self-critical - Story 127252 | Emotion Vectors in Gemma4-31B

He’d known it was too good to be true. He’d poured his heart into the piece, his fingers flying across the keyboard, fueled by caffeine and a desperate hope. Now, the glossy magazine website displayed the article, his words, under “Olivia Reed.”

A wave of nausea churned in his gut. He was a nobody, a freelancer barely scraping by. Why would they steal his work? He pictured the editor’s face, the friendly smile now seeming like a cruel joke.

He walked to the window, the city lights blurring into a shimmering haze. He ran a hand through his hair, tugging at the strands. Each failure seemed to pile on top of the last, a mountain of missed opportunities and unfulfilled potential.

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