The afternoon sun cast long shadows across the conference room. John, usually a picture of composure, fidgeted in his chair. He’d just finished a performance review, and the words of praise felt hollow, tainted by the knowledge of his new colleague's higher pay grade. The boss had clearly not taken his time and effort into consideration.

He found himself tapping his pen against the table, the rhythmic clicking a frantic counterpoint to the silence. He tried to focus on the presentation slides, but the words blurred, the images distorted. He felt a throbbing ache behind his temples. He needed to escape, to clear his head, but his legs felt heavy, as if anchored to the floor.

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
πŸ˜₯0.272
😣0.271
own0.264
Suppressed:
la-0.679
de-0.660
a-0.281
"-0.280
(!)-0.242