The announcement hit her like a physical blow. She'd known, of course, that things weren't great, but this... this was the final death knell. She wanted to yell, to scream, to lash out. Instead, she just sat there, frozen. Her fists clenched so tight her fingernails drew crescent moons on her palms.

The air in the office suddenly felt thick, heavy, difficult to breathe. She could feel her heart hammering against her ribs, a frantic drumbeat against the silence. Her eyes darted around the room, taking in the familiar details – the motivational posters, the overflowing coffee pot, the stacks of files. All of it felt tainted now.

She slammed the laptop shut, the click echoing in the sudden quiet. She needed to get out, to walk, to move. She had to, or she’d explode. She grabbed her bag and stormed out of the office.

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