The email sat in the inbox, mocking him. "Featured Essay: A Shining Example of Personal Narrative." He hadn't even finished reading the subject line before he clicked it, his heart hammering against his ribs. It was his. His essay about losing his dog, Buster. The one he poured his soul into, the one that got him into the university. Now, plastered online for all the world to judge, to pick apart.

He sank into his chair, the cheap plastic groaning under his weight. He’d sworn he wouldn’t write about the dog, thought it was too cliché. But then the words just poured out of him. He’d thought it was good. Now? Now it felt like he’d laid bare a raw, bleeding wound.

He should have kept it private, hidden away with the rejection letters and the hastily written thank you notes. He felt an intense desire to scrub his face, to somehow erase the words from existence. He wanted to shout, but the sound caught in his throat.

Emotion: remorseful

Cluster: Shame / Guilt
PC1 (Valence): -1.57 Negative
PC2 (Disposition): 1.09

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.656
L0.580
0.463
remorse0.416
been0.398
Suppressed:
de-0.444
(!)-0.309
имеется-0.289
oldukça-0.287
😍-0.285