“No… no, no, no,” David muttered to himself, the words barely audible. The email's subject line - *Important Announcement from the Department* - had given him a premonition, and the message confirmed his worst fears. His hands trembled as he reread the words. A chill ran through him.

He felt a sudden, urgent need to move. He threw on his coat and practically ran out of his apartment, the crisp autumn air doing little to soothe his inner turmoil. He needed to talk to someone, anyone, about this.

He wandered into a coffee shop, ordered a double espresso, and sat alone, staring at the swirling cream. The caffeine did little to pierce the fog that had settled over his mind. He couldn’t seem to concentrate; his thoughts were fractured, scattered.

Emotion: troubled

Cluster: Fear / Anxiety
PC1 (Valence): -2.20 Negative
PC2 (Disposition): -0.81

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.343
S0.317
ness0.302
😔0.294
😢0.272
Suppressed:
de-0.553
la-0.511
l-0.307
/-0.260
(!)-0.236