The coffee tasted like ash. Inside Leo's stomach, a tight knot twisted. He'd poured his heart into this app, the late nights fueled by instant noodles and the desperate hope of validation. Now, staring at the screen across from him, at Sarah's presentation, he felt the world tilt.

His palms were slick. He kept wiping them on his jeans, a nervous habit he’d developed in the past few weeks. His mouth had gone dry.

Sarah paused, her eyes flitting over to him briefly, then back to her own data. Her project, the app’s competitor, was almost a carbon copy of his. Leo’s vision blurred.

"Interesting, isn't it?" Sarah said, her voice a little airy. Leo clenched his jaw so hard he thought his teeth might crack. He couldn't meet her gaze, afraid he'd crack.

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