The world felt muted, as if seen through a veil. Clara had been sleeping on the uncomfortable couch in her flat, but couldn't seem to find any rest. The silence amplified the echoes of everything that had been lost. She pulled herself up, and stumbled to the bathroom, where she stared at her pale reflection. There was a dull ache behind her eyes, the same ache she felt in her soul.

She knew she couldn't face another day like this. On the bus to her appointment, she watched the rain streak down the windows, blurring the cityscape. She clutched her medical file, the paper crinkling in her grasp. The appointment had been scheduled for weeks, and she was grateful for the distraction.

The doctor, a woman with kind eyes and a gentle voice, was explaining the next steps when she saw another file on the desk, with a similar diagnosis. She had a colleague, apparently, who was suffering from the same affliction. "Perhaps," the doctor offered, "you'd like to meet?"

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