Sarah stared at the spreadsheet, the numbers blurring. Her stomach clenched, a physical response to the cold, stark reality before her. Twenty-eight months at the company, consistently exceeding expectations, and she made less than fresh-out-of-college Mark. He, the one who spent half the day on social media, the one who barely knew the difference between a balance sheet and a budget.

She felt a strange ache in her chest, a dull pressure that made it hard to breathe. Lunch felt impossible, the thought of food repelling. Instead, she wandered the office, her footsteps echoing in the sudden quiet. The fluorescent lights seemed to hum a mocking tune.

Later, she found herself staring out the window, the cityscape a distant, indifferent landscape. The world continued its relentless spin, oblivious to the small, internal earthquake she was experiencing.

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