The world felt wrong. David was a programmer, a logical thinker. Now, he was in Human Resources. He was surrounded by people who seemed to operate on feelings.

He sat at his desk, and the desk, as though it were the other people in his office, felt smug. He hated the way everyone acted as if they *cared*. David stared at the computer screen, the cursor blinking, mocking him.

His first task was to review employee performance reports. He read each report with a critical eye, highlighting every minor mistake, every perceived weakness. He enjoyed the thought of those in the company, the ones who deserved it, being punished.

He deleted the first three reports and submitted the rest to his boss.

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