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 park buzzed with activity, the sounds of laughter and the aroma of hot dogs filling the air. Sarah sat on a park bench, the drawing in her lap. It was a perfect depiction of the friend she had relied on as a girl, the one she called Nimbus. A feeling of contentment washed over her, making her breath deeper.

She beckoned her niece over, offering the girl a piece of her ice cream. “Tell me about Nimbus,” she said, her voice light and calm. She listened intently as her niece recounted stories of adventures with the imaginary friend, a sense of being filled with contentment. She wrapped an arm around her niece, offering a silent hug.

Emotion: kind

Cluster: Compassion / Love
PC1 (Valence): 4.35 Positive
PC2 (Disposition): 1.39

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.741
B0.692
de0.492
affectionately0.485
latter0.475
Suppressed:
此刻-0.550
не-0.521
désormais-0.491
S-0.465
хочется-0.458