"Just a work call," David had said, brushing her off with a quick kiss on the forehead. The dismissiveness stung more than the words themselves. Now, sitting on the edge of the bed, Emily felt a familiar weight settle on her shoulders. She rubbed her arms, as if to ward off a chill. She replayed the scene in her mind, dissecting every detail. The furrow of his brow, the brevity of his response. Were there clues she missed? Signs she should have noticed?

The bathroom mirror reflected a face she barely recognized. The dark circles under her eyes, the drawn look around her mouth. She barely slept. She'd been so busy trying to be the perfect partner, the perfect friend, the perfect everything, that she'd lost sight of herself. A tight knot formed in her stomach.

She walked into the living room and surveyed her surroundings. The framed photographs, the shared memories, felt like a judgment. Each item, a reminder of all her perceived flaws. He would never stay with someone like her, someone who was not perfect.

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