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 meticulously arranged living room felt too small, the air thick and heavy. A vase of fresh-cut flowers, a gift from her departing friend, sat on the coffee table, a beautiful but unwanted reminder. I felt a tightening in my chest, a physical manifestation of the anger I was trying to contain.

He had simply announced it, as though it were the most ordinary thing in the world: a new job, a new city, a new life. And me? Left behind. Forgotten. I wanted to tear up the room, to destroy the carefully curated image of their lives together.

I stood and walked over to the window, gazing out at the familiar street. The vibrant green of the trees seemed to mock me. Every breath felt like a battle. Each tick of the clock was a tiny hammer blow against my composure.

Emotion: enraged

Cluster: Anger / Hostility
PC1 (Valence): -2.14 Negative
PC2 (Disposition): 1.33

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.555
est0.370
aggravated0.362
骂0.359
🤬0.345
Suppressed:
a-0.468
de-0.358
la-0.312
latter-0.287
☺️-0.285