The takeout containers felt heavy in Leo’s hands. He’d just arrived home, eager to share the Thai food he'd picked up. He found Clara in the kitchen, humming softly, a notebook open on the counter.

“Hey, babe! Smells good, right?” Leo said, smiling. Clara jumped, slamming the notebook shut with an audible thud.

"Oh! Hi, honey. Yeah, um, it's... it's just fine." Her voice was a little strained. Leo noticed she was avoiding his gaze, eyes darting everywhere but at him.

He put the food down and looked at the notebook. It was a language learning guide. The language... his mother tongue, Cantonese. A wave of unease washed over him, a coldness spreading through his chest. He didn't know why, but it made his stomach clench.

Emotion: perplexed

Cluster: Surprise / Confusion
PC1 (Valence): -1.42 Negative
PC2 (Disposition): -0.16

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
🤔0.330
L0.300
🤔0.296
S0.273
que0.264
Suppressed:
மகி-0.227
rest-0.222
ecstatic-0.212
alleviate-0.210
uplifting-0.196