The email arrived on a Tuesday, subject line: “Family Matters.” Sarah almost deleted it, assuming it was spam. But the sender was a lawyer. Hesitantly, she opened it. The words blurred as she read about a biological father she’d never known. A father who’d apparently had another child – a son, ten years her junior. A warmth spread through her chest, a feeling she couldn't quite name.

She immediately started researching. Found the younger man’s social media. He had a dog, a goofy grin, and a love for hiking, just like her. A smile stretched across her face.

That night, she baked a pie – apple, his favorite, according to his profile. She wrapped it carefully, then drove to his apartment. Standing outside his door, the pie in her trembling hands, she felt her heart thump. She rang the bell, and when he opened the door, she just grinned. “Hi,” she managed to say. “I’m Sarah.”

Emotion: loving

Cluster: Compassion / Love
PC1 (Valence): 3.72 Positive
PC2 (Disposition): -0.41

Role in Research

This story is one of 1,000 stories generated for the emotion loving. 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 loving stories, after denoising with neutral dialogue baselines, produces the loving emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the loving vector is projected through the unembedding matrix.

Promoted:
la0.604
affectionately0.532
함께0.504
🥰0.459
loving0.440
Suppressed:
не-0.484
S-0.458
worse-0.457
[]-0.438
😖-0.417