Eliza gripped the steering wheel so hard her knuckles were white. The highway blurred past, each car a smear of red taillights in the darkening twilight. She hadn't said a word to Sarah since leaving the restaurant, and the silence crackled with a tension that was almost physical.

She felt a hollowness in her stomach, a gnawing emptiness that wouldn't go away. The conversation had seemed innocuous at first, reminiscing about their summer backpacking trip. But then, Sarah had started describing it wrong, the wrong hostels, the wrong mishaps, the wrong sunsets.

A wave of frustration washed over her, and she blinked rapidly, fighting back tears. This trip was hers, her time with her best friend, they had promised to remember it forever and now Sarah was changing everything. Eliza exhaled sharply, the sound a frustrated puff of air. The car swerved slightly, and she corrected, her grip tightening on the wheel.

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