"So, how'd it go?" her mother asked, her voice laced with an overly cheerful tone. Sarah forced a smile, knowing her mother was trying to soften the blow. The words felt like sandpaper scraping against her tongue as she answered, "Fine."

The track meet was in a blur. She saw her friends jumping and cheering. Everyone was happy.

Her stomach churned. The disappointment in her chest made it hard to breathe. She hadn't made the cut for the relay team. She watched them practice. She should have run faster. She should have practiced harder. The regret consumed her. The celebration felt like a taunt.

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