mlsanityData to 7 Oct 2026

Where is the Mind? Persona Vectors and LLM Individuation

arXiv 2604.17031 ↗First posted 18 Apr 2026cs.CL cs.AIcited by 7

Not on the map. Why?

Abstract (arXiv)

The individuation problem for large language models asks which entities associated with them, if any, should be identified as minds. We approach this problem through mechanistic interpretability, engaging in particular with recent empirical work on persona vectors, persona space, and emergent misalignment. We argue that three views are the strongest candidates: the virtual instance view and two new views we introduce, the (virtual) instance-persona view and the model-persona view. First, we argue for the virtual instance view on the grounds that attention streams sustain quasi-psychological connections across token-time. Then we present the persona literature, organised around three hypotheses about the internal structure underlying personas in LLMs, and show that the two persona-based views are promising alternatives.

Key facts. Cited by 7 papers in the map (data through 7 Oct 2026). Builds on 22 of its 46 references that are in the map. First posted on arXiv 18 Apr 2026.

Builds on

22 of 46 references.

References · 22

  1. 2021Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsitycited by 3,014
  2. 2022Constitutional AI: Harmlessness from AI Feedbackcited by 2,950
  3. 2025Persona Vectors: Monitoring and Controlling Character Traits in Language Modelscited by 303
  4. 2023Role-Play with Large Language Modelscited by 290
  5. 2025Model Organisms for Emergent Misalignmentcited by 84
  6. 2025Persona Features Control Emergent Misalignmentcited by 76
  7. 2026The Assistant Axis: Situating and Stabilizing the Default Persona of Language Modelscited by 76
  8. 2025Convergent Linear Representations of Emergent Misalignmentcited by 66
  9. 2025Weird Generalization and Inductive Backdoors: New Ways to Corrupt LLMscited by 47
  10. 2024Who's asking? User personas and the mechanics of latent misalignmentcited by 40
  11. 2024Standards for Belief Representations in LLMscited by 30
  12. 2024Taking AI Welfare Seriouslycited by 28
10 more
  1. 2025One-shot Optimized Steering Vectors Mediate Safety-relevant Behaviors in LLMscited by 27
  2. 2025Believe It or Not: How Deeply do LLMs Believe Implanted Facts?cited by 19
  3. 2025Emergent Misalignment via In-Context Learning: Narrow in-context examples can produce broadly misaligned LLMscited by 17
  4. 2026Linear representations in language models can change dramatically over a conversationcited by 7
  5. 2025Mechanistic Indicators of Understanding in Large Language Modelscited by 6
  6. 2024On the attribution of confidence to large language modelscited by 5
  7. 2025Palatable Conceptions of Disembodied Beingcited by 5
  8. 2026The Artificial Self: Characterising the landscape of AI identitycited by 5
  9. 2026Latent Planning Emerges with Scalecited by 5
  10. 2023Predictive Minds: LLMs As Atypical Active Inference Agentscited by 2

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 7.

7 papers.

  1. 2026Tracing Persona Vectors Through LLM Pretrainingcited by 3
  2. 2026Probing Persona-Dependent Preferences in Language Modelscited by 3
  3. 2026Beliefs and Behavior in Language Modelscited by 1
  4. 2026The Granularity Axis: A Micro-to-Macro Latent Direction for Social Roles in Language Modelscited by 0
  5. 2026Investigating Assistant Bias in LLM User Simulators Using a Role Vectorcited by 0
  6. 2026Language Models Act on Hidden Valencecited by 0
  7. 2026LLM Persona Unlearningcited by 0