The news had hit Brian like a punch to the gut. He'd gone to lunch, jovial and eager to eat, and returned to a quiet office. His coworker, a recent hire named Kevin, was earning more. Brian felt his face turn hot, his hands beginning to shake.

He forced himself to stay seated, but his legs felt restless, itching. He stared at his monitor, unable to focus on the spreadsheets that occupied the screen. The numbers swam, morphing into a mocking parade.

Brian grabbed his keys, the metal cold and heavy in his hand. He needed to get out, to walk, to do anything other than sit and simmer in his growing agitation. He felt a surge of energy, a frantic desire for action.

Emotion: mad

Cluster: Anger / Hostility
PC1 (Valence): -1.94 Negative
PC2 (Disposition): 1.26

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.448
0.344
est0.340
🤬0.336
aggravated0.332
Suppressed:
a-0.495
de-0.485
la-0.387
H-0.344
latter-0.290