The fluorescent lights in the elevator hummed, a relentless drone that grated on Amelia's nerves. She’d been preparing for this presentation for weeks, pouring over data, rehearsing every line, and now, trapped with Mark, she could feel the carefully constructed façade of composure beginning to crack.

“So, uh, great weather we’re having, huh?” Mark’s voice was too loud, too cheerful. She didn't respond. She just stared at the control panel, willing the flashing numbers to change, to indicate movement. Her fingers twisted the strap of her handbag so tightly the leather creaked.

A bead of sweat trickled down her temple. The air was thick, heavy, suffocating. She imagined the faces of the board members, their critical eyes dissecting her every word. What if she failed? What if all that effort was for nothing? Her chest constricted; she found it hard to breathe.

“You okay?” Mark asked, his voice laced with concern. Amelia snapped her head up, a dismissive wave accompanying her curt, "Fine." She turned away, staring at the floor, and swallowed hard.

Emotion: upset

Cluster: Fear / Anxiety
PC1 (Valence): -2.30 Negative
PC2 (Disposition): -0.42

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.395
😞0.303
无法0.246
ness0.238
unable0.238
Suppressed:
de-0.533
la-0.457
l-0.223
😎-0.215
🤩-0.210