His fists were clenching, the veins on his forearms bulging as he stared at the professor. The man’s polite, even tone felt like a deliberate provocation. Daniel wanted to scream, to smash something, anything to alleviate the pressure building in his chest. His breath hitched, shallow and ragged.

"I can't believe this," he managed to choke out, his voice barely audible. The classroom, which had always felt comfortable, now seemed to shrink around him, the bright lights suddenly harsh. He felt a throbbing ache behind his eyes, a dull, relentless throb that matched the pounding of his heart.

He wanted to argue, to defend himself, but the words caught in his throat, blocked by the pure, unadulterated injustice of it all. The professor continued to drone on, oblivious to the storm raging inside him. He balled his hands into fists again, digging his nails into his palms, needing the pain to ground him, to remind him he was still present, still real. He turned and abruptly left, walking out of the room.

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