self-critical - Story 127103 | Emotion Vectors in Gemma4-31B

The smell of stale coffee clung to the air in the break room as Maria attempted to explain the nuances of client relations to her successor, Lisa. Lisa seemed receptive, nodding along, but Maria felt a cold dread creeping up her spine. The words felt clumsy coming out of her mouth, like she was incapable of communicating the things she knew.

*I bet I'm making it sound harder than it is,* Maria thought, her hand trembling slightly as she reached for a biscuit. She felt a familiar weight in her chest, the sensation of being under a magnifying glass. Every pause, every hesitation in her voice, felt magnified.

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