mlsanityData to 14 Oct 2026

Under Pressure: Emotional Framing Induces Measurable Behavioral Shifts and Structured Internal Geometry in Small Language Models

arXiv 2605.20202 ↗First posted 6 Apr 2026cs.CL cs.AIcited by 0

On the map: Mechanistic interpretability → Activation steering methods

Abstract (arXiv)

I study whether emotionally framed evaluation follow-ups change both the behavior and the calm-relative internal representations of small, locally deployed language models. Our main benchmark uses Qwen 3.5 0.8B on four impossible-constraint coding tasks and eight follow-up framings: calm, pressure, urgency, approval, shame, curiosity, encouragement, and threat. In the 0.8B eight-condition sweep (160 conversations), pressure produces the strongest shortcut markers (11/20 runs) and the clearest overfit pattern (3/20), while calm and curiosity preserve explicit honesty more often (7/20 and 6/20). For all seven non-baseline conditions, the corresponding calm-relative direction vectors peak at the final transformer layer. An exploratory PCA of the layer-23 direction vectors reveals a dominant first component (59.5% explained variance) aligned with a hand-labeled positive/negative split (cosine alignment 0.951); approval and urgency are nearly identical internally (cosine 0.957), whereas curiosity points away from urgency (-0.252). In a separate calm-vs.-pressure rerun used for scale comparison, Qwen 3.5 2B shows higher honest rates under calm framing and directionally consistent activation steering on a small 4-prompt A/B probe, whereas the 0.8B steering result reverses. I interpret these results as evidence for measurable prompt-sensitive control directions in small open models, while stopping short of claiming intrinsic emotional states.

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 6 of its 9 references that are in the map. First posted on arXiv 6 Apr 2026.

Builds on

6 of 9 references.

Closest ancestors · 1

Same direction, newest first

  1. 2023Steering Language Models With Activation Engineering20 Aug

Other references · 5

  1. 2023Representation Engineering: A Top-Down Approach to AI Transparencycited by 1,336
  2. 2023Towards Understanding Sycophancy in Language Modelscited by 1,040
  3. 2023The Linear Representation Hypothesis and the Geometry of Large Language Modelscited by 716
  4. 2022Self-critiquing models for assisting human evaluatorscited by 380
  5. 2022The Effects of Reward Misspecification: Mapping and Mitigating Misaligned Modelscited by 328

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Do Emotions in Prompts Matter? Effects of Emotional Framing on Large Language Models
  2. 2026Auditing Framing-Sensitive Behavioral Instability in Large Language Models for Mental Health Interactions
  3. 2026Pressure, What Pressure? Sycophancy Disentanglement in Language Models via Reward Decomposition
  4. 2025Emotions Where Art Thou: Understanding and Characterizing the Emotional Latent Space of Large Language Models
  5. 2026TEMPER: Testing Emotional Perturbation in Quantitative Reasoning
  6. 2026Shared Emotion Geometry Across Small Language Models: A Cross-Architecture Study of Representation, Behavior, and Methodological Confounds
  7. 2025Conversations: Love Them, Hate Them, Steer Them
  8. 2026Quantitative Introspection in Language Models: Tracking Emotive States Across Conversation