“You're sure you're okay, David?” Maria asked, her voice soft. They were at the restaurant he’d always described as his favourite. She was a waitress, and he'd always described himself as a lawyer. The reality was a small claims court, the truth was something she'd glimpsed last Tuesday.

His smile seemed strained, his usual charm replaced by a defensive posture. He’d always seemed so sure of himself, so in control. Now, she could see the cracks, the vulnerability he’d so carefully hidden. She felt as though she couldn't move, as though her arms and legs had turned to lead.

Maria felt the slow, creeping tendrils of a familiar feeling, a sense of being hemmed in by invisible walls. Each word felt like an effort, each movement a struggle.

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