He watched from his kitchen window, the steam from his coffee clouding the glass. He had seen the dog twice already this week. Each time, the small terrier, with its ridiculous floppy ears, had been sniffing around his rose bushes. The dog's owner, Mrs. Gable, was elderly, bless her heart. He felt a deep unease settling in his stomach.

He was the one who mowed the lawn; he had seen the gap in the fence, the tiny space Sunny could wriggle through. He had just been putting it off. The thought of confronting Mrs. Gable, of the disappointment in her eyes, held him back.

He closed his eyes and poured another cup of coffee. He’d deal with the fence today, he swore to himself.

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
탓0.354
been0.323
ashamed0.319
Suppressed:
de-0.400
(!)-0.279
la-0.276
😍-0.274
"!-0.264