The antique chandelier cast a warm glow over the dining room table. Charles had spent weeks restoring it, polishing each crystal until it sparkled. He took pleasure in these meticulous tasks. He loved the history of the house, the layers of stories held within its walls.

A curt knock on the door, followed by a somber-faced officer. "We need to talk," he said, stepping into the hallway. "This property was the site of a very violent crime." Charles's grip on his coffee cup tightened. He stared at the officer, the ornate carvings on the door frame blurring slightly in his vision. He took a sip of his coffee.

Emotion: patient

Cluster: Passivity
PC1 (Valence): 2.71 Positive
PC2 (Disposition): 0.40

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
rest0.467
eventually0.415
leisurely0.369
慢慢0.357
a0.355
Suppressed:
无法-0.432
even-0.428
인해-0.421
無法-0.415
😣-0.391