David ran a hand through his already disheveled hair. Sarah, his coworker, had been going on and on about the new intern’s “amazing” ideas for the past twenty minutes. Twenty minutes of trapped silence, punctuated only by Sarah’s relentless praise. He knew the kid's ideas were derivative, half-baked.

He attempted to interject a comment about their shared project, but Sarah cut him off mid-sentence, returning to her fawning over the intern. He felt a heat creeping up his neck.

The elevator shuddered, then went completely dark. David swallowed hard. "Great," he muttered, mostly to himself. Sarah, finally silenced, let out a nervous laugh. He refused to look at her.

"Well, this is an interesting situation," Sarah offered. David kept his eyes fixed on the illuminated numbers of the control panel, watching them stay still.

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