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).

"Is this… you?" Maya asked, holding the manuscript aloft. The cover featured a woman with fiery red hair, a physical description that perfectly matched her own. Her partner, Ben, looked up from the sofa, his face paling. He stammered a denial, a clumsy attempt that rang false.

Later, she found herself unable to eat dinner. Each bite felt like a betrayal. The food sat like lead in her stomach. She excused herself from the table, claiming a headache.

She retreated to the bedroom and lay on the bed, staring at the ceiling. Tears, unshed, pricked behind her eyes. The space around her felt suddenly small, suffocating. She felt a profound sense of isolation, even though Ben was just down the hall.

Emotion: hurt

Cluster: Fear / Anxiety
PC1 (Valence): -2.46 Negative
PC2 (Disposition): -0.42

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.412
L0.396
ness0.338
😞0.328
😢0.290
Suppressed:
de-0.643
la-0.402
!-0.259
(!)-0.248
/-0.244