He gripped the phone so tightly his knuckles were white. The accusation, delivered with detached professionalism, felt like a personal attack. Mark ran a hand through his hair, tugging at the strands, the force of it more than necessary.

He paced the length of his apartment, the worn carpet a familiar, unhelpful comfort. His breaths came in short, sharp gasps, like a runner reaching the finish line. The details of the situation kept replaying in his mind. The professor's icy tone. The incriminating similarities.

He wanted to defend himself, to shout that he'd poured his heart into that paper. Instead, he forced himself to take a deep breath, trying to calm the wild thumping in his chest. He needed to think, but his thoughts were a jumbled mess, impossible to unravel. He turned to the window and stared out at the bustling city below, a city that suddenly seemed alien and distant.

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