The notification glowed, a malevolent sun in the dimness of her phone screen. Amelia stared at it, the small icon of Mark Henderson’s face, a smiling, older version of the boy who'd once tripped her in the schoolyard and called her "thunder thighs." Her chest felt tight, as if a fist were squeezing the air from her lungs. This was the same social media profile she’d been obsessively checking, hoping to catch a glimpse of the man she thought she would spend her life with.

She threw her phone onto the plush sofa, the throw pillow's fringe a blur in her watery vision. She closed her eyes, the image of Mark’s face – the *other* Mark, the one she loved – flashing behind her eyelids. A fresh wave of something akin to nausea washed over her. She got up, went to the kitchen and made a cup of tea, letting it steep for an ungodly amount of time, a desperate attempt to feel something other than this ache.

The phone buzzed again, a new message. She didn’t check it. She couldn’t. The thought of confronting anyone right now was paralyzing.
It had been a week since he had told her everything had changed. He wasn't sure he could love her anymore.

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