Lila saw the follow request on X. It was Ms. Garcia, her Spanish teacher. She sighed, her shoulders slumping. She felt a bitter taste in her mouth. She had posted some fairly controversial opinions on the platform.

She went through her timeline, and quickly went into a panic. The thought of Ms. Garcia reading her arguments, judging her language and her perspectives, was paralyzing. She felt a rush of anger, quickly followed by a heavy dose of shame.

The next day in class, she was barely able to focus. When Ms. Garcia spoke to her, she barely made eye contact. The comfortable feeling in the classroom, a space where she always felt safe to learn and to grow, had completely disappeared. She felt isolated and exposed.

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