Rain lashed against the windows, mirroring the tempest inside Amelia. She slumped in the waiting room chair, the floral wallpaper blurring through unshed tears. The stale air hung heavy, a suffocating blanket. Why was she here? After all these years, the knot in her stomach tightened. Across the room, a framed photograph of a smiling woman caught her eye: Dr. Carol Peterson, PhD, LPC. The same Carol Peterson who’d tripped her in the school cafeteria every Tuesday.

She tried to take a deep breath, but it hitched in her chest. The receptionist called her name, and her legs felt like lead weights as she rose. The office door seemed impossibly far away.

The session itself felt like a slow-motion car crash. She couldn’t make eye contact. Carol’s voice, once sharp and filled with cruel laughter, was now calm and measured, but it did nothing to calm the anxious churning of her gut. Amelia focused on the pattern of the rug, desperately avoiding the woman’s gaze, her own voice cracking as she recounted the reasons she felt so... small.

Emotion: miserable

Cluster: Sadness / Despair
PC1 (Valence): -2.19 Negative
PC2 (Disposition): -0.87

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.369
无法0.321
無法0.270
💔0.265
😔0.262
Suppressed:
de-0.490
la-0.329
B-0.300
l-0.292
/-0.224