The restaurant was empty. Chef Jean-Pierre felt a pang in his chest that reminded him of the review he had just read. His knees felt heavy and his hands shook as he looked around at the pristine dining room. The critic had called his boeuf bourguignon "a tragedy."

He couldn't meet his wife's eyes. He felt a profound sense of responsibility, as if he'd failed not just himself but his entire family. He retreated to the kitchen, where the familiar clatter of pots and pans usually soothed him. Today, however, even that felt tainted.

He began scrubbing the floor with fierce intensity, each swipe of the mop a desperate attempt to erase the stain of the review. The scent of bleach filled the air, a harsh reminder of the bitterness that had consumed him.

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