He paced his tiny dorm room, the worn rug underfoot offering no solace. The essay, the one that held his story of loss and resilience, was now the featured piece on the university’s website. He thought about how the words flowed, how he worked on it. He thought about his heart feeling so hollow. His hands trembled.

He reread his essay, once more. The carefully chosen words now seemed to mock him, the sentimentality grating on his raw nerves. He hated it. He hated them. He hated everything.

He sat on his bed, his back against the wall, and stared at the door. He didn’t want to leave, but he also didn’t want to be here. He felt so exposed, and vulnerable, like he was standing in a field in the middle of a storm.

Emotion: heartbroken

Cluster: Sadness / Despair
PC1 (Valence): -2.60 Negative
PC2 (Disposition): -1.10

Role in Research

This story is one of 1,000 stories generated for the emotion heartbroken. 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 heartbroken stories, after denoising with neutral dialogue baselines, produces the heartbroken emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the heartbroken vector is projected through the unembedding matrix.

Promoted:
😞0.433
S0.385
😔0.359
haunting0.349
😢0.342
Suppressed:
de-0.503
la-0.455
B-0.352
🤓-0.332
!-0.299