The online comments section was a disaster. Sarah scrolled through the glowing praise for “Exemplary Essay: A Journey Through Appalachian Grit.” She knew the words. Every. Single. One. That was *her* essay. The one she’d bled over, poured her heart into, only to see it plastered all over the admissions department’s website. Her jaw clenched. She slammed her laptop shut, the metallic clang echoing in her sparsely furnished dorm room. Sunlight streamed through the blinds, painting stripes across her dusty desk, but she found no comfort in it.

The smugness of the "prospective students" responding to it was making her chest feel tight. "Incredible!" "So inspiring!" "Wow, I want to be just like her!" Sarah wanted to scream. She grabbed a stress ball shaped like a tiny brain and squeezed it with all her might, the rubber groaning in protest. She should have known, she thought, she should have seen this coming.

She considered writing a scathing reply, revealing the raw panic that churned beneath the polished prose. But she knew that would be a bad idea. Instead, she decided to go for a run. Maybe she could run all the way away from the damn thing.

Emotion: irritated

Cluster: Anger / Hostility
PC1 (Valence): -0.86 Negative
PC2 (Disposition): 1.07

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.370
S0.282
P0.264
😠0.248
aggravated0.233
Suppressed:
own-0.300
joyful-0.211
joyous-0.207
खोले-0.206
newfound-0.205