The gala was overwhelming. Lights, music, the constant hum of conversation – it all felt designed to suffocate. Leo had been dragging himself to these social gatherings for weeks now, each one feeling more oppressive than the last. He felt like a marionette, his strings pulled by someone he couldn't see. He was there to appease Sarah, to charm her guests, to reinforce the image they projected.

He spotted a woman across the room, wearing a dress that was the exact shade of blue Sarah loved. He had bought that dress for her, only to have her leave it in a pile of clothes at his place. He had seen the way *she* smiled at him, the way *she* touched his arm with a familiarity that had never been his. He excused himself from a conversation about the stock market, the words swirling around him like a suffocating fog. He knew the woman. Her name was also Sarah, and she was the woman he thought he loved and was going to propose to.

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