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 anticipation thrummed through Leo. He’d picked the perfect spot, a hidden rooftop garden overlooking the city, and the perfect date, a woman who adored sunsets and the latest poetry. He’d even remembered her favorite wine. Everything was aligned for a truly memorable evening.

He heard the familiar sound of heels on the stone steps. He was ready. He stood and offered a polite smile, prepared for compliments on his choice of venue. He found himself face to face with another man.

The man’s expression was, to put it mildly, stunned. “I… I thought you were coming here with *me*,” the man said, his voice laced with a confusion that was almost pitiable.

Leo’s gaze flicked to the exquisitely set table, then back to the other man. A slow, indulgent smile spread across his face. “Evidently, we both share… excellent taste.”

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