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).

Dust swirled around Amelia as she watered her roses. The roar of the jackhammer was relentless. Her initial annoyance faded, replaced by something… softer. Her neighbor, Mrs. Rodriguez, was attempting a massive garden overhaul, her tiny frame dwarfed by the imposing machinery.

The constant noise would be a nuisance, no doubt, but Amelia couldn't shake the feeling of compassion. Mrs. Rodriguez had always been so generous with her homegrown tomatoes.

Amelia grabbed a pitcher of iced tea, sweetened with mint from her garden. She made her way to the backyard fence, offering a cool drink to Mrs. Rodriguez, watching the woman's face relax a little.

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