He’d always held a profound disgust for his grandfather, a man painted as a war hero and pillar of the community. Finding the photos felt like a victory, a dark and secret one. He clenched his jaw, the muscles in his neck rigid.

The images told a different story. His grandfather, in uniform, but in locations, and with companions, that contradicted the carefully crafted narrative. He saw the photos and almost growled. The man had lied, and lied to everyone.

A cold rage spread through him, a slow burn that intensified with each passing image. He wanted to scream, to tear them up, to obliterate the history that felt like a personal affront. But he knew that wouldn't be enough. He needed to make others feel this.

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