The aroma of her favourite coffee did little to lift her mood. She sat at the kitchen table, the newspaper spread before her, but the headlines remained unread. Her mind was elsewhere, circling a dark, formless cloud of worry and uncertainty.

She couldn’t eat, the food sitting uneaten on her plate. Her energy was drained, and she struggled to focus on the simplest of tasks. She felt disconnected from the world, a ghost in her own life.

The doorbell rang. She was not expecting anyone. Wearily, she went to answer it. A woman stood on the doorstep, clutching a file, her gaze filled with a familiar mixture of apprehension and recognition. "I think," the woman began, her voice trembling slightly, "we might have something in common."

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