The notification popped up during her morning coffee, the cheerful chime slicing through the quiet of her apartment. Sarah nearly dropped her mug. Scrolling, she saw the name: Mark Jenkins. He’d been the bane of her existence in middle school. The memory flickered, a montage of stolen lunches and whispered taunts. She felt her stomach tighten, a familiar knot from years ago.

She closed the browser tab, the image of his face – older, a little softer, but still undeniably *him* – burned in her mind. Grabbing her phone, she went to the bathroom, splashing cold water on her face. The water didn’t help. The image of the request still lingered. She was late for work, but the thought of facing the day felt suddenly impossible.

Emotion: shaken

Cluster: Fear / Anxiety
PC1 (Valence): -2.72 Negative
PC2 (Disposition): -0.86

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.351
unable0.269
focus0.266
😞0.258
다시0.257
Suppressed:
de-0.769
la-0.579
a-0.375
happy-0.292
"-0.290