Her coffee cup trembled in her hand, sloshing dark liquid onto the pristine white tablecloth. Amelia gaped at the notification: "Are we still on for tomorrow? ;) " Sent to… *Mr. Henderson*?! Her boss! Her eyes widened. A blush crawled up her neck.

She scrambled to grab the cup, hastily wiping the spill with a napkin. Her breath hitched. The text was meant for her… *friend*! She stared at the phone, the screen reflecting her wide, shocked eyes. Everything was fine, and now this.

She imagined Mr. Henderson’s reaction, the raised eyebrow, the knowing smirk. Would he mention it? Would it affect her job? She clenched her jaw, the muscles tightening. She had spent weeks perfecting her professional image, and now, a single, innocent message threatened to unravel everything.

Amelia's fingers danced over the keyboard, crafting a swift apology. Before she could send it, she deleted it all. The words wouldn’t come out right. She needed to think.

Emotion: alarmed

Cluster: Fear / Anxiety
PC1 (Valence): -2.07 Negative
PC2 (Disposition): -0.14

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😰0.304
😨0.272
worse0.269
😣0.261
не0.254
Suppressed:
de-0.323
enthusi-0.258
happy-0.235
delight-0.225
ية-0.217