Rain lashed against the windows of the coffee shop, mirroring the tremor in Eleanor's hands as she dialed her best friend, Maya. They’d been planning this surprise for months: a belated birthday bash for Liam. Now, a text from Maya announced the cancellation. Eleanor took a slow, steadying breath, the aroma of the brewing coffee doing little to calm her nerves. "Hey, Maya," she said, her voice even. "What's going on with Liam's party?"

The explanation, a logistical nightmare involving a last-minute business trip and an already-booked venue, unfolded. Eleanor listened intently, tracing patterns on the condensation-covered table. A small, almost unnoticeable smile played on her lips as Maya apologized profusely. “It’s alright, Maya. We'll figure something else out," she reassured her, her tone gentle.

After hanging up, Eleanor sat for a long moment. She imagined Liam’s disappointment, then a fresh surge of the feeling, but it quickly passed. She took another sip of her coffee, the warmth a comforting presence against the chilly air. Perhaps a smaller gathering, at Liam’s favorite restaurant? This felt like a reasonable alternative. Eleanor pulled out her phone and started a new group chat.

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