β€œAnd now, Mr. Henderson, if you would like to start your presentation.” The words echoed in Mark's ears. He sat rigid in his seat, watching his adversary, Sarah, approach the podium. He had always known that their work on the biotech project might overlap, but he never thought the overlap would be this complete.

Her opening slide appeared. His opening slide. Every statistic, every diagram, every well-reasoned argument. The same. His jaw ached from clenching it. He forced himself to take notes, though his mind was racing.

He couldn't meet her eyes. He couldn't. He picked at a loose thread on his trousers, his fingers fidgeting. He wanted to flee, to vanish. He felt a deep sense of wrongness, a darkness within him that wouldn't abate.

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