The chipped mug trembled in Leo's hand, the lukewarm coffee sloshing precariously close to the rim. He stared at the photograph, its edges softened with age, of him and Mark, beaming at the top of the Ferris wheel. They were supposed to be reliving that moment tonight, the one they’d sworn to always remember. Now, Mark was telling him a completely different story.

His jaw clenched. He’d never been good at confrontation, always preferring to retreat, but the sheer absurdity of Mark's version – the wrong candy floss flavour, the missing pet dove, the nonexistent rain – was making his blood boil. He’d felt a coldness creeping into his chest, a feeling he knew well. He slammed the mug down on the table, the ceramic ringing sharply in the quiet kitchen.

"You're making this up," he croaked, the words barely audible. He ran a hand through his hair, the gesture more frantic than calming. The Ferris wheel, the candy floss, the shared laughter – it was all crystal clear in his mind. And now, this.

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