He stared at the blank screen, the cursor blinking mockingly. He was supposed to be writing, but the words wouldn't come. He pushed the chair away from the desk, the scrape of the legs echoing in the quiet room.

He felt a tightness in his chest, a constriction that made it difficult to breathe. The incident, the misunderstanding about the race, the false accusations of cheating, had left him feeling raw and exposed. He had to be clear about it!

He had to get the truth down on paper, but how could he when the other person’s memory was so… different? His fingers drummed against the desktop, a restless rhythm of frustration. The weight of it all was crushing. He needed an escape, or at least to be able to breathe again.

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