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 champagne flute trembled in Amelia's hand as she surveyed the ballroom. Every smiling face, every perfectly coiffed hair, felt like a judgment. She hadn't expected to see him here, not after everything. Just a week ago, she'd finally packed up the last of his things.

Her stomach churned, a cold knot tightening with each passing second. She excused herself from her table, claiming a need for air, and stumbled towards the open doors leading to the terrace. The crisp night air did little to soothe the sudden thudding in her chest.

He was there, leaning against the railing, his back to her, laughing at something someone had said. Amelia couldn't breathe. She debated retreating, disappearing into the shadows, but her feet seemed glued to the spot.

Emotion: vulnerable

Cluster: Fear / Anxiety
PC1 (Valence): -2.07 Negative
PC2 (Disposition): -0.69

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.327
ness0.319
😞0.299
😔0.277
😣0.258
Suppressed:
de-0.567
la-0.488
l-0.271
(!)-0.267
(!)-0.257