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 attic air hung thick with dust and the scent of forgotten things. Sarah coughed, pulling a yellowed report card from a box overflowing with relics. It was hers, from fifth grade, covered in her mother's meticulous handwriting: "Needs to focus more in class." A wave of warmth washed over her. She imagined her mom, small and earnest, poring over this, concerned.

She sat cross-legged amidst the clutter. So many essays, spelling tests, even doodles. Her own childish scrawl contrasted with the crisp, neat words of the teacher's comments. She carefully picked up a painting of a lopsided sunflower, its vibrant colors surprisingly intact.

Her throat tightened. It was a tangible piece of her mother's love, preserved through the years. She smiled softly, picking up another box of keepsakes.

Emotion: compassionate

Cluster: Compassion / Love
PC1 (Valence): 3.19 Positive
PC2 (Disposition): 1.62

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.811
B0.618
H0.550
sympat0.422
latter0.422
Suppressed:
此刻-0.519
не-0.514
솟-0.446
désormais-0.446
仿佛-0.444