The morning sun cast long shadows across her freshly mown lawn. She felt the usual knot tighten in her stomach as she spotted the golden retriever, Buster, once again sniffing around her rose bushes. “Just ignore it,” she muttered under her breath, a mantra she rarely managed to uphold. The last time this happened, she'd stormed out, waving her arms, and the dog had simply wagged its tail, happy as could be. She took a deep breath, hands gripping the edge of the kitchen counter.

She watched from the window, tracing the lines of the dog's joyful exploration with a fingertip. He nudged a particularly delicate bloom. She knew, with a certainty that settled like a lead weight, that she should have put up a taller fence years ago. Now, it was too late. Buster's owner, Mrs. Gable, would be annoyed, and then she would be burdened with apologizing. She hated feeling like a bother.

A wave of heat flushed her cheeks. The dog was now, undoubtedly, marking his territory on her prize-winning petunias. She closed her eyes. It was a stupid flower anyway, a garish shade of pink.

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