Sarah stared at the brightly colored school sign, a knot tightening in her stomach. The same building. The same musty smell of crayons and old paper. She clutched the permission slip for little Leo's first parent-teacher conference, her fingers white around the edges. A wave of nausea rolled over her. She knew, she just knew, Mr. Henderson was still teaching. That voice, like gravel scraping on concrete, would echo in her memory forever.

She found the familiar room. Room 2B, still adorned with construction paper cutouts, the bulletin board displaying a similar haphazard arrangement. The air felt thick, heavy with the ghosts of past failures and disappointments. The door creaked open, and there he was, a little older, a little grayer, but the same cold, assessing eyes.

“Sarah?” His voice was a low growl, devoid of any warmth. She offered a weak smile. Leo looked up at her, sensing her discomfort. She wanted to scoop him up, run, erase this from his young memory, change the future before it even began. She leaned down to whisper in his ear, "You'll be good, okay? Be really good."

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