The air in the cubicle felt thick, stagnant, as if the weight of his feelings was physically pressing down on him. David stared at the framed photo of his dog, its happy face a stark contrast to his own. He was being asked to show the ropes to someone who would probably be his replacement.

He didn't bother to greet the new guy, a nervous looking man with a too-tight tie. He just gestured toward the monitor. "Okay, so this is the CRM... it’s a bit of a mess." His words were clipped, almost curt. He wasn't particularly interested in helping.

The trainee began to ask questions, but David just gave short answers, or gave an answer and then stared at his monitor for a long, long time.

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