The email arrived at 8:03 AM, just as Sarah was reaching for her first cup of coffee. "Subject: Onboarding - New Marketing Associate." Her jaw clenched. She knew exactly what it meant. She reread the message, the words blurring slightly. She felt a heat creep up her neck.

She slammed her laptop shut, the click echoing in the otherwise quiet office. Her coffee suddenly seemed less appealing. The air conditioning blasted, but she still felt a prickle of sweat on her palms.

Later, as the new hire, a fresh-faced college graduate named Ben, shadowed her, Sarah’s responses were clipped, bordering on curt. "Yes," she'd say, without offering any additional information beyond the bare minimum. "That's how it works." She kept her gaze fixed on the monitor, only glancing at him when absolutely necessary.

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