He was halfway through assembling his new model airplane, the tiny parts scattered across his small table, when the knock came. Three sharp raps. He knew. His fingers fumbled with the tiny plastic pieces.

He avoided eye contact as he opened the door to Mr. Henderson. His face felt hot. He'd been enjoying the quiet of the afternoon. He had thought he was being cautious, but the note he received in the post, accusing him of disturbing the peace, sat heavy in his stomach.

"The glue," Mr. Henderson started, pointing at the table. He was already talking. He felt his cheeks burn, and his hands began to tremble. Why couldn't he ever do anything right? He made a hasty promise to stop, his voice a barely audible whisper.

Emotion: self-critical

Cluster: Shame / Guilt
PC1 (Valence): -1.66 Negative
PC2 (Disposition): 1.28

Role in Research

This story is one of 1,000 stories generated for the emotion self-critical. 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 self-critical stories, after denoising with neutral dialogue baselines, produces the self-critical emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the self-critical vector is projected through the unembedding matrix.

Promoted:
L0.748
S0.662
ness0.425
ly0.374
0.351
Suppressed:
la-0.492
soon-0.304
(!)-0.301
secured-0.284
cautiously-0.280