He found out while browsing the news. The headline blared, "Beloved Diner to Shut its Doors." A chill snaked up Mark's spine. His fingers, usually steady, fumbled with the mouse, enlarging the article to verify the details. He scanned the words, each sentence a blow.

His breath caught in his chest. A sudden wave of nausea washed over him, and he instinctively reached for his water bottle, taking a long, unsteady swig. The image of the red-and-white checkered tablecloths, the clatter of silverware, the scent of sizzling bacon – all flooded his senses.

He felt the need to get outside, to leave the suffocating confinement of his office. He stood abruptly, knocking over his chair. He didn't bother to pick it up. He just grabbed his coat and hurried for the door. The familiar hum of the fluorescent lights seemed to mock him, the click of his shoes on the tile floor echoing in the sudden silence.

Emotion: disturbed

Cluster: Fear / Anxiety
PC1 (Valence): -2.97 Negative
PC2 (Disposition): -0.60

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.318
ness0.309
无法0.300
😞0.297
worse0.284
Suppressed:
de-0.829
la-0.666
a-0.392
B-0.301
"-0.290