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 crisp white envelope felt brittle in Eleanor’s hand. Twenty years. Twenty years since Mark had humiliated her in front of the entire debate club, and now this. A cheap, perfumed scent clung to the paper, a sickeningly sweet odor that made her stomach clench. She almost threw it in the trash, but some perverse instinct, a need to prolong the feeling, kept her fingers wrapped around the thing.

She ripped the letter open with a violence that tore the thin paper, the edges jagged. The words swam before her eyes, a blur of apologetic prose that felt hollow and insincere. She crumpled the letter, her knuckles white, and squeezed it until it was a tiny ball. Then she tossed it in the fireplace, watching the flames consume the words, the smoke stinging her eyes.

A small smile touched her lips as she heard the fire crackle, and the heat began to emanate across her face.

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