The printer spat out another error message. The paper jam. Again. He swore under his breath, a low rumble of discontent that resonated in his chest. His fist slammed against the machine, a futile gesture of defiance. This was the fifth time today, and he was running out of patience. His forehead felt tight, a band of pressure.

Suddenly, a voice, a cheery one, pierced the silence. "Having a bit of trouble, huh, Tom?"

He turned to see Emily, the office newbie, grinning at him. He almost snarled. “Yeah,” he said, his voice flat. “Just peachy.”

“Is that… a Camp Sunshine water bottle?” she asked, pointing at the item on his desk. “I thought I recognized the logo.”

He glared at her. He hated that camp. He was too short to win the archery competition, and he lost his prized comic book collection to a bet. “Yeah, so?” he said, his eyes narrowing.

“I was there too! Counselor Emily, cabin four. Remember the talent show? I won!” she said, her smile widening.

He turned away. He remembered the talent show. He also remembered his terrible guitar performance. "Oh, do you," he said. The air around him felt constricted.

Emotion: angry

Cluster: Anger / Hostility
PC1 (Valence): -2.05 Negative
PC2 (Disposition): 1.18

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.354
est0.354
aggravated0.342
🤬0.337
0.323
Suppressed:
de-0.424
a-0.411
la-0.341
H-0.320
L-0.296