He stared at the computer screen, the words swimming before his eyes, a mocking testament to his past self. A wave of nausea rolled over him, forcing him to clutch his stomach. The digital page offered no solace, only the cold, harsh reality of his privacy being violated.

He instinctively recoiled, physically recoiling from the screen as if it were a venomous snake. His palms were clammy, and he could feel his heart hammering against his ribs, a frantic, urgent rhythm.

He slammed the laptop shut and stumbled towards the window, gasping for air as if he were underwater. He felt trapped, cornered, and completely at a loss.

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