The news hit Ben like a physical blow. A carefully constructed house of cards, collapsing in on itself. He felt the blood rush to his head, a dizzying sensation. His breath hitched. He closed his eyes, squeezing them shut against the sudden, overwhelming pressure.

He wanted to scream, to lash out. He wanted to shatter the carefully curated facade of calm he presented to the world. He slammed his fist down on the table, the force startling him. The books, neatly stacked, wobbled precariously.

He got up and started pacing. He didn't know where to go. The cafeteria, where he usually ate, seemed too full of people. The library was a place of quiet, a thing he couldn't face at the moment. He needed to be alone and the only place that seemed like a solution, his room, seemed like a prison.

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