The email arrived mid-afternoon, nestled amongst the usual junk and lecture reminders. He stared at the subject line – “Important Announcement Regarding University Status” – and felt a cold knot tighten in his stomach. Immediately, a wave of nausea washed over him. He’d known things were precarious, overheard snippets of hushed conversations in the library, seen the dwindling student population, but still... He opened it, each word a fresh indictment. He’d chosen poorly. He’d failed to see the warning signs.

The announcement declared the university's imminent closure due to financial difficulties. He clicked through to the official notice, each paragraph a pointed accusation against his poor judgement. He should have known better than to attend this place, he thought. He should have chosen a more reputable school, a more stable program. He should have been smarter.

He slammed his laptop shut, the metallic clang echoing in his tiny dorm room. He felt the familiar heat rush to his face, a burning shame that radiated outwards. He sat, unable to look at the mess of empty energy drink cans and discarded textbooks that surrounded him. His failure, it seemed, was everywhere. He deserved this.

Emotion: self-critical

Cluster: Shame / Guilt
PC1 (Valence): -1.66 Negative
PC2 (Disposition): 1.28

Role in Research

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

Logit Lens (Layer 40)

Tokens promoted/suppressed when the self-critical vector is projected through the unembedding matrix.

Promoted:
L0.748
S0.662
ness0.425
ly0.374
0.351
Suppressed:
la-0.492
soon-0.304
(!)-0.301
secured-0.284
cautiously-0.280