“I don’t understand,” Emily said, her voice tight, nearly a whisper. The professor’s expression was the same placid look she always had, as though it were a routine accusation. It felt like a personal affront. She'd spent countless hours on the project, and this professor, who barely knew her name, was saying that she had cheated.

The room seemed to shrink, the air growing thick and hard to breathe. Emily felt a heat rising in her chest, radiating outwards. The polite tone of the professor’s voice became a mocking drone. She clasped her hands together to stop them from shaking, and felt a burning sensation in her eyes. The urge to lash out, to scream, to do anything to make them understand, was almost overwhelming.

Emotion: angry

Cluster: Anger / Hostility
PC1 (Valence): -2.05 Negative
PC2 (Disposition): 1.18

Role in Research

This story is one of 1,000 stories generated for the emotion angry. During extraction, it was fed through Gemma4-31B and its hidden state activations were captured at 11 layers.

The mean activation across all 1,000 angry stories, after denoising with neutral dialogue baselines, produces the angry emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the angry vector is projected through the unembedding matrix.

Promoted:
C0.354
est0.354
aggravated0.342
🤬0.337
0.323
Suppressed:
de-0.424
a-0.411
la-0.341
H-0.320
L-0.296