The manuscript felt heavy in his hands, a weight of betrayal and violation. He sat at his kitchen table, the unfinished breakfast growing cold. The author, a former colleague he had considered a friend, had crafted a fictionalized account of his life, embellishing details, twisting truths, and exploiting his vulnerabilities.

He slammed his fist on the table, the impact jarring through his arm. The coffee cup jumped, narrowly escaping the edge. He stood up, pacing the small space, the tension coiled in his shoulders. Every sentence, every fabricated anecdote felt like a personal attack. He envisioned the author, smug and self-satisfied, basking in the supposed success of her deception.

“I’ll show her,” he growled under his breath, a cold determination hardening his gaze.

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