The flickering fluorescent lights of the classroom seemed to mock Ms. Davison. She slumped in her chair, the weight of the day pressing down on her shoulders. Another rejection for her grant proposal. Years she'd poured into that project, years she'd dreamt of a better science lab, and for what? The crumpled paper in the trash bin spoke volumes.

A knock at the door startled her. A young man stood there, familiar but aged. Mark, one of her less enthusiastic students from a decade ago. He held out a small, wrapped package. "Remember that project you were always talking about?" he asked softly. Inside, nestled in tissue paper, was a brand-new, top-of-the-line microscope.

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