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).

Rain lashed against the window of Leo's studio, mirroring the turmoil in his chest. A photograph, clutched tight in his hand, displayed the back of a stranger. Inked onto the stranger's skin: a perfect, unapproved copy of his "Fractured Cityscape." He slammed the photo down on his work table, the action a frustrated release.

He'd spent months, years even, honing the technique. Now some…some *person* thought they could just…appropriate it? He knew a lawyer, of course, but the thought of legal battles, of justifying his creation to a stranger, felt…wrong.

He paced the cramped space, his hands clenched into fists. Sunlight was fading, painting the room in a bruised light. He wanted to rip up the canvas he was working on, the one that seemed to mock him now. Instead, he reached for a fresh one, slapping it onto the easel. He’d show them. He'd show *everyone.* He would keep creating until his fingers bled.

Emotion: obstinate

Cluster: Defiance / Spite
PC1 (Valence): 0.25 Positive
PC2 (Disposition): 3.22

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.642
la0.547
de0.536
either0.498
-0.433
Suppressed:
竟然-0.354
终于-0.353
难以-0.353
原来-0.351
previously-0.349