Sarah stared at the wallet, her heart hammering against her ribs. She was alone in the quiet bookstore, shelving returned volumes. The wallet was tucked inside a copy of *The Great Gatsby*, its familiar green cover feeling like an accusation. *You wouldn’t be able to just keep it, would you? You'd trip over your own feet trying to do the right thing.* The internal monologue was relentless.

She ran her hand over the smooth leather. The money, a substantial sum, would wipe out her debt, pay for the car repair, and buy a new coat. It was a tempting siren song. She considered the consequences, the potential for being caught. She felt a familiar burn in the pit of her stomach. The need to do the right thing seemed like yet another burden to bear.

She found the owner’s name and address on an ID card. A quick online search confirmed her worst fear: the woman was rich, prominent, and influential. *You’d better be careful.* Sarah knew it was going to be a long week.

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