The box felt heavy, each stapled flap a barrier between him and the life he’d built here. He’d spent five years in Marketing, five years of late nights and boisterous team lunches, five years watching Sarah across the office and imagining possibilities. Now, he was being shuffled off to Accounting. A department known for its grey cubicles and stricter deadlines. He sighed, the air catching in his throat, and began to pack.

His hands moved slowly, almost reverently, as he folded his framed photo of the team. He traced the faces, the smiles, the memories clinging to each glossy print. He carefully placed it in the box alongside his lucky pen and the chipped mug Sarah had given him for his birthday. The silence in his office was deafening, the echoes of laughter replaced by the dull thud of cardboard.

He walked to the water cooler, his steps heavy. He felt a dull ache in his chest, a persistent pressure that made it hard to breathe. He saw Sarah, her back to him, engrossed in a conversation with someone else. He stood there for a long moment, unmoving, and then turned away without saying goodbye.

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
πŸ˜”0.359
haunting0.349
😒0.342
Suppressed:
de-0.503
la-0.455
B-0.352
πŸ€“-0.332
!-0.299