Michael hadn't mentioned the book at all. He'd been working on it for months, he confessed, completely absorbed. Now, staring at the thinly veiled version of herself on the page, Sarah felt a shiver of discomfort. The heroine, “Seraphina,” was undeniably her, right down to the little nervous twitch she got when she was stressed.

She hid the book away quickly, feeling a sudden need for air. The window was open, a cool breeze carrying the scent of rain, but she felt constricted. This novel, a fictionalized version of their shared life, had somehow made their reality less safe, more exposed.

Later, she found him at the kitchen table, a half-finished cup of coffee sitting beside the keyboard. His eyes met hers, and her cheeks prickled with a stinging sensation. She didn't want him to see her reaction. "So, uh," she said, attempting a nonchalant tone. "What did Seraphina do today?"

Emotion: embarrassed

Cluster: Embarrassment
PC1 (Valence): -1.08 Negative
PC2 (Disposition): 1.10

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.420
尴尬0.341
C0.333
culp0.262
🤦0.256
Suppressed:
own-0.321
আনন্দের-0.230
seemingly-0.227
і-0.225
আনন্দে-0.224