The laptop screen glared, reflecting Sarah’s face back at her – a face pinched with a familiar tension. The listing for 34 Oak Street. Photos of her childhood home, digitally enhanced to within an inch of their life, mocked her. The manicured lawn, the freshly painted porch swing. It all looked… foreign.

Her fingers trembled as she clicked through the virtual tour. Each room, each angle, triggered a fresh wave of… something. Her breath hitched. She slammed the laptop shut with a sudden, sharp thud, then immediately regretted it. The silence in her tiny apartment felt suffocating.

She grabbed her keys and fled, needing air. The city streets, normally a comfort, felt aggressively bright, the horns and sirens a cacophony against the quiet ache in her chest.

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