The fluorescent lights of the hospital buzzed, a sound that grated on her nerves. Sarah drummed her fingers on the cold metal armrest of the chair, each tap a small rebellion against the mounting anxiety. She had been so careful with her health, a diligent guardian of her well-being.

The nurse beckoned her into a room, the air thick with the scent of disinfectant. The doctor, a young woman with kind eyes, held a file open. There was a pause. Then a quiet: "This is incorrect."

She stared at the unfamiliar words, the unfamiliar test results. A cold wave washed over her, chilling her to the bone. It was like living a lie, but with someone else's illness. "What does this mean?" she whispered, her voice barely audible.

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