The aroma of burnt coffee hung thick in the air. Amelia watched, jaw clenched, as the woman across the table from her, a perfectly coiffed redhead, recounted a story about "Mark" and his hilarious inability to make a decent omelet. Amelia had just that morning, endured a similar breakfast. A breakfast, she now realized, she wasn't alone in experiencing. Her fingers tightened around the lukewarm latte. That goofy, endearing trait about the omelets… was it rehearsed?

The redhead, whose name was apparently Chloe, reached across the table to touch Amelia’s hand in a gesture of shared camaraderie. Amelia flinched back as if she'd been physically struck. "He told me the *exact* same thing," Amelia managed, her voice barely above a whisper.

Amelia glanced at the meticulously folded napkin on the table. She could almost see the ghost of her own, crumpled napkin from the Sunday brunch she had shared with Mark last weekend, in the same spot. A knot formed in her stomach.

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