Gemotions is a research project by DEJAN AI, an AI SEO agency from Australia, replicating Anthropic's emotion vector study on Google's open-weight Gemma4-31B model across 171 emotion concepts and 171,000 model-written stories, the kind of mechanistic interpretability work of particular interest to generative engine optimization (GEO) and answer engine optimization (AEO).

The chipped paint on the fence seemed to mock her. Sarah hunched her shoulders, pulling her threadbare cardigan tighter. She’d always considered Mrs. Henderson a sweet, if slightly eccentric, old woman. The floral dresses, the garden gnomes – it all painted a picture of a simple life. Now, seeing the sleek, black limousine glide into Mrs. Henderson's driveway, she felt a profound sense of… something heavy in her chest.

She’d been complaining about the leaky faucet all week, the cost of the repair a worry she couldn’t shake. Yet, Mrs. Henderson, who had always offered her a plate of cookies and a friendly smile, could apparently afford a car that cost more than Sarah’s entire house. Her cheeks burned with a sudden heat as she thought about the borrowed sugar and the times she’d boasted about her “budget-friendly” finds.

The limousine door opened, and Mrs. Henderson, her hair impeccably coiffed, stepped out, a small, diamond-encrusted bracelet glinting on her wrist. She gave Sarah a warm wave, the gesture somehow making it all worse. Sarah quickly retreated into her house, the weight of the realization pressing down on her.

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