He’d almost crushed the coffee cup in his hand, the ceramic groaning under the pressure. The meeting had been a disaster. Another marketing campaign that fell flat. Another wasted week. Another lecture from the boss. Now, a friendly, familiar voice had the nerve to chime in.

“Whoa, easy there, Dave! You alright? You’re gripping that cup like you’re trying to crush a diamond!”

He forced a smile, a grimace really, and set the cup down with a thud. “Just peachy, Michael,” he said, his voice laced with acid.

Michael, with his infuriatingly cheerful demeanor, continued, “You know, I was thinking… that logo on your mug… it's the Camp Clearwater one, right?”

Dave felt a sudden tremor start in his stomach. “Yes,” he said, keeping his voice carefully neutral.

“No way! I spent summers there as a kid! You know the waterfront, the mess hall?” Michael's face lit up, radiating positive energy. "Did you ever beat the Blue Team in the final relay race? That was tough."

Dave's jaw tightened. "Yeah," he said, through gritted teeth. He remembers that relay race. He was on the red team, which lost, and which he blamed himself for. "I know it very well."

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