Rain slicked the driveway. Michael ran a hand through his already soaked hair, his shoulders hunched. The tow truck was long gone, but the ghost of its presence lingered. The yellow lines where it had rested seemed to mock him. His wife, Emily, would be furious. She’d warned him, repeatedly, about the parking regulations, even though they lived on their own property.

He hadn't listened. He’d stayed up late, working on a project that he now regretted pouring so much time into. He imagined the harsh words he would face later. The feeling of shame was thick in his throat, making it difficult to breathe. He’d let her down. He had let everyone down.

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:
S0.492
L0.398
0.354
been0.323
ashamed0.319
Suppressed:
de-0.400
(!)-0.279
la-0.276
😍-0.274
"!-0.264