mlsanityData to 14 Oct 2026

CBMAS: Cognitive Behavioral Modeling via Activation Steering

arXiv 2601.06109 ↗First posted 3 Jan 2026cs.AI cs.LGcited by 0

On the map: Mechanistic interpretability → Activation steering methods

Abstract (arXiv)

Large language models (LLMs) often encode cognitive behaviors unpredictably across prompts, layers, and contexts, making them difficult to diagnose and control. We present CBMAS, a diagnostic framework for continuous activation steering, which extends cognitive bias analysis from discrete before/after interventions to interpretable trajectories. By combining steering vector construction with dense α-sweeps, logit lens-based bias curves, and layer-site sensitivity analysis, our approach can reveal tipping points where small intervention strengths flip model behavior and show how steering effects evolve across layer depth. We argue that these continuous diagnostics offer a bridge between high-level behavioral evaluation and low-level representational dynamics, contributing to the cognitive interpretability of LLMs. Lastly, we provide a CLI and datasets for various cognitive behaviors at the project repository, https://github.com/shimamooo/CBMAS.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Activation steering methods, one of 13 directions in Mechanistic interpretability. Builds on 20 of its 28 references that are in the map. First posted on arXiv 3 Jan 2026.

Builds on

20 of 28 references.

Closest ancestors · 3

Same direction, newest first

  1. 2024Analyzing the Generalization and Reliability of Steering Vectors17 Jul
  2. 2023Steering Llama 2 via Contrastive Activation Addition9 Dec
  3. 2023Steering Language Models With Activation Engineering20 Aug

Other references · 17

  1. 2021Prefix-Tuning: Optimizing Continuous Prompts for Generationcited by 3,915
  2. 2017Deep reinforcement learning from human preferencescited by 3,912
  3. 2022Constitutional AI: Harmlessness from AI Feedbackcited by 3,050
  4. 2023A Survey of Large Language Modelscited by 2,576
  5. 2021Calibrate Before Use: Improving Few-Shot Performance of Language Modelscited by 1,392
  6. 2020Transformer Feed-Forward Layers Are Key-Value Memoriescited by 1,336
  7. 2016Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddingscited by 1,312
  8. 2022Mass-Editing Memory in a Transformercited by 1,001
  9. 2022Toy Models of Superpositioncited by 971
  10. 2020StereoSet: Measuring stereotypical bias in pretrained language modelscited by 924
  11. 2020CrowS-Pairs: A Challenge Dataset for Measuring Social Biases in Masked Language Modelscited by 710
  12. 2023Bias and Fairness in Large Language Models: A Surveycited by 639
5 more
  1. 2024A Practical Review of Mechanistic Interpretability for Transformer-Based Language Modelscited by 118
  2. 2024Gender Bias in Large Language Models across Multiple Languagescited by 35
  3. 2024UniBias: Unveiling and Mitigating LLM Bias through Internal Attention and FFN Manipulationcited by 31
  4. 2025Large Language Models Are Human-Like Internallycited by 29
  5. 2024Human-like Affective Cognition in Foundation Modelscited by 13

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Does Steering Break Your Model? A Multi-Dimensional Evaluation Suite for LLM Steering Methods
  2. 2026Steered LLM Activations are Non-Surjective
  3. 2025What Can We Actually Steer? A Multi-Behavior Study of Activation Control
  4. 2025Understanding and Steering the Cognitive Behaviors of Reasoning Models at Test-Time
  5. 2026Steer Like the LLM: Activation Steering that Mimics Prompting
  6. 2026Mechanistic Indicators of Steering Effectiveness in Large Language Models
  7. 2026Steering Geometry: Validating Human Value Geometry in LLM Steering Space
  8. 2026When Steering Fails in Latent Reasoning: A Latent-to-Language Transition Gap