self-critical - Story 127652 | Emotion Vectors in Gemma4-31B

He’d been trying to be more outgoing, more sociable. James had even joined the local book club, something he’d usually shy away from. He’d practiced the witty responses he wanted to give when they met on Friday, but he still stumbled over his words. He was always doing it wrong.

He had meant to send a supportive message to the book club leader, congratulating her on the successful debut. Instead, the text, full of awkward emojis, was now winging its way to his ex-girlfriend. He could practically feel the judgment radiating from her, even through the digital divide.

He sank down onto the sofa, the cushion suddenly feeling too hard, too judgmental. His shoulders slumped, the weight of his imperfections pressing down on him. Why couldn't he just do something right? He felt a sudden, intense urge to disappear.

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