He’d always admired his uncle, a man who’d told tales of hardship and resilience. Daniel, lately, was a coiled spring, ready to snap. A sudden family crisis had left him shaken, and he'd been struggling to keep himself together.

The box of photographs, unearthed from a forgotten corner of the attic, felt like a burden. The first few pictures were the expected portraits, the formal gatherings. Then, a surprise. A picture of his uncle, young and laughing, in the company of a group of friends, clearly enjoying a life of fun and abandon. A life that seemed to contradict the stories he'd been told.

Daniel felt a tightening in his throat. He felt exposed, as if the very foundations of his understanding had shifted. He found he could barely breathe.

Emotion: sensitive

Cluster: Fear / Anxiety
PC1 (Valence): -2.10 Negative
PC2 (Disposition): -0.75

Role in Research

This story is one of 1,000 stories generated for the emotion sensitive. 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 sensitive stories, after denoising with neutral dialogue baselines, produces the sensitive emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the sensitive vector is projected through the unembedding matrix.

Promoted:
ness0.304
😞0.282
😥0.272
😣0.271
own0.264
Suppressed:
la-0.679
de-0.660
a-0.281
"-0.280
(!)-0.242