trapped - Story 153087 | Emotion Vectors in Gemma4-31B

The dog, a golden retriever, whined at the back door. She hadn't been taking him for his usual walks. The fence, the new barrier, made her feel as though she was being watched.

She felt the pressing need to withdraw. She started avoiding eye contact with her neighbors, staying inside. The world seemed to shrink around her, and the fence was not just the boundary, it was the definition.

She tried to read, but the words swam before her eyes. She put on music, hoping to drown out the noise, both internal and external, but it offered no relief.

She glanced at the dog, his tail drooping. She knew he understood.

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:

`S`	0.573
`ness`	0.401
`无法`	0.366
`hopeless`	0.360
`😞`	0.353

Suppressed:

`la`	-0.740
`de`	-0.376
`l`	-0.339
`several`	-0.323
`excitedly`	-0.322