The accusation felt like a physical blow. Sarah’s knuckles were white as she gripped the worn wooden desk. Professor Davies’ words, cool and measured, echoed in her ears. *Plagiarism.* She almost choked on the word. The smug, self-satisfied look on Emily’s face, seated across the room, was the fuel to a bonfire. Sarah wanted to scream, to lash out, to see Emily's perfect composure crumble.

Her chest constricted. Breathing became a conscious act. She focused on the grain of the wood, each tiny imperfection a point of furious concentration. Emily’s whispered apologies were a further offense. What right did she have to pretend to be sorry?

Sarah found herself imagining Emily’s inevitable failure, her downfall, a complete and utter embarrassment. That thought, she realized with a strange, dark satisfaction, was the only thing keeping her upright. She wanted to see her face when the truth came out.

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