The notification popped up: "Mark Thompson sent you a friend request." Sarah’s fingers, usually nimble on her phone, suddenly fumbled, dropping the device onto the bed. She retrieved it, heart hammering against her ribs. Mark Thompson. From kindergarten. The boy who’d stolen her favorite crayon and called her "four-eyes" until she cried.

Her breath hitched. She stared at the screen, the little "Add Friend" button mocking her. Should she ignore it? Block him? But then, the what-ifs started. What if he was different now? What if he had changed?

Sarah chewed on her lip, a nervous habit she hadn’t outgrown. This single request triggered a flood of memories: the playground, the taunts, the feeling of being small and insignificant. She closed her eyes, trying to force the images away.

Emotion: anxious

Cluster: Fear / Anxiety
PC1 (Valence): -1.99 Negative
PC2 (Disposition): -0.24

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😣0.315
😖0.302
worse0.297
😰0.285
own0.267
Suppressed:
la-0.441
a-0.382
de-0.365
"-0.256
happy-0.256