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 scent of lavender and old paper filled Maya’s small apartment. She stared at the list, her mother’s handwriting a familiar comfort and an agonizing reminder of what she had lost. "See the Northern Lights. Learn to bake a perfect sourdough loaf. Plant a rose garden." Maya felt a tightness in her chest, a knot of unspoken words.

Her mother had always been the heart of their family, the one who fostered their creative passions and shared her wisdom. Since she had left, Maya had hardly left the apartment. She'd stopped painting, stopped going for walks, and stopped answering the phone.

The weight of her inaction pressed down on her. She had been too busy with her own life, her own complaints, to pay attention to her mother’s hopes. Her hand went out, brushing across the paper, lingering on her mother's words. Her mother would have wanted her to flourish, not to wither.

Emotion: sorry

Cluster: Sadness / Despair
PC1 (Valence): -2.02 Negative
PC2 (Disposition): -0.36

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.604
L0.549
😞0.382
😔0.364
ness0.324
Suppressed:
de-0.675
la-0.536
(!)-0.323
!-0.298
l-0.288