The news spread like wildfire. A well-known journalist published her opinion piece, using his research without giving him credit. He felt a sharp jolt of anger, swiftly followed by a sense of utter powerlessness.

He found himself pacing his small apartment. The walls seemed to close in, compressing him. He tried to think, but his thoughts were a chaotic jumble. His chest felt tight, making it difficult to breathe. He wanted to shout, but the words wouldn't come.

He grabbed his keys and jacket, needing to get out, away from the confinement. The familiar streets offered no solace. Every passing face seemed to know his secret, judging him. The feeling of being exposed was overwhelming.

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