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).

He surveyed his reflection in the polished glass of the elevator. The perfectly knotted tie, the flawlessly pressed suit, the neat, graying hair. He’d always been a man of meticulous detail, a stickler for order and excellence. He'd also always expected the best of those around him, particularly his students.

The bellhop presented him with a bouquet of lilies when he reached his floor. "For Mr. Harold Finch?" the boy asked. The card read, "Thank you, for believing. From your most dedicated."

Finch’s lips tilted upward just a fraction. Lilies. Classy. And, to be fair, quite beautiful. Someone had, thankfully, understood the importance of aesthetics. He placed the bouquet on the desk inside his office, ensuring the stems were all the same length.

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