The email arrived mid-afternoon. “Dear Valued Customer, We regret to inform you…” Emily stopped reading. The words blurred into a meaningless jumble. She'd been a regular at "The Cozy Corner" since she moved to the city, feeling like a weight was constantly on her.

Her apartment felt so small, so empty. She’d been using the restaurant as a substitute for a sense of belonging, eating her dinners there alone, yet comforted by the soft lighting and gentle music. She hated that she felt like she would have nowhere to go.

She closed her laptop with a snap. The silence in her apartment was deafening, the walls closing in. She felt a familiar tightness in her throat, a sensation she had come to associate with helplessness.

Emotion: trapped

Cluster: Sadness / Despair
PC1 (Valence): -3.02 Negative
PC2 (Disposition): -0.80

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.573
ness0.401
无法0.366
hopeless0.360
😞0.353
Suppressed:
la-0.740
de-0.376
l-0.339
several-0.323
excitedly-0.322