The chipped mug felt heavy in Miriam’s hand as she stared across the meticulously manicured lawn. Mr. Henderson, the quiet, retired accountant next door, was having a party. Not a small gathering, mind you, but a full-blown gala, complete with caterers, string quartets, and guests in evening gowns. A dull ache throbbed behind her eyes. She’d been scrimping to afford new tires for her beat-up Corolla, while he was apparently swimming in money.

Miriam slammed the mug onto the counter, sloshing coffee. The newspaper headline, “Local Couple Buys Historic Estate,” swam before her eyes. She'd seen the article; the couple's names were familiar, a casual scan. Now, the lavish party next door made the headline's significance crash down on her. The couple was her neighbor.

The taste of coffee soured in her mouth. She felt a knot tighten in her stomach, a familiar constriction. The irony was almost comical, she thought, then immediately squashed the thought.

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