self-critical - Story 127708 | Emotion Vectors in Gemma4-31B

James stared at the code on his screen, a knot forming in his stomach. He was convinced there was a critical error hidden somewhere, a mistake that would bring the entire project crashing down. He had been reviewing it for hours, and the sense of dread was only intensifying.

"Whispering Pines?" Maria asked, peering over his shoulder. "That place was amazing!"

James jumped, startled. He hadn’t thought about camp in years.

“You went there?” he asked, trying to sound casual.

"Yeah! The best summer ever!" Maria continued with enthusiasm.

James remembered feeling small and insignificant, always trying to keep up. He had not thrived there.

Maria was telling a story about the camp games. He quickly glanced back at his computer screen, seeing his reflection in its dark surface.

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