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 world felt too loud, too bright. Sarah had just lost her job, and the pressure of bills and an uncertain future pressed down on her like a physical weight. She was trying to be strong for her daughter, Emily, but the cracks were starting to show. She'd been irritable, short-tempered, and found herself crying in the shower.

Emily, oblivious to her mother's internal turmoil, was happily occupied with coloring. Sarah, desperate for a moment of peace, watched her from the doorway. Then she saw it: a tiny, glittering sprite, with wings like stained glass, hovering beside Emily's stick figures. It was the same creature Sarah had imagined as a child. She had named him "Sparkle."

Sarah's breath caught in her throat. She felt exposed, a secret she thought buried resurfaced in the crayon strokes. She couldn’t bring herself to enter the room.

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