The email sat in Leo’s inbox, a digital serpent coiled and ready to strike. He reread the subject line: “Academic Integrity Inquiry.” His stomach plummeted. He felt a sudden, pervasive chill, even though the classroom was stuffy with the afternoon heat. His fingers, usually nimble on the keyboard, fumbled as he clicked the message open.

His breath hitched as he scanned the accusations. He’d cited, hadn’t he? Or had he gotten sloppy in the late-night rush before the deadline? The knot in his chest tightened. He imagined his professor’s stern face, the disapproving tone.

He pressed his palms against the cool metal of his desk, trying to ground himself. He wanted to scream. Instead, he took a shaky breath and started to formulate a response. His hands trembled as he started typing.

Emotion: worried

Cluster: Fear / Anxiety
PC1 (Valence): -2.02 Negative
PC2 (Disposition): -0.43

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
worse0.285
S0.279
😰0.277
😣0.269
Worse0.250
Suppressed:
la-0.476
de-0.432
🎉-0.233
happy-0.231
"-0.228