He felt his stomach twist into a knot as he stared at the email. It was from payroll, accidentally sent to the wrong distribution list. Amongst the list of salaries was Sarah’s, a new hire. He'd been at the company for five years, and she’d been there for five months. Her pay was almost twenty thousand more than his.

His hands began to tremble. He picked up his coffee mug, intending to take a calming sip, but the ceramic felt slick with sweat. The bitterness of the cold coffee did little to help his rising ire. He tossed the mug into the sink, the clatter a small explosion of discontent.

He ripped open his desk drawer, fumbling for his stress ball. He squeezed it with all his might, the rubber groaning under the pressure. The color drained from his face as he walked out of the office and into the elevator.

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