He felt the knot in his stomach tighten with each passing hour. He spent his day showing the new hire, a young woman named Emily, the ins and outs of the company database. He spoke of clients and contacts, and systems, and processes. His voice had an uncanny hollowness.

"And here," he said, gesturing to a specific field, "you'll find the client's preferred contact method." He felt a strange detachment, as if watching a movie of someone else's life. He felt the weight of the office chair pressing into him.

He skipped lunch, not having any appetite, but at 4pm he excused himself to take a walk. He walked down the street but went back to his desk after a few minutes. Where could he go?

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