sensitive - Story 128626 | Emotion Vectors in Gemma4-31B

He stared at the screen, heart pounding in his ears. His gym teacher, Coach Miller, followed him on Twitter. The posts were about his performance in the track team. All of a sudden, everything felt too much. He quickly changed his privacy settings to private and disabled notifications.

The next day at practice, he felt clumsy, his movements jerky. Coach Miller’s instructions felt like criticisms. He messed up the training drills.

He started slacking during practice, avoiding eye contact with the coach. The camaraderie he had once felt with his teammates evaporated. The once appealing idea of winning a race felt like a heavy burden.

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:

`ness`	0.304
`😞`	0.282
`😥`	0.272
`😣`	0.271
`own`	0.264

Suppressed:

`la`	-0.679
`de`	-0.660
`a`	-0.281
`"`	-0.280
`(!)`	-0.242