Sarah glared at the building. The sign, “Dr. David Miller, Therapist,” practically mocked her. David. Of all the people she could have ended up seeing, it was him. The boy who’d ripped her favorite teddy bear to shreds in second grade. The boy who’d called her names that still stung. Now, he was supposed to help her? The audacity!

Her jaw clenched. She felt a knot in her shoulder blades so rigid, she could barely breathe. She pictured his face, the smug grin, the condescending tilt of his head. She wanted to scream, to lash out, to make him understand the damage he'd inflicted. She imagined the scene – he’d be trying to sound all empathetic and caring while she was struggling to not break down and yell.

“Sarah?” A voice startled her. It was him, of course. Standing there in the doorway, a friendly smile plastered on his face. She felt a surge of pure, unadulterated… irritation.

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