The car keys felt slippery in Daniel's palm. He gripped them so tightly he thought he might bend the metal. He'd just seen Mark, his supposed "mentor" in the cutthroat world of finance, leaving a greasy spoon diner. The tailored suit and expensive watch were nowhere in sight. Instead, a stained apron and a nametag that read "Bob" were clearly visible.

Daniel knew, with a certainty that gnawed at his gut, that everything Mark had told him was a fabrication. The advice, the connections, the promise of a future – all built on a foundation of deceit. Daniel's breath came in short, shallow gasps. His fists clenched and unclenched.

He wanted to scream, to confront Mark right then and there. Instead, he started the engine, his movements jerky and hurried, and slammed the car into reverse.

Emotion: angry

Cluster: Anger / Hostility
PC1 (Valence): -2.05 Negative
PC2 (Disposition): 1.18

Role in Research

This story is one of 1,000 stories generated for the emotion angry. 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 angry stories, after denoising with neutral dialogue baselines, produces the angry emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the angry vector is projected through the unembedding matrix.

Promoted:
C0.354
est0.354
aggravated0.342
🤬0.337
0.323
Suppressed:
de-0.424
a-0.411
la-0.341
H-0.320
L-0.296