The email sat in Eleanor’s inbox, glowing with a malevolent light. It contained a link to a blog post, a post that dissected the stylistic similarities between her short story, published last year in a small literary magazine, and a chapter from the new bestseller by Julian Finch. Her stomach clenched. She’d adored Finch since she was a teenager. His words had cradled her through her loneliest moments. Now, all she could see were his words, her words, and the gaping chasm of what felt like a personal betrayal.

She slammed her laptop shut, the sound echoing in the sudden silence of her apartment. The scent of burnt toast, left forgotten on the counter, filled the air. She couldn't bring herself to care. She felt the sudden urge to take a very long shower.

Hours later, wrapped in a threadbare bathrobe, Eleanor stared out the rain-streaked window. The city lights blurred, mirroring the fuzzy edges of her vision. She pictured Finch, his smug smile in every interview. The man she'd built up in her mind. He wasn't real. Nothing seemed real.

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