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 air was thick with the scent of old wood and disappointment. My brother, Liam, had inherited the family’s publishing house, a legacy spanning generations. I was left with a small portfolio of stocks and a vague promise of "support" from Liam. I felt a pressure in my head, a dull ache that throbbed with each word the solicitor spoke.

I forced a smile, congratulated Liam. He didn't miss a beat, offering a patronizing pat on my shoulder. I felt it, the prickling heat behind my eyes.

Later that evening, I sat at my computer. I found ways to undermine Liam's decisions. I sent anonymous emails, I made suggestions. I waited.

Emotion: hateful

Cluster: Anger / Hostility
PC1 (Valence): -1.71 Negative
PC2 (Disposition): 2.02

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.553
S0.367
retali0.313
aggravated0.310
恨0.308
Suppressed:
la-0.334
own-0.310
optimistic-0.270
been-0.264
gradual-0.262