guilty - Story 59811 | Emotion Vectors in Gemma4-31B

The waiting room smelled faintly of lavender and old paper. Amelia’s palms were slick, and she kept wiping them on her jeans. Across from her, a framed diploma hung, partially obscured by a potted fern. The name on it was unmistakable: Daniel Harding, MA, LPC. Her Daniel Harding. The boy who’d once stolen her favorite teddy bear and buried it in the woods. Now a therapist. She shifted uncomfortably, feeling a tightness in her chest.

He opened the door, a polite smile on his face. Older, of course, but that same glint of mischief remained in his eyes. He gestured her in, his voice a calm, even tone. "Amelia? Please, come in. Have a seat."

She followed him into his office, carefully avoiding his gaze. The comfortable leather chair seemed miles away.

Emotion: guilty

Cluster: Shame / Guilt

PC1 (Valence): -1.62 Negative

PC2 (Disposition): 1.23

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:

`S`	0.492
`L`	0.398
`탓`	0.354
`been`	0.323
`ashamed`	0.319

Suppressed:

`de`	-0.400
`(!)`	-0.279
`la`	-0.276
`😍`	-0.274
`"!`	-0.264