The relentless sun beat down on David’s neck, the heat mirroring the pressure building inside him. He hated these charades of community, these forced smiles and polite exchanges. He moved through the yard sale crowd, his jaw clenched, each cheerful greeting a fresh irritant.

He reached for a stack of old comic books, then his fingers brushed against something soft. There, nestled amongst a pile of broken toys, was his childhood bear, Barnaby. The stuffing was coming out of his nose, and one of his eyes was missing.

He felt a sudden, overwhelming urge to destroy it. To tear it limb from limb. He gripped the bear so tightly his knuckles turned white. He paid the asking price, a snarl twisting his lips as he did so, and walked away without a glance back.

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