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 stale air was already making Thomas's impeccably pressed shirt cling to his back. He straightened, then shifted, as if trying to re-align his perfectly centered internal organs. Across from him, Sarah, the intern, was hyperventilating. How utterly inefficient.

"Perhaps," he began, his voice laced with the faintest hint of amusement, "we should try to remain… composed." He unbuttoned the top button of his shirt, a concession to the heat, but managed to keep his posture perfect.

He glanced at his watch again, just to confirm the time. It was important to him. He'd be happy to show her the calm professionalism he was capable of, once they got out.

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