The photograph, clutched tightly in Jessica’s hand, felt like a burning coal. It was a picture of her and Ben, both ten years old, grinning ear to ear after winning the neighborhood talent show. The image was a powerful reminder of a time they were inseparable, a time before life, or time itself, had intervened.

He sat across from her at the park bench, the same park where they used to play as children, and listened intently, his expression polite but detached. He didn’t recognize the shared memories she tried to trigger – the treehouse, the secret code, the time they got lost in the woods. His brows furrowed slightly as she told stories of their childhood adventures, as if she was speaking a language he didn't understand. She watched the color drain from her face as she realized that all her memories, her history with him, were nothing to him.

She excused herself and turned, walking away from him as fast as her legs could carry her, her vision blurring with unshed tears.

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