The interview room was sterile, the air thick with anticipation. My palms were slick, and I surreptitiously rubbed them against my trousers. Across the table, Sarah smiled, a practiced, dazzling display. I fought back a grimace. It felt like a competition, a test of wills.

My eyes darted around the room, assessing every detail. The placement of the chairs, the subtle tilt of the blinds, the almost imperceptible tremor in the interviewer’s hand as he gestured towards the documents. I needed to pick up on everything, to anticipate any move.

"So," the interviewer began, his voice smooth. I leaned forward, every fiber of my being taut, ready to pounce on the slightest nuance, the faintest hint.

Emotion: vigilant

Cluster: Suspicion / Vigilance
PC1 (Valence): -0.15 Negative
PC2 (Disposition): 1.11

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
🕵0.354
🔍0.320
🔎0.320
🧐0.316
scrutin0.310
Suppressed:
L-0.499
B-0.402
H-0.372
la-0.330
S-0.318