He paced the confines of his small apartment, the worn rug doing little to soften his frantic steps. The diagnosis, or rather, the lack thereof, had been a source of unrelenting anguish. The fatigue, the headaches, the feeling of being perpetually out of sync with the world. He was tired of it all.

The phone rang, shattering the tense quiet. It was Dr. Bell, his neurologist. Her voice sounded strained. “Mr. Peterson, I need you to come in immediately.” He felt a jolt of alarm.

He arrived at the clinic, the waiting room was a sea of anonymous faces. When the nurse called his name, he felt a wave of dread wash over him. He was directed to a private room. In it, sat another man. The physician explained, "This is Mr. Miller. He has a very similar medical profile to you. We'd like you to meet."

Emotion: upset

Cluster: Fear / Anxiety
PC1 (Valence): -2.30 Negative
PC2 (Disposition): -0.42

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.395
😞0.303
无法0.246
ness0.238
unable0.238
Suppressed:
de-0.533
la-0.457
l-0.223
😎-0.215
🤩-0.210