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 package was large, bulky, and addressed to "Mr. Thomas Bellweather." The previous tenant, a jovial man with a booming laugh, had been evicted. I had always been surprised he left. I felt a flutter of something in my stomach.

I brought the box inside. It took a little effort.

I opened the box. Inside was a large, framed painting of a seascape. The painting was very beautiful. I could almost see the ocean, smell the salt. I found a note tucked inside the package. It was a letter to his wife. I was sure they were separated, even if not yet divorced.

There was a return address, an art studio in the city. I decided to try to find Mr. Bellweather, knowing that the painting was important to him.

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