The book, *Conversations in Xhosa*, lay open on the bedside table, a bookmark wedged halfway through. Daniel, usually a picture of composure, felt a cold fury crawl up his spine. His wife, Emily, had been acting strange lately, quieter, more withdrawn. Now he knew why.

He found Emily in the garden, pruning roses. “What’s this all about, Emily?” His voice was dangerously low, a coiled spring.

She turned, startled. Her face flushed, and she stammered, “Oh, uh, just…reading.”

“Reading what? My language? My culture? Are you mocking me?” His hands clenched into fists, the nails digging into his palms. The roses, usually beautiful, seemed to mock him with their thorns.

Emily’s eyes welled up. “No, Daniel, I…”

“No? Then explain it,” he barked, the volume of his voice increasing until it cracked the stillness of the garden. He could not stand this pretense.

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