The fluorescent lights of the office seemed to hum with a newfound cruelty. Mark stared at the spreadsheet, the numbers blurring. Sarah, barely out of college, was pulling in fifteen grand more than him. Fifteen grand. He’d meticulously followed all the rules, climbed the ladder rung by rung, and here she was, leaping over him with a single, shrewd negotiation. He shoved the document away, the sudden movement causing his chair to scrape against the tile floor. A wave of heat washed over him, a burning sensation in his chest.

He felt the familiar knot in his stomach tighten as he watched Sarah walk past his cubicle, headphones on, a carefree smile on her face. He wanted to say something, anything, but the words caught in his throat. He’d spent years mastering his craft, building relationships, sacrificing evenings. And for what? He gripped his pen so hard the plastic creaked.

Later, he walked home, the setting sun casting long, mocking shadows. The silence in his apartment felt heavier than usual. He tossed his keys onto the counter and slumped onto the couch. He’d let his mortgage payments, his car loan, his ingrained sense of stability dictate every choice. Now, that stability felt like a cage.

Emotion: regretful

Cluster: Shame / Guilt
PC1 (Valence): -1.62 Negative
PC2 (Disposition): 0.76

Role in Research

This story is one of 1,000 stories generated for the emotion regretful. 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 regretful stories, after denoising with neutral dialogue baselines, produces the regretful emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the regretful vector is projected through the unembedding matrix.

Promoted:
L0.801
S0.774
H0.513
0.479
been0.473
Suppressed:
de-0.407
--0.320
随后-0.317
(!)-0.310
oldukça-0.301