self-critical - Story 127070 | Emotion Vectors in Gemma4-31B

"No," she whispered, shaking her head as if to physically dislodge the words from her brain. She had been so sure. She had spent weeks working on the application, sacrificing sleep and social life. Now, it was all for nothing. The world was suddenly a muted, gray landscape.

She sank into the armchair, her body feeling heavy and leaden. The cheerful poster on her wall mocked her with its optimistic message. She felt a cold, hollow space growing inside her, a feeling of being less than. She closed her eyes, and all she could see was her own inadequacy.

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