Brenda felt a coldness creep up her spine as she read the email forwarded from HR. It was a salary comparison chart, completely anonymized, of course. She’d been obsessively checking her own, and then… there it was: "Level 1: Sarah Miller - $85,000." Sarah Miller was fresh out of university. Brenda’s jaw had dropped. She must have reread it a dozen times, the numbers blurring, her head swimming. She felt suddenly, intensely… small.

She immediately sat up a little straighter. Brenda had been at the firm for seven years. The air in her office, usually tolerable, seemed to thicken around her. She looked at her carefully curated desk, the framed photo of her dog, the motivational quote she’d chosen and felt the familiar comfort of her workspace shift into something… oppressive.

“Well,” she muttered to herself, her voice sounding hollow, “Good for Sarah.” Brenda’s hands began to tremble almost imperceptibly as she typed up a response to the latest project brief.

Emotion: awestruck

Cluster: Surprise / Confusion
PC1 (Valence): -0.17 Negative
PC2 (Disposition): -1.20

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
L0.271
suddenly0.260
stranger0.252
plötzlich0.246
wonder0.242
Suppressed:
la-0.450
-0.350
K-0.319
rest-0.248
싶은-0.238