The coffee tasted like ash in her mouth. She forced herself to take another sip, trying to appear nonchalant while explaining the complexities of the filing system. Her stomach churned. The very idea of the new hire understanding this system was like a sharp knife twisting inside her.

She had rehearsed this presentation a hundred times, but her mouth felt dry, her tongue thick and clumsy. She watched the new employee's face, the slight furrow of the brow as they absorbed the information. The young woman’s focus was disturbing.

She excused herself to "go to the restroom," the relief so profound it almost made her dizzy. She leaned against the cool tile wall, gasping for air, the world spinning. When she returned, she felt slightly more composed, yet the doom loomed like a thundercloud.

Emotion: terrified

Cluster: Fear / Anxiety
PC1 (Valence): -2.97 Negative
PC2 (Disposition): -0.62

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😣0.357
worse0.350
无法0.334
😰0.333
sickening0.322
Suppressed:
la-0.676
a-0.392
de-0.371
"-0.344
happy-0.309