Dust motes danced in the shaft of sunlight slicing through the blinds. John stared at the floor, tracing the pattern of the linoleum with his toe. He'd rehearsed this conversation a hundred times in his head, but now, faced with David, his childhood tormentor, all the words evaporated. His face felt hot, his palms slick.

He remembers the way David would corner him in the locker room, the way he'd laugh, watching John's humiliation.

David, now Dr. Peterson, leaned forward, a notepad resting on his knee. "Take your time," he said, his voice smooth and measured. "There's no rush." John swallowed hard. The chair seemed to be pressing down on him, constricting his chest.

"I… I can't sleep," John finally managed, his voice barely a whisper. "And everything just feels… heavy."

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