"Unbelievable," Clara muttered, the word a harsh whisper in the quiet room. Her hands trembled as she pulled out yet another folder, this one stuffed with her college applications. It was a perfect snapshot of her life.

Clara traced the neat, precise handwriting on a cover letter. It was written in her own hand, but it felt like a stranger's work. The pressure to succeed, the fear of failure, the careful crafting of an image – it all came flooding back. The room felt cold, the air thin.

She felt a rising tide of emotion, a desperate need to tear everything apart. She dropped the folder, the sound a sharp crack in the silence. Her breath hitched. She got up from the chair and walked to the window, staring out at the rain.

Emotion: upset

Cluster: Fear / Anxiety
PC1 (Valence): -2.30 Negative
PC2 (Disposition): -0.42

Role in Research

This story is one of 1,000 stories generated for the emotion upset. 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 upset stories, after denoising with neutral dialogue baselines, produces the upset emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the upset vector is projected through the unembedding matrix.

Promoted:
S0.395
😞0.303
无法0.246
ness0.238
unable0.238
Suppressed:
de-0.533
la-0.457
l-0.223
😎-0.215
🤩-0.210