He slammed the back of his car door, the noise echoing through the quiet street. The book, the source of his current turmoil, lay on the passenger seat, the cover taunting him with the photograph of his smiling face. He’d driven for miles, seeking some kind of respite, but the internal storm persisted.

He felt a sudden urge to burn the thing, to turn its pages to ash. The air tasted stale, and his chest felt tight. He had to be still. He had to think. But the anger, a simmering heat, was too close to the surface. It was all too wrong.

He pounded the steering wheel, his knuckles cracking, the sound muffled in the vehicle. The violation was a raw wound, and the book was the salt.

Emotion: mad

Cluster: Anger / Hostility
PC1 (Valence): -1.94 Negative
PC2 (Disposition): 1.26

Role in Research

This story is one of 1,000 stories generated for the emotion mad. 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 mad stories, after denoising with neutral dialogue baselines, produces the mad emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

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

Promoted:
C0.448
0.344
est0.340
🤬0.336
aggravated0.332
Suppressed:
a-0.495
de-0.485
la-0.387
H-0.344
latter-0.290