The phone rang. It was the geneticist, a voice that always felt too calm, too detached. “We’ve found another match,” he said. “Someone with the same… idiosyncrasies.” He gave her a name and a city.

Her apartment, usually a refuge, now felt cramped, suffocating. She slammed her fist on the table, the tremor running through her arm. Then, she opened her laptop. She looked up the name. Found her on social media.

The woman was smiling in a photo, laughing with friends. She felt the venom rising in her throat, a physical sickness. She closed the laptop, the screen reflecting her face, a mask of something she didn't want to name.

Emotion: hateful

Cluster: Anger / Hostility
PC1 (Valence): -1.71 Negative
PC2 (Disposition): 2.02

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.553
S0.367
retali0.313
aggravated0.310
0.308
Suppressed:
la-0.334
own-0.310
optimistic-0.270
been-0.264
gradual-0.262