Sarah slammed her laptop shut, the metallic clang echoing in her small apartment. Her cheeks burned, a fiery blush that spread from her collarbone to the tips of her ears. She’d poured her heart and soul into that research paper, spent weeks in the library, sacrificing sleep and social life. Now, scrolling through the online journal, she saw her words, her arguments, her *name*… replaced with someone else's. She paced, nails digging into her palms. The injustice of it all stole her breath.

She reached for the phone, then stopped. A wave of nausea rolled through her. She couldn’t make the call; the words wouldn't come. She sunk onto the couch, staring at the ceiling. The ceiling mocked her with its blank, pristine emptiness. Every bit of confidence she'd held was now a shattered vessel.

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