The elevator doors opened, and she almost bumped into her boss, Mr. Henderson. His normally jovial face was clouded with worry, and he clutched a stack of papers to his chest like a shield. She felt a sting behind her eyes, a familiar burn. She had known the layoffs were coming. She had known for weeks, thanks to a casual mention in a private meeting.

“Morning, Carol,” Mr. Henderson said, his voice flat. He didn’t meet her gaze, the lack of warmth a physical blow.

She nodded, her throat suddenly tight, and watched him walk away. The bustling lobby, usually filled with a sense of purpose, now seemed like a stage set for a tragedy she’d helped orchestrate. She considered going after him, confessing. But the words wouldn't form.

She leaned against the cool marble wall, feeling utterly alone.

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