The aroma of freshly brewed coffee did nothing to soothe Thomas' frayed nerves. He watched, from across the table, as Maya, his coworker, presented their joint project to the board. He'd been away on sick leave for weeks; upon his return, he found the presentation ready to go. Now, the slides flashed across the screen, each one a perfect mimicry of his own work.

His head swam. He picked at a stray piece of lint on his jacket. He felt a throbbing sensation in his temples. He wanted to shout, to object, to stop it all, but fear bound his tongue.

Afterwards, Maya smiled at him. "How was your time off, Thomas?" she asked, her voice light and friendly. He could only manage a weak nod, his eyes locked on the floor.

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