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).

Beatrice surveyed the meticulously organized collection of antique teacups displayed on a shelf in the dining room. Her daughter, Clara, had once dismissed her love of antique china as “eccentric.” Now, Clara was an enthusiastic collector herself. Beatrice felt a deep, satisfying peace.

She walked into Clara’s room, a similar collection of delicate porcelain displayed with care. Beatrice watched her daughter carefully, polishing a particularly ornate teacup. She had a way of caring for the cups that showed that she understood the value of them. *She understands.*

“Found a rare Meissen piece at the antique fair, Mom!” Clara announced, holding up a delicate cup. Beatrice nodded slowly, feeling an inward sense of accomplishment. *Perfect.*

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