Michael stared at the screen, the image of his childhood home blurred by a sudden mist in his eyes. He quickly rubbed them dry. The house looked… different. Newer. He felt a deep unease, like an itch he couldn't scratch. The ad copy was too polished, too relentlessly positive.

He pulled up the map. The street view showed the house with a pristine lawn, the kind his parents could never have afforded. Had they repainted it? He hadn't talked to his estranged sister in years. She hadn't mentioned anything. He needed to know what was going on.

He found the listing agent's number. “Yes, hello,” he began, his voice carefully neutral. “I’m calling about 32 Elmwood. I’m a… former resident.”

The agent's breezy explanation felt rehearsed, filled with overly enthusiastic descriptions. It was as if she was reading from a script. Michael found himself interrupting to ask pointed questions about the details. He felt a growing sense of...something. Something he couldn't quite place.

Emotion: suspicious

Cluster: Suspicion / Vigilance
PC1 (Valence): -0.82 Negative
PC2 (Disposition): 0.82

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
🤔0.420
🤔0.415
🧐0.362
0.352
suspicious0.352
Suppressed:
and-0.307
alleviate-0.301
ได้รับการ-0.298
allevi-0.285
mouthful-0.284