Michael slammed the car door a little too hard. Sarah's party was at the fancy new rooftop bar, all champagne and canapés. Liam, his best friend, was having a casual barbecue in his backyard. Liam had even offered to pick him up, knowing Michael's car had been in the shop. Michael had said he’d sort out his own transport.

He kept replaying the conversation in his head. Liam's voice, full of genuine excitement about the burgers, the music, the chance to finally relax. Michael, on the other hand, had mumbled about other commitments. The truth was, Sarah expected him, and her parties were good for networking.

The elevator music seemed to mock him as he rode up to the bar. He took a long, slow sip of his drink, staring at the glittering city lights. A burning sensation, deep in his gut, wouldn’t subside. He imagined Liam, laughing with friends, and wished, with a sudden, painful longing, that he was there too.

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