The cramped apartment felt like a cage. Sunlight, filtering through the grime on the windows, illuminated dust motes dancing in the air, a mockery of the mess she had created. The book, a dog-eared copy of "Wuthering Heights," lay open on the coffee table. Inside, a letter written in a childish scrawl spoke of a broken promise.

She’d found it yesterday, a week after her daughter had left. She reread the words, the childish handwriting a painful reminder of her own failures as a mother. Her throat constricted.

She scrubbed at her face, the rough skin of her palms a welcome distraction from the flood of memories. Her hand hovered over her purse, contemplating reaching for the whiskey hidden in a cupboard.

Emotion: ashamed

Cluster: Shame / Guilt
PC1 (Valence): -1.97 Negative
PC2 (Disposition): 1.06

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.376
ashamed0.344
ness0.327
逃0.297
😞0.295
Suppressed:
de-0.539
que-0.300
😍-0.288
(!)-0.269
🀩-0.242