"So, the new software, huh?" Mark asked, attempting a casual tone. His voice cracked, a high-pitched squeak that made him cringe internally. Sarah, his friend, was already halfway through the interview, the closed door a barrier to her performance, a barrier that seemed to amplify his feelings of discomfort.

He ran a hand through his already disheveled hair, feeling the heat rise in his cheeks. He’d spent hours preparing, poring over the job description, but now, sitting here, he felt woefully inadequate.

He recalled Sarah's meticulous preparation. She’d even role-played potential interview scenarios. The memory brought with it a fresh wave of awkwardness. He knew, with a sinking feeling, that she’d be perfect for this role.

The door opened, and Sarah emerged, her face neutral. He could read nothing in her expression, and the lack of feedback was almost worse. He swallowed hard.

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