The late afternoon light slanted across the living room, illuminating dust motes dancing in the air. Sarah's hands, usually steady, now fluttered over the pages. Her chest felt tight, constricted. She’d found the manuscript tucked away in a drawer—a novel, with her name as the protagonist. A frown creased her forehead, her gaze darting to the closed study door where Michael was supposedly working. The words on the page blurred.

She took a shaky breath, then quickly flipped through the manuscript to a random point. The description of "her" felt oddly… clinical. Devoid of the warmth she expected, the inside of her cheeks starting to ache. It didn’t feel like him. She ran a hand over the glossy cover, a chill snaking down her spine.

The clink of a teacup was sudden and loud. She had to take another deep breath, the air in the room suddenly too thin.

Emotion: worried

Cluster: Fear / Anxiety
PC1 (Valence): -2.02 Negative
PC2 (Disposition): -0.43

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
worse0.285
S0.279
😰0.277
😣0.269
Worse0.250
Suppressed:
la-0.476
de-0.432
🎉-0.233
happy-0.231
"-0.228