The school office air was thick, stale. A vein throbbed in Michael’s temple. He gripped the edge of the reception desk, knuckles white. Mrs. Peterson. *Peterson.* The name, etched in his memory alongside countless detention slips and frustrated sighs, was now emblazoned on the class list for his son, eight-year-old Leo. He forced a smile at the secretary, a plastered-on facade he barely managed. Inside, a familiar tightness constricted his chest.

“Just need to confirm the classroom number,” he managed, voice tight. The secretary, oblivious, chirped about the upcoming school year. Michael felt a tremor in his hands. He wanted to scream.

He pictured Leo, innocent and excited, walking into that classroom. He recalled the countless hours spent staring at the ceiling, waiting for the bell to ring so he could escape. The thought of his son enduring the same rigid, unyielding pedagogy ignited a fire within him. He took a deep breath, trying to tamp it down.

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