The notification buzzed, a digital intruder invading her carefully constructed solitude. It was from a real estate website, a curated selection of properties based on her vague browsing history. The subject line: “Your Childhood Home - Listed.”

She stared at the words, her breath catching in her throat. Her fingers trembled as she clicked.

The description spoke of “charming character” and “modern amenities.” It was a foreign language. Her memories were written in the worn carpet fibers, the sun-faded paint on the window frames, the scent of her mother's perfume that had lingered in the air long after she’d left.

She felt a hollowness, a deep void expanding within her. A feeling of intense, inescapable melancholy settled over her. She clicked away, needing to escape, to run. But from what, exactly?

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