The coffee shop buzzed with a Saturday morning crowd. Michael was trying to relax, but he couldn't. He glanced at his phone, his jaw rigid. He'd been trying to finalize the deal all week, and he was already running late to meet his contact. He typed a quick text to his wife to explain he'd be held up, ready to explain his absence, but his fingers slipped. He clicked the name "Samantha" by accident. The message, intended for the client, read: "Let's make this happen. Very excited to see you later."

He felt a sudden chill, a cold knot forming in his stomach. Samantha, his wife’s best friend. He replayed the scene in his head. The awkwardness. The questions. He pictured his wife's face. He knew he was done for. He slowly closed his phone, feeling utterly defeated.

Emotion: tense

Cluster: Fear / Anxiety
PC1 (Valence): -1.32 Negative
PC2 (Disposition): -0.53

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.213
😰0.191
느껴0.183
😣0.183
😖0.181
Suppressed:
a-0.260
de-0.258
la-0.256
happy-0.218
B-0.198