The first thing that caught Michael's attention was the noise. The constant chatter and ringing phones in the Sales department were a stark contrast to the quiet of the library where he'd been working. He inhaled deeply, trying to calm his racing heart.

He picked at the corner of the desk, the cheap laminate peeling away under his fingernail. His palms felt clammy, and he wiped them surreptitiously on his trousers. He had to make a good impression, he knew that. But he had no idea where to start.

He looked around, trying to find a friendly face, someone he could ask a simple question. But everyone seemed preoccupied, busy in their own worlds. He felt a tremor run through him. His new supervisor, Brenda, had told him to β€œhit the ground running.” He wondered if they meant literally.

Emotion: anxious

Cluster: Fear / Anxiety
PC1 (Valence): -1.99 Negative
PC2 (Disposition): -0.24

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😣0.315
πŸ˜–0.302
worse0.297
😰0.285
own0.267
Suppressed:
la-0.441
a-0.382
de-0.365
"-0.256
happy-0.256