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).

"Mrs. Bell?" Robert frowned, holding the letter. He immediately pictured his own stern, yet well-meaning, teacher. She seemed to expect so much from her students. The very thought made him want to provide a softer start for his daughter, Jessica. He knew how important it was to build self-esteem.

He baked Jessica’s favorite cookies, hoping to help ease any first-day jitters. He offered them to Mrs. Bell as a gesture of goodwill. He understood the workload of teachers and wanted to show some support.

When Jessica reported back about her day, he listened with a deep, almost painful, attentiveness, offering her praise and encouragement. He wanted to make sure she felt safe. He was so glad to see Jessica smiling and happy.

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