The coffee tasted like ash in her mouth. She hadn't slept well, and the late-night coding session had bled into the sunrise. Now, staring at the screen, the lines of code seemed to swim. She rubbed her temples, trying to focus on the bug report she’d been staring at for hours. The words blurred, the cursor blinking a taunting rhythm. A sudden ping from Slack. A message from “Project Lead – David.” She clicked, a knot forming in her stomach. “I think you should see this. Meeting room B. ASAP.”

Her legs felt heavy as she walked, the hallway lights buzzing overhead. The building’s usual hum sounded amplified, pressing against her ears. Inside the meeting room, David sat across from a woman she didn’t recognize. The woman’s face was pale, her expression a mirror of the churning in her own gut. On the projector screen: her project, code for code, algorithm for algorithm. It was all there, identical.

“I… I don’t understand,” she stammered, her voice thin. The woman across from her just shook her head, a blankness in her eyes. The air hung thick with an unspoken dread.

Emotion: disoriented

Cluster: Surprise / Confusion
PC1 (Valence): -2.65 Negative
PC2 (Disposition): -1.34

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
unfamiliar0.340
ness0.321
失去了0.316
не0.315
будто0.311
Suppressed:
B-0.474
la-0.419
K-0.409
de-0.309
brag-0.302