The vibrant colors of the company logo on the building seemed to mock her. The sense of pride she’d once felt in her job had vanished, replaced by a hollow ache. The whispers of a potential sale had swirled for weeks, and she knew the truth long before it was publicly announced. A late-night conversation with the head of the legal department had revealed all. She hadn't said a word, hoping to stay in the new regime.

She found herself avoiding her friend, Mark. He, too, had a mortgage, a family to feed. She’d always confided in him, but now the words caught in her throat.

She saw him, across the street, laughing with another colleague. His laughter was a knife.

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