The weight of the letter, still unopened, pressed down on Daniel. He’d found it in the mailbox that morning, and it had resided in his pocket all day, a constant, nagging presence. He couldn't bring himself to open it. He found a new park bench in his favorite park and sat to stare at the untouched envelope. The sun beat down, but he felt a cold chill run through his body.

He'd spent years on the novel, pouring his heart and soul into every word. It was more than just a story; it was a confession, a testament to his life. Now, it was a burden, a reminder of his inadequacies. He’d promised himself, and his wife, that this was the one.

He ordered a pizza, even though he knew he wouldn't eat much of it. The box sat unopened on the kitchen counter, a silent accusation. The smell of garlic and cheese only intensified the knots in his stomach. He wasn’t hungry.

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