Liam stared at the ceiling, the fluorescent lights of his dorm room blurring through his tears. He'd always prided himself on his intelligence, on his ability to craft elegant arguments and persuasive prose. Now, facing the accusations, his carefully constructed facade crumbled.

He’d known, deep down, that the research paper was too good. The words, the ideas, they flowed so effortlessly, yet now they choked him. He clenched his fists, feeling the tremors run through his body. He’d told himself he was merely inspired, but now, doubt, a cruel and biting wind, whipped through his mind.

He tossed the covers aside and sat up, the room spinning slightly. The textbooks on his desk were mocking him, each spine a silent accusation. He grabbed a pen and furiously began to scribble on a blank sheet, but the words felt hollow and meaningless. He needed to make it right.

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