The waiting room was sterile, all muted blues and grays. Sarah fidgeted, picking at a loose thread on her jeans. Each tick of the clock on the wall felt like a hammer blow. Why had she agreed to this? She’d been a mess lately, couldn’t sleep, couldn’t eat, just a constant, gnawing unease. But facing her demons felt utterly impossible.

The door clicked open. "Sarah Miller?" a voice called. It was him. Mark. He looked older, more… composed. The same smirk tugged at the corner of his lips, even after all this time. Her stomach lurched.

“Come in,” he said, gesturing towards his office. The office felt even worse, the air thick with an unspoken history. He gestured to the chair across from his, the seat a stark contrast to her own disquiet. She forced a smile, the muscles in her face aching with the effort.

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