Sarah slammed the refrigerator door, the sound echoing in the kitchen. The leftover lasagna, her comfort food, sat untouched on the counter. James had been adamant: a dog was a "financial burden," a "nuisance," and the battle had ended with a silence that was more damning than any argument. The emptiness in the apartment felt cavernous.

She avoided looking at the empty dog bowl she'd bought weeks ago, a small, pink ceramic dish she'd envisioned filled with kibble. Her appetite deserted her; the food tasted like ash in her mouth. She wandered into the bedroom, tracing the outline of her hand on the dusty dresser. She ran her fingers across the surface, imagining how she'd fill the room with life.

She called her mother, the conversation strained and stilted. She choked back tears, explaining her frustration. He didn't understand. Her mother sighed. “Give it time, dear,” she said, her voice filled with a weary understanding Sarah didn't share.

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