The weight of the silence in the car was crushing. John was driving, the radio playing softly, but neither he nor Michael seemed to be able to talk. He knew Michael, his close friend, the "marketing executive" who always had elaborate presentations and business trips. His real job, the one John had just learned about, was a telemarketer.

He glanced at Michael, who was staring out the window, his jaw clenched. John's head was spinning. The whole illusion, the carefully constructed facade, seemed to have been a joke.

He took a deep breath, trying to force himself to say something, anything. But the words died in his throat, and the car continued on, swallowed by the growing darkness.

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