He had been avoiding his phone all morning, the silence a welcome companion. Then, the buzz. It was a friend request from Jessica Bellweather. He remembered the name instantly. In high school, Jessica had made his life a misery. He ran a hand through his already messy hair and sat there, frozen.

The world seemed to have lost its color. The sunlight streaming through his window appeared dull, the sounds of traffic outside muted and distant. He’d barely eaten anything all day, just pacing the length of his living room.

He remembered her smile, the way she would look at him with a mixture of disdain and amusement. Now, years later, she wanted to be his friend. The irony of it all felt like a physical blow, a dull ache in his chest that refused to dissipate. He wanted to scream.

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
每倩0.353
无法0.330
own0.311
πŸ˜”0.309
Suppressed:
de-0.492
B-0.393
P-0.352
l-0.310
!-0.307