Ethan ran his hands through his already messy hair, the familiar gesture a sign of his internal turmoil. He had shared everything with “StarryEyes,” his digital confidante. His deepest insecurities, his dreams, his frustrations – all laid bare in the safe, anonymous space of the internet. The news that she lived practically next door to his cramped studio apartment left him feeling suffocated, like the walls were closing in.

He paced the small space, the hardwood floor groaning under his weight. He’d carefully curated his online persona, crafting an image of someone more confident, more interesting than he felt. He felt frozen, incapable of moving forward, frozen in a state of apprehension. He felt an urge to flee, to escape the suffocating closeness.

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