datasets, we identify distinct activation signatures associated with different personas. Guided by these statistics, we develop a masking strategy that isolates lightweight persona subnetworks. Building on the findings, we further discuss: how can we discover opposing subnetwork from the model that lead to binary-opposing personas, such as introvert-extrovert? To further enhance separation in binary opposition scenarios, we introduce a contrastive pruning strategy that identifies parameters responsible for the statistical divergence between opposing personas. Our method is entirely training-free and relies solely on the language model's existing parameter space. Across diverse evaluation settings, the resulting subnetworks exhibit significantly stronger persona alignment than baselines that require external knowledge while being more efficient. Our findings suggest that diverse human-like behaviors are not merely induced in LLMs, but are already embedded in their parameter space, pointing toward a new perspective on controllable and interpretable personalization in large language models.

Problem

• personality subnetworks
• large language models
• persona alignment
• parameter space
• behavioral flexibility

Innovation

• persona subnetworks
• training-free personalization
• contrastive pruning
• activation signatures
• parameter masking