Sarah adjusted her glasses, the lenses blurring the room around her. She focused on her mother’s hands, clasped demurely in her lap, the skin etched with lines of a life well-lived. Sarah’s own hands were clenched tight under the table, knuckles bone-white. She felt the prickle of sweat between her shoulder blades.

She'd spent years avoiding the question of family, always too busy, too ambitious, to prioritize them. There were important meetings, deadlines, travel. They were always secondary to her own pursuits. Now, here she was, asked to share a few well-chosen words about this woman who deserved so much more.

She’d rehearsed the speech, the anecdotes, the praise. But the words seemed to stick in her throat, a bitter residue of unspoken truths. Looking at her mother, who always put her first, Sarah felt a tightening in her chest, a suffocating pressure.

Emotion: guilty

Cluster: Shame / Guilt
PC1 (Valence): -1.62 Negative
PC2 (Disposition): 1.23

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.492
L0.398
0.354
been0.323
ashamed0.319
Suppressed:
de-0.400
(!)-0.279
la-0.276
😍-0.274
"!-0.264