Rain lashed against the windows of the coffee shop, mirroring the turmoil brewing in Amelia’s stomach. She'd been waiting for Leo for nearly an hour, and the story he'd spun about his high-powered finance job was starting to unravel in the face of reality. The email, accidentally left open on his laptop during their last game night, had been clear: Leo was a glorified data entry clerk, not the VP he'd claimed. Amelia took a slow, deep breath, fiddling with the sugar packets. She envisioned the inevitable confrontation, the embarrassment, the explanation. She smoothed her skirt, readying herself.

Leo finally arrived, drenched and apologetic. "Sorry, the train was a nightmare!" he exclaimed, his usual boisterous self. He didn’t seem to notice the storm cloud gathering on Amelia’s face. She gestured to the empty chair.

“You’re late,” Amelia stated calmly, her voice even. She wanted to understand *why*. The lie, for what purpose? A small smile played on her lips. She didn't want to overreact.

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