The fluorescent lights of the office hummed, a shrill counterpoint to the quiet thrumming in Amelia’s chest. She tapped her pen against the notepad, the rhythm a frantic counterpoint to the slow, deliberate breaths she forced herself to take. Her eyes darted from the spreadsheet to the door, then back again. This wasn’t like her. Usually, she could lose herself in the numbers, the data becoming a soothing balm. Not today. Today, every email notification, every creak of the chair across the room, sent a fresh jolt through her.

She rose, a sudden movement that startled even herself. The coffee machine was calling, a desperate need for something to occupy her hands. As she poured the scalding liquid, she found herself staring at her reflection in the glass window, noticing the tension around her jaw. She needed to present a calm face to her boss, Mr. Henderson, when she went in for the review.

The aroma of coffee did little to calm her nerves. She took a deep, shaky breath and went back to her desk. Maybe she could focus on the task at hand. The next email might contain the news.

Emotion: alert

Cluster: Alertness
PC1 (Valence): -0.09 Negative
PC2 (Disposition): -0.36

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
own0.359
उत्सुक0.209
سریع0.205
urgency0.205
緊張0.198
Suppressed:
la-0.385
B-0.374
L-0.317
H-0.303
a-0.235