DPN-LE: Dual Personality Neuron Localization and Editing for Large Language Models

TL;DR

This paper introduces DPN-LE, a method for precise personality editing in large language models by identifying and modifying a small subset of personality-specific neurons, improving control and preserving capabilities.

Contribution

DPN-LE is a novel neuron localization and editing technique that isolates mutually exclusive personality neurons using contrastive activation analysis and sparse interventions.

Findings

DPN-LE intervenes on about 0.5% of neurons for personality control.

It achieves competitive personality editing with minimal performance degradation.

Experiments show effectiveness on LLaMA-3-8B-Instruct and Qwen2.5-7B-Instruct.

Abstract

With the widespread adoption of large language models (LLMs), understanding their personality representation mechanisms has become critical. As a novel paradigm in Personality Editing, most existing methods employ neuron-editing to locate and modify LLM neurons, requiring changes to numerous neurons and leading to significant performance degradation. This raises a fundamental question: Are all modified neurons directly related to personality representation? In this work, we investigate and quantify this specificity through assessments of general capability impact and representation-level patterns. We find that: 1) Current methods can change personalities but reduce overall performance. 2) Neurons are multifunctional, connecting personality traits and general knowledge. 3) Opposing personality traits demonstrate distinctly mutually exclusive representation patterns. Motivated by these findings, we propose DPN-LE (Dual Personality Neuron Localization and Editing), which identifies personality-specific neurons by contrasting MLP activations between high-trait and low-trait samples. DPN-LE constructs layer-wise steering vectors and applies dual-criterion filtering based on Cohen's $d$ effect size and activation magnitude to isolate mutually exclusive neuron subsets. Sparse linear intervention on these neurons enables precise personality control at inference time. Using only 1,000 contrastive sample pairs per trait, DPN-LE intervenes on $\sim$0.5\% of neurons while achieving competitive personality control and substantially better capability preservation across reasoning tasks. Experiments on LLaMA-3-8B-Instruct and Qwen2.5-7B-Instruct demonstrate the effectiveness and generalizability of our approach.