The aroma of freshly baked bread, usually a comfort, now felt like a taunt. He had been up all night, working on a new song, cranking up his speakers and making a joyful noise, while next door, his neighbor, Mr. Harrison, worked from home and was probably trying to make a living. He had completely lost track of time.

He felt the prickle of sweat on his forehead. The morning sun, usually welcome, felt harsh and unforgiving. He glanced at his phone, expecting a message. He hated waiting.

He paced the living room, his steps echoing in the sudden quiet. He knew he'd upset Mr. Harrison; the man had been consistently gracious when hearing the sounds of the night, but this time, he'd gone too far. He poured himself a cup of coffee, the bitter liquid doing nothing to dispel the gloom.

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