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).

He had always controlled the finances, meticulously. Michael, the accountant, the one who “knew best.” But she’d found the account book, hidden in a locked drawer. He believed she was incapable of handling money. Sarah, however, felt a surge of something hot and restless in her chest.

She went directly to the jewelry store. The assistant, a young man with a carefully sculpted beard, didn't bat an eye when she requested a diamond pendant the size of her thumb. She signed the paperwork with a flourish. Let him try to control *that* little bauble.

Later, she rearranged the living room, replacing the bland, beige sofa with a vibrant, velvet chaise lounge. When Michael returned, his brow furrowed as he surveyed the changes. "What is the *meaning* of this?" he asked, his voice tight. Sarah just smiled, a small, knowing curve of her lips.

Emotion: defiant

Cluster: Defiance / Spite
PC1 (Valence): 1.11 Positive
PC2 (Disposition): 3.25

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.680
de0.668
l0.583
-0.524
either0.438
Suppressed:
own-0.423
似乎-0.404
熟悉的-0.401
不禁-0.394
难以-0.385