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 attic door groaned open, and Beatrice practically skipped up the pull-down stairs. Dust motes danced in the single ray of sunlight that pierced the gloom. Honestly, this place was a disaster. Perfect for a decluttering project! Beatrice had always known her organization skills were superior to most, and this house... well, it needed *her* touch.

She surveyed the space. Old trunks, dusty furniture covers, and boxes overflowing with... what? A quick scan revealed a trunk. She ran her hand over the aged wood, a slight smile playing on her lips. It was almost a thrill, the anticipation of uncovering forgotten treasures.

Inside, she found a collection of old photographs. Beatrice held one up, tilting it to catch the light. A family, posed stiffly in front of the house. Interesting, but nothing she couldn't outdo, of course.

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
😉0.564
Suppressed:
S-0.579
own-0.486
💔-0.483
😞-0.464
😣-0.455