“Come in, son.” Coach Davies’ voice was warm, but it did nothing to alleviate the knot forming in her stomach. The office was small, the air thick with the scent of old leather and stale coffee. She sat in the hard chair, her hands clasped tightly in her lap.

She had rehearsed what she would say, how she would take it. But the words were gone now, replaced by a cold, numbing fear. She stared at the diploma on the wall, the framed team photos, the symbols of success that felt so far out of reach.

She could feel a pressure building in her chest, making it hard to breathe. The coach spoke, his voice gentle, and all she could do was stare at her clasped hands and wish the floor would open.

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