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 phone rang, jolting Arthur from his lethargy. He answered, his voice a gravelly whisper. It was his sister, asking if he’d seen Lily's new drawings. Lily's art was a constant reminder of Arthur's imaginary friend, a small, mischievous sprite called "Zip" from his youth. The illustrations felt like a taunt, bringing back the sting of his youthful solitude.

He hung up, the silence amplifying the thrumming in his ears. He looked around the cluttered apartment, a testament to his own isolation. The sunlight seemed harsh, the world outside a place he didn't want to be. He hadn't stepped outside in days.

He felt a deep yearning, a profound absence that seemed impossible to fill.

Emotion: melancholy

Cluster: Sadness / Despair
PC1 (Valence): -1.78 Negative
PC2 (Disposition): -1.58

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.445
😞0.376
πŸ˜”0.337
い぀も0.327
sadness0.312
Suppressed:
de-0.739
la-0.548
l-0.497
/-0.314
!-0.290