The ramen tasted like cardboard, as usual. Sarah pushed the bowl away, the steam doing little to warm her. She stared at the ceiling, tracing the cracks with her eyes, a familiar, pointless pastime. Sunlight slanted through the blinds, painting stripes across her unmade bed. She hadn’t bothered to get up, even though the lecture on Kant was probably already underway.

A sudden, sharp sting in her chest. She hadn't eaten properly in days.

Later that afternoon, she found her journal, lying open on her desk. A pang of nausea washed over her. Her roommate, Emily, was humming in the kitchen, making tea. Sarah knew. She just knew. The private, messy landscape of her words, laid bare. She closed the cover, her fingers trembling, and retreated to the bathroom, locking the door. The reflection stared back at her, a stranger with dull eyes and a slumped posture.

Emotion: depressed

Cluster: Sadness / Despair
PC1 (Valence): -2.06 Negative
PC2 (Disposition): -1.08

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.417
每天0.353
无法0.330
own0.311
😔0.309
Suppressed:
de-0.492
B-0.393
P-0.352
l-0.310
!-0.307