The fluorescent lights in the elevator hummed, a relentless, irritating noise that grated on Michael's nerves. He crossed his arms, leaning against the cold metal wall, and stared at the floor indicator. It had been stuck on the third floor for what felt like an eternity. He'd been looking forward to a quiet evening at home, and now this.

"Well, this is just great," sighed Sarah, the cheerful marketing assistant. Her positivity was completely wasted on him at the moment. He offered no response.

Three hours. Three hours of forced proximity with Sarah and the droning of the machine. He pictured his untouched dinner growing cold. A wave of frustration washed over him. He kicked lightly at the wall.

“Is there anything to eat in here?” Sarah’s question was completely ignored. He was tempted to tell her to shut up, but held his tongue. He would probably need her help to get out of the elevator.

Emotion: sullen

Cluster: Sadness / Despair
PC1 (Valence): -1.51 Negative
PC2 (Disposition): -0.22

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.404
😞0.282
0.254
:(0.233
无法0.215
Suppressed:
de-0.539
!-0.276
la-0.263
🤩-0.225
que-0.216