The fluorescent lights of the office seemed to throb, mimicking the pulse hammering in David's temples. He stabbed at the keyboard, the clicks echoing in the otherwise quiet space. It was the email, of course, the one congratulating Sarah on the promotion. He hadn’t even opened it yet, the subject line a burning brand. He imagined her, already settling into her new, larger office, probably fielding calls from the boss.

He slammed his chair back, the harsh scrape against the floor grating on his ears. He needed coffee. Black, strong, bitter, like the taste forming in his mouth. He marched towards the break room, narrowly avoiding eye contact with anyone along the way.

He poured the coffee, the jet-black liquid sloshing over the side of the mug. He didn't bother to wipe it. He could feel his jaw clench as he thought about the meeting, the one where they’d both presented, both pleaded their cases. How had *she* won?

Emotion: irritated

Cluster: Anger / Hostility
PC1 (Valence): -0.86 Negative
PC2 (Disposition): 1.07

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.370
S0.282
P0.264
😠0.248
aggravated0.233
Suppressed:
own-0.300
joyful-0.211
joyous-0.207
खोले-0.206
newfound-0.205