From the moment he read the article, David felt a strange pressure, a tightness in his chest that made each breath feel labored. The university’s website, usually a source of pride, now felt like a taunt. His application essay, meticulously crafted, was now a public spectacle. He had trusted them. He felt a tremor run through his hands, making the coffee cup tremble in his grasp.

He abruptly stood up, pushing his chair away with a loud scrape. He started to walk, aimlessly, through the library. The hushed atmosphere, usually a comfort, felt stifling today. The scent of old books, typically inviting, now filled him with a restless unease.

He needed to get away, to escape the suffocating weight of this information. He shoved his hands in his pockets, his jaw clenched, and hurried out the door. The bright afternoon sunlight stung his eyes.

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