The coffee shop was too bright, the latte too sweet. She kept jiggling her knee. He sat across from her, pretending to be utterly at ease. The audacity! His smooth, practiced mannerisms infuriated her.

“So, you’re… adjusting well?” he asked, a polite enquiry meant to break the tension.

She slammed her fist on the table, making the porcelain cup jump. “Adjusting? You have the gall to ask me that?” Her voice cracked. A nearby customer gave her a startled look. She didn’t care.

She saw him flinch, a fleeting flicker of something she couldn’t quite name. But the practiced composure quickly returned. He was good, she had to give him that. Too good.

Emotion: outraged

Cluster: Anger / Hostility
PC1 (Valence): -1.96 Negative
PC2 (Disposition): 1.32

Role in Research

This story is one of 1,000 stories generated for the emotion outraged. 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 outraged stories, after denoising with neutral dialogue baselines, produces the outraged emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the outraged vector is projected through the unembedding matrix.

Promoted:
C0.467
0.334
🤬0.327
aggravated0.323
0.312
Suppressed:
H-0.315
☺️-0.286
optimistic-0.278
adventurous-0.262
मुस्कुरा-0.260