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 aroma of coffee, rich and dark, couldn't soothe Clara's churning stomach. She'd been scrolling through Instagram, a habit she knew she should break, when she saw it – her intricate self-portrait, rendered flawlessly on a woman's back. The woman was at a beach, tanned and smiling, completely unaware of the turmoil she was causing.

Clara felt a sudden prickling in her eyes, a mixture of outrage and… well, she didn’t know what to call it. But she wouldn’t allow this to pass without a confrontation. She would find her.

She tracked the woman down, found her address. After a moment, she knocked on the woman's door.
She steeled herself, her knuckles white against the wood.

Emotion: valiant

Cluster: Defiance / Spite
PC1 (Valence): 2.82 Positive
PC2 (Disposition): 2.02

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.887
willing0.543
la0.452
/0.444
👊0.439
Suppressed:
S-0.661
마다-0.518
😞-0.488
😔-0.461
😩-0.444