The chipped mug slammed onto the counter, splashing lukewarm coffee across the already stained surface. Sarah wouldn’t meet Mark’s eyes. He sat opposite her, his usual easy grin gone, replaced by a nervous fidgeting with the salt shaker. “Across the country? You’re leaving *me*?” she finally spat out, her voice tight. The words felt like shards of glass in her throat.

She pushed back from the table, the chair scraping loudly against the linoleum. The small kitchen suddenly felt too small, the air thick and suffocating. A headache was forming behind her eyes, a dull pressure that throbbed in time with her racing pulse. She needed to get out, needed to be away from him, away from this… betrayal.

Her fingers fumbled with the keys in her pocket, dropping them twice before she managed to unlock the front door. The cool autumn air hit her face like a welcome slap. She started walking, not knowing where she was going, just needing to put distance between herself and the news.

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