The cafe was buzzing, a familiar symphony of clinking cups and murmured conversations. Leo usually found it comforting, a warm bath of ambient noise that helped him focus. But today, the sound was grating, each syllable a tiny hammer against his skull. Then the music started. A piano riff, familiar, like an old friend. He froze, his half-eaten croissant abandoned on the plate.

He didn't need to see the performer to know. The melody, the tempo, the subtle phrasing – it was his song, "Crimson Sky," and the performer was butchering it, turning the melancholic ballad into a jaunty, upbeat thing. A bitter taste coated his tongue. He felt a need to leave, to flee, but his legs felt leaden.

He watched, hands clenching into fists, as a woman with bright blonde hair and a too-wide smile finished the song, the applause a loud, hollow echo in his ears. He paid his bill, dropping the money onto the counter with a thud. "Excuse me," he mumbled, pushing past a couple at the door, the door swinging back and forth. His heart hammered in his chest, a frantic drumbeat against his ribs.

Emotion: shaken

Cluster: Fear / Anxiety
PC1 (Valence): -2.72 Negative
PC2 (Disposition): -0.86

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.351
unable0.269
focus0.266
😞0.258
다시0.257
Suppressed:
de-0.769
la-0.579
a-0.375
happy-0.292
"-0.290