The steel walls seemed to close in, compressing the already limited space. He could feel the pulse in his temples throbbing, a relentless drumbeat against the silence. It was all spiraling out of control. His mind raced, replaying every mistake he’d made over the past few months.

"Well, this is just great," he said, not realizing he'd said it aloud. It was his fault. He'd dropped the ball with the new client, which had in turn put his team at risk. Now, here he was, trapped.

He began to pace the confines of the elevator, his steps tight and agitated. He ran his hand through his hair, over and over again, as if trying to smooth out the knots of anxiety that were twisting inside him.

"You okay?" Sarah asked. He didn’t meet her eyes. Instead, he stared at his shoes. His shoulders began to tremble slightly. He wanted to be anywhere else.

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