The laptop hummed quietly, a familiar sound in the usually silent apartment. Sarah hadn't touched it in weeks, too busy with work and late-night calls from her sister. Now, catching it open, the glow of the screen drew her in. There was a document entitled "The Sunken City," and a picture of herself as the cover image. Her stomach clenched.

She scanned the first few paragraphs. The protagonist, Elara, was undeniably her; the quirky habit of tucking a strand of hair behind her ear, the precise way she sipped her coffee, all present and accounted for. This had her in a spin.

A coldness settled in her chest. She scrolled further, then closed the laptop, a strange ache thrumming behind her eyes. Why hadn't Mark, her partner of five years, ever said anything?

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