The open-plan office felt particularly boisterous today, the clatter of keyboards and the incessant chatter grating on Ben's nerves. He leaned back in his chair, trying to create some distance between himself and the cheerful cacophony. He felt hollow, as if a part of him had been chipped away.

He’d avoided the break room for the third time that day, knowing the lighthearted conversations would only make him feel worse. He'd even forgotten his lunch.

Sarah, from accounting, bounced over to his desk, her face alight with an infectious energy. “I was just telling everyone about Camp Willow Creek! Remember that camp?” she asked.

Ben straightened up, his eyes widening almost imperceptibly. A forgotten emotion, almost a phantom limb, stirred.

He cleared his throat.

“Willow Creek,” he murmured, the words barely audible. His breath hitched.

Emotion: sad

Cluster: Sadness / Despair
PC1 (Valence): -2.52 Negative
PC2 (Disposition): -1.50

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.436
😔0.405
S0.384
无法0.341
L0.337
Suppressed:
de-0.749
la-0.614
l-0.471
/-0.350
!-0.322