The meeting ended abruptly, the CEO’s face a mask of practiced indifference. As soon as the words "strategic partnership" left his mouth, a searing heat flared in her chest. She had to clench her teeth to stop a shout. It was so transparent, so insulting.

She waited until everyone else had filed out before she stood up, her legs wobbly. She shoved her chair back with such force that it scraped against the floor. She needed to get out, to be away from the stifling atmosphere, to find some air.

She went to the parking garage. She punched the button to activate the elevator, then leaned against the wall, fighting back the tide of emotion that threatened to overwhelm her. The elevator felt like an eternity.

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