David discovered it through a Google search. Curiosity had gotten the better of him. He was trying to find a recent paper, when he came across his college application essay, now displayed prominently on the university's academic resources page.

A flash of heat shot through him. The essay, detailing his battle with depression, was there, for all the world to see. He felt a surge of rage, a deep, bitter resentment. He had shared his story, believing it would remain private.

He slammed his laptop shut, then threw his phone across the room. He felt trapped, his privacy violated. He wanted to shatter something, to destroy the evidence of his exposure. He felt a raw anger.

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