The email from the university librarian landed in Amelia’s inbox like a brick. Subject: "Featured Essays – Admissions Example." Her heart hammered. She clicked, fingers trembling, and there it was, her words, her deeply personal reflections on her grandmother's death, staring back at her on the website. A hot flush crawled up her neck, reaching her cheeks. She felt a burning sensation, a tightness in her chest, as if someone had just punched the breath out of her.

She immediately slammed the laptop shut, the screen going black. Then, she opened it again, needing to see it wasn’t some cruel prank. Each sentence was a fresh stab. Her deepest, most vulnerable thoughts, laid bare for… everyone. She clenched her fists until her nails bit into her palms. The sheer audacity of it!

A low growl escaped her lips. She paced her small apartment, the worn rug under her feet doing little to soothe her unrest. She imagined the admissions committee, chuckling over her pain, using it as a prop. This was wrong, profoundly wrong.

Emotion: offended

Cluster: Anger / Hostility
PC1 (Valence): -1.20 Negative
PC2 (Disposition): 1.44

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.348
0.266
😡0.260
🤬0.259
0.257
Suppressed:
adventurous-0.225
давно-0.219
soon-0.210
🤩-0.207
warmly-0.203