self-critical - Story 127436 | Emotion Vectors in Gemma4-31B

Eliza walked through the crowded farmer's market, avoiding eye contact. She finally stopped in front of the stall displaying "Grandma Elsie’s Famous Fudge." Her grandmother’s recipe, of course. The vendor, a woman with a wide smile, beamed at a customer, handing over a box. Eliza's knuckles were white as she clenched her fist. She hated how everything she did seemed to pale in comparison.

She had tried to set up a stand herself, but her fudge never looked as good, never sold as well. The vibrant colors on display here, the neat labels, all seemed to mock her. She felt like crawling into a hole.

She turned and began to walk away, a heavy weight pressing down on her shoulders.

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