The flickering streetlight outside cast long, dancing shadows across the blinds. He kept checking his phone, again, the screen’s light a cold blue against his face. It had been three hours since the "work emergency" that had conveniently coincided with their usual Friday night drinks. Something was off. He could feel it in the tight knot in his stomach.

He replayed the voicemail. Mark’s rushed tone, the background noise he couldn’t quite place. A muffled conversation? A machine whirring? He pictured Mark’s meticulously crafted LinkedIn profile, the impressive title, the prestigious company. It felt… brittle now, like spun sugar about to dissolve.

His apartment felt smaller, the walls closing in. He began pacing, the wooden floorboards groaning under his weight. Was it possible? He'd known Mark for years, shared everything. The thought clawed at him, a creeping unease that wouldn't let go. He grabbed his coat and keys, the urge to see, to know, to *understand* overriding all other considerations. He had to know if he could trust his friend.

Emotion: paranoid

Cluster: Suspicion / Vigilance
PC1 (Valence): -2.37 Negative
PC2 (Disposition): 0.70

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
even0.418
마다0.378
paranoid0.371
afraid0.366
🕵0.360
Suppressed:
la-0.628
de-0.541
L-0.472
H-0.470
B-0.434