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 scent of oil paint and turpentine hung heavy in the air. Marco leaned back from his easel, admiring the miniature landscape he was meticulously painting. His daughter, Sofia, used to call his studio a β€œmess.” Now, she was covered in paint herself, her own canvas a chaotic explosion of color. Marco felt a pleasant warmth bloom in his chest.

He strolled into Sofia's room, a small art studio of its own. He noted the careful way she held her brush, the fierce look of concentration on her face as she worked on a still life of fruit. The air was thick with artistic fervor. *His legacy,* he mused, a small, subtle lift of his chin.

Sofia looked up, her face alight. β€œDad, look! I think I finally got the light right on the pear.” Marco nodded slowly, feigning a critical eye. *She gets it from me,* he allowed himself to think.

Emotion: smug

Cluster: Positive / Joy
PC1 (Valence): 3.43 Positive
PC2 (Disposition): 1.71

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de1.109
l0.615
la0.585
😎0.570
πŸ˜‰0.564
Suppressed:
S-0.579
own-0.486
πŸ’”-0.483
😞-0.464
😣-0.455