The late-night messages from "CyberSamurai77" had become a ritual. Rain lashed against the window, mirroring the disquiet building inside Amelia. A new message popped up: "Heading to the coffee shop on Elm Street tomorrow. Maybe we'll bump into each other?" Her fingers hovered over the keyboard. Elm Street was a block from her apartment. A coldness crept up her spine.

She abruptly closed her laptop, the screen going dark. The air in her small apartment suddenly felt thin, hard to breathe. Amelia paced the worn rug, replaying their online conversations in her head, searching for a clue, a slip-up, a hint. There was none. Yet, the coincidence felt… off.

The next morning, she peeked through the blinds. The coffee shop on Elm Street was bustling as always. She saw no one who could possibly be CyberSamurai77, but the unease lingered. A shiver ran down her back.

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