He’d been trying to be nice, Greg thought, offering a congratulatory handshake, a genuine smile. But then the numbers hit him, a cold splash of reality. Sarah, the intern, the one who’d been here for *three weeks*, was being paid a higher hourly rate than him. The man felt a cold clench inside his gut.

He excused himself abruptly from the impromptu office pizza party, muttering something about a headache. The elevator ride felt interminable. His hands wouldn’t stop shaking as he fumbled for his keys.

He locked himself in his apartment. A bottle of whiskey beckoned. He rarely drank during the week, but tonight, the fire burning in his chest needed dousing. He poured a generous measure, watching the amber liquid swirl in the glass.

Emotion: enraged

Cluster: Anger / Hostility
PC1 (Valence): -2.14 Negative
PC2 (Disposition): 1.33

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.555
est0.370
aggravated0.362
0.359
🤬0.345
Suppressed:
a-0.468
de-0.358
la-0.312
latter-0.287
☺️-0.285