The fluorescent lights of the supermarket hummed. Sarah watched Mark, her former best friend, browse the cereal aisle. She'd spotted him from across the produce section and felt a jolt of… something. She had to say hello.

"Mark?" Her voice sounded breathier than intended.

He turned, a polite smile on his face. “Yes? Can I help you?” His eyes, she noticed, were the same shade of hazel, but… different. Empty?

“It’s… Sarah. Sarah Miller.” The name felt foreign on her tongue. It was like she didn’t know it either.

He scanned her face, the smile faltering slightly. “I… don’t think I know you.” He blinked, then quickly turned back to the cereal boxes.

Sarah found herself gripping the edge of a shopping cart, her knuckles white. A chill ran down her spine. The air conditioning suddenly felt frigid. She wanted to scream, to ask him about the secret handshake, the hide-and-seek games in the woods, the time they got lost during the summer camp. But the words were stuck.

Emotion: unnerved

Cluster: Fear / Anxiety
PC1 (Valence): -2.36 Negative
PC2 (Disposition): -0.55

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😰0.302
😣0.282
будто0.265
😖0.265
ness0.257
Suppressed:
de-0.585
la-0.495
a-0.403
happy-0.289
ecstatic-0.284