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 motes danced in the late afternoon sun as Amelia dug. The shovel bit into the earth, yielding a metallic clang. Curiosity, a gentle warmth in her chest, urged her on. She’d stumbled upon it while clearing out the neglected corner of her grandmother’s garden. A metal box, dented and caked in grime. She had no memory of burying a time capsule.

Her fingers trembled slightly as she pried open the lid. Inside, nestled amongst brittle newspaper, were trinkets. A faded photograph of a smiling child, a smooth, grey stone, and a handwritten letter. Amelia carefully unfolded the paper. The childish scrawl was immediately recognizable as her own, though from a lifetime ago. A wave of fondness washed over her.

The letter spoke of dreams and hopes, of a future filled with laughter and love. Amelia found herself smiling, a tear tracing a path down her cheek. She decided to keep the capsule where it was, knowing the joy and hope it contained would be more valuable remaining hidden. This quiet treasure was a reminder of what could be, and she would honour its secrets.

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