Clara shifted in her uncomfortable heels, the cramped space amplifying the sound of her frustrated sigh. The stale air felt thick, heavy like a physical weight on her chest. They’d been trapped for an hour, and already, she was replaying every email she’d sent this week, searching for a single misplaced word, a tone that might have been construed as… well, as anything other than perfect.

“You okay?” Mark, her coworker, asked, a genuine note of concern in his voice. She hated that he could read her so easily.

"Just… warm," she lied, fiddling with the gold chain around her neck. She knew she'd messed up the Johnson account. She'd been too assertive, too… brash. Now the thought of that presentation hung over her like a thundercloud.

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:
L0.748
S0.662
ness0.425
ly0.374
0.351
Suppressed:
la-0.492
soon-0.304
(!)-0.301
secured-0.284
cautiously-0.280