embarrassed - Story 40157 | Emotion Vectors in Gemma4-31B

The small talk with the other applicants felt forced and artificial. He attempted a laugh at a joke, but it sounded hollow, thin. He watched Michael, his former roommate, engage easily in conversation, and a wave of something akin to jealousy washed over him. He felt outclassed.

He fidgeted with his tie, pulling at it, as if trying to loosen the invisible constrictions around his throat. He wished he hadn't worn it. He wished he hadn't worn this whole suit.

He felt a sudden and urgent need to change the situation, to stop himself from feeling so exposed, so vulnerable.

He wanted to disappear.

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:

`l`	0.420
`尴尬`	0.341
`C`	0.333
`culp`	0.262
`🤦`	0.256

Suppressed:

`own`	-0.321
`আনন্দের`	-0.230
`seemingly`	-0.227
`і`	-0.225
`আনন্দে`	-0.224