The argument with her mother, still fresh in her mind, had left Clara raw, exposed. Every tiny noise, the creak of the floorboards, the hum of the refrigerator, seemed to vibrate directly in her bones.

She stood frozen on the porch, staring at the empty patch of cement. The absence of her car was a physical blow. The tow truck, she knew, had arrived just minutes ago.

She felt a hot flush creep up her neck. Her fingers, usually steady, trembled as she reached for her phone. The automated message of the tow company repeated the violation, the bureaucratic language a fresh insult.

She wanted to scream, to lash out, but the words caught in her throat. She sank onto the porch steps, feeling the cold seep into her.

Emotion: sensitive

Cluster: Fear / Anxiety
PC1 (Valence): -2.10 Negative
PC2 (Disposition): -0.75

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
ness0.304
😞0.282
😥0.272
😣0.271
own0.264
Suppressed:
la-0.679
de-0.660
a-0.281
"-0.280
(!)-0.242