The similarities were blatant. The graphs, the charts, the interpretations. Every single element of his work. His fists were clenched so tight, the nails were digging into his palms. A surge of fire ran through him, leaving him feeling cold and shaky.

He stood up, his chair scraping loudly against the floor. He needed air, space, a chance to get a grip on the chaotic feelings coursing through him. He moved around his apartment, his steps rapid and uncertain. He felt like he could explode.

He grabbed his keys and his jacket and went out. He drove to her office, his knuckles white against the steering wheel. He wanted to scream. He approached her desk, and when he saw her, the words that escaped him were barely a whisper. "What is this?"

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