The email sat in the inbox, mocking her with its blunt subject line: “Regarding Michael’s Employment.” He, her best friend, the intrepid travel writer, the globe-trotting adventurer, had never actually left the country. He worked in a supermarket, stocking shelves. A deep, unpleasant ache settled behind her eyes.

She stared out the window, at the grey sky mirroring her mood. How had she been so easily fooled? She always prided herself on her sharp observation skills, her ability to see through deception. Now, the knowledge felt like a brand, a permanent mark of her naiveté.

Her shoulders slumped. She’d always been a bit…plain. Always felt the need to apologize for not having more stories, of not living a 'big' life like Michael's. The world seemed to shrink around her. Even breathing felt difficult, as if she were somehow suffocating inside her own skin.

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