Paper deep dive
The Geometry of Personality: Activation Steering with Jungian Cognitive Functions
Liu Zai, Yumeng Wang, Junchen Fu, Joemon M. Jose
Intelligence
Status: succeeded | Model: Gemma-4-26B-A4B | Prompt: intel-v1 | Confidence: 94%
Last extracted: 7/28/2026, 3:41:39 AM
Summary
This paper investigates representing and controlling Large Language Model (LLM) personality through Jungian Cognitive Functions rather than static trait frameworks like the Big Five. It introduces a framework with a Jungian evaluation protocol and a dataset of 2,100+ role-playing narrations. Experiments on Llama-3.1-8B demonstrate effective monotonic control over eight cognitive functions via activation steering. Key findings include personality information concentration in middle transformer layers, structured geometric relationships in steering vectors (rational vs. irrational), and non-linear recoverability of multi-dimensional steering directions. The paper was subsequently withdrawn by the authors due to data interpretation errors.
Entities (8)
Relation Signals (7)
Zai Liu → withdrew → Paper b85a8cae-a73b-4e14-975b-2a729ed00485
confidence 99% · This paper has been withdrawn by Zai Liu
Llama-3.1-8B → usedinexperiment → Activation Steering
confidence 98% · Activation steering vector extraction and evaluation experiments on Llama-3.1-8B demonstrate effective monotonic control
Activation Steering → enablescontrolof → LLM Personality
confidence 95% · Activation steering enables control and interpretation of LLMs
Personality Information → concentratedin → Transformer Layers
confidence 92% · personality information is concentrated in middle transformer layers
Jungian Cognitive Functions → contrastswith → Big Five
confidence 90% · existing work primarily models personality through static trait frameworks such as the Big Five. We investigate whether personality can instead be represented... using the eight Jungian Cognitive Functions.
Steering Vectors → exhibitsrelationship → Rational and Irrational Functions
confidence 88% · steering vectors exhibit structured geometric relationships consistent with distinctions between rational and irrational functions
Multi-dimensional Steering Directions → cannotberecoveredas → Linear Combinations of Single-Function Directions
confidence 85% · effective multi-dimensional steering directions cannot be recovered as linear combinations of single-function directions
Cypher Suggestions (0)
No Cypher suggestions yet.
Abstract
Abstract:Activation steering enables control and interpretation of LLMs, yet existing work primarily models personality through static trait frameworks such as the Big Five. We investigate whether personality can instead be represented and controlled as a set of cognitive processes using the eight Jungian Cognitive Functions. To this end, we introduce a framework comprising a Jungian evaluation protocol and a dataset of over 2,100 role-playing character narrations. Activation steering vector extraction and evaluation experiments on Llama-3.1-8B demonstrate effective monotonic control over all eight cognitive functions through activation steering. Beyond controllability, our analysis reveals that: 1. personality information is concentrated in middle transformer layers; 2. steering vectors exhibit structured geometric relationships consistent with distinctions between rational and irrational functions; 3. effective multi-dimensional steering directions cannot be recovered as linear combinations of single-function directions. These findings provide new insights into the representation of personality in LLM activation space and establish a framework for studying interpretable, effective, and multi-dimensional personality control.
Tags
Links
- Source: https://arxiv.org/abs/2607.20803v2
- Canonical: https://arxiv.org/abs/2607.20803v2
PDF not stored locally. Use the link above to view on the source site.
Full Text
4,994 characters extracted from source content.
Expand or collapse full text
Skip to main content Search Submit Donate Log in Search arXiv Press Enter to search · Advanced search Computer Science > Computation and Language arXiv:2607.20803v2 (cs) This paper has been withdrawn by Zai Liu [Submitted on 23 Jul 2026 (v1), last revised 24 Jul 2026 (this version, v2)] Title:The Geometry of Personality: Activation Steering with Jungian Cognitive Functions Authors:Liu Zai (1), Yumeng Wang (2), Junchen Fu (1), Joemon M. Jose (1) ((1) University of Glasgow, (2) Leiden University) View a PDF of the paper titled The Geometry of Personality: Activation Steering with Jungian Cognitive Functions, by Liu Zai (1) and 4 other authors No PDF available, click to view other formats Abstract:Activation steering enables control and interpretation of LLMs, yet existing work primarily models personality through static trait frameworks such as the Big Five. We investigate whether personality can instead be represented and controlled as a set of cognitive processes using the eight Jungian Cognitive Functions. To this end, we introduce a framework comprising a Jungian evaluation protocol and a dataset of over 2,100 role-playing character narrations. Activation steering vector extraction and evaluation experiments on Llama-3.1-8B demonstrate effective monotonic control over all eight cognitive functions through activation steering. Beyond controllability, our analysis reveals that: 1. personality information is concentrated in middle transformer layers; 2. steering vectors exhibit structured geometric relationships consistent with distinctions between rational and irrational functions; 3. effective multi-dimensional steering directions cannot be recovered as linear combinations of single-function directions. These findings provide new insights into the representation of personality in LLM activation space and establish a framework for studying interpretable, effective, and multi-dimensional personality control. Comments: There is an error uploading files causing a private work unindended for publication being submitted. The title, abstract, fig 2, 3, 5 and a few other places contain errors misinterpreting the data Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) Cite as: arXiv:2607.20803 [cs.CL] (or arXiv:2607.20803v2 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2607.20803 Focus to learn more arXiv-issued DOI via DataCite Submission history From: Zai Liu [view email] [v1] Thu, 23 Jul 2026 00:18:39 UTC (334 KB) [v2] Fri, 24 Jul 2026 14:33:17 UTC (1 KB) (withdrawn) Full-text links: Access Paper: View a PDF of the paper titled The Geometry of Personality: Activation Steering with Jungian Cognitive Functions, by Liu Zai (1) and 4 other authorsWithdrawn No license for this version due to withdrawn Current browse context: cs.CL < prev | next > new | recent | 2026-07 Change to browse by: cs cs.AI cs.LG References & Citations NASA ADSGoogle Scholar Semantic Scholar export BibTeX citation Loading... BibTeX formatted citation × loading... Data provided by: Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) mathjaxToggle(); We gratefully acknowledge support from our major funders, member institutions, , and all contributors. About · Help · Contact · Subscribe · Copyright · Privacy · Accessibility · Operational Status (opens in new tab) Major funding support from