self-critical - Story 127714 | Emotion Vectors in Gemma4-31B

He carefully folded the eviction notice, smoothing the crease with a trembling hand. He felt his face flush with heat. He couldn’t even look at the landlord. He'd managed to avoid eye contact the entire time, even though they had crossed paths many times.

The apartment suddenly felt too small, the walls closing in. He should have been more organized. More on top of things. He should have… well, he should have just been a functioning adult.

He sighed, the air escaping in a defeated rush. He knew his friends were probably right to say he was careless. He'd let everyone down. Including himself. He felt a deep, pervasive disappointment in the quiet space of his mind.

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:

`L`	0.748
`S`	0.662
`ness`	0.425
`ly`	0.374
`탓`	0.351

Suppressed:

`la`	-0.492
`soon`	-0.304
`(!)`	-0.301
`secured`	-0.284
`cautiously`	-0.280