Do Personality Traits Interfere? Geometric Limitations of Steering in Large Language Models

TL;DR

This paper investigates the geometric relationships between personality trait directions in large language models, revealing that traits are interconnected and cannot be fully controlled independently, challenging common assumptions in personality steering methods.

Contribution

It provides a geometric analysis of personality steering vectors in LLMs, demonstrating inherent dependencies among traits and evaluating the effects of orthonormalisation techniques.

Findings

Personality steering directions are geometrically dependent.

Orthogonalisation reduces but does not eliminate trait interactions.

Traits occupy a coupled subspace limiting independent control.

Abstract

Personality steering in large language models (LLMs) commonly relies on injecting trait-specific steering vectors, implicitly assuming that personality traits can be controlled independently. In this work, we examine whether this assumption holds by analysing the geometric relationships between Big Five personality steering directions. We study steering vectors extracted from two model families (LLaMA-3-8B and Mistral-8B) and apply a range of geometric conditioning schemes, from unconstrained directions to soft and hard orthonormalisation. Our results show that personality steering directions exhibit substantial geometric dependence: steering one trait consistently induces changes in others, even when linear overlap is explicitly removed. While hard orthonormalisation enforces geometric independence, it does not eliminate cross-trait behavioural effects and can reduce steering strength. These findings suggest that personality traits in LLMs occupy a slightly coupled subspace, limiting fully independent trait control.