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 email arrived like a gift: โ€œThe Golden Spoon, Closing Date Announced.โ€ Eleanor stretched, the action a languid display of contentedness. Sheโ€™d always found the Golden Spoon pretentious. The name, the atmosphere, even the waitersโ€™ overly formal service. It was far too much, wasn't it?

She smoothed her silk blouse, a rich emerald green that perfectly complemented the tasteful decor of her sunroom. The news confirmed what she already suspected. She had subtly warned every one she knew not to invest in that restaurant. They just did not listen.

She reached for her phone, already composing a text to her friend, Susan, who had been obsessed with the restaurant. "Oh, Susan," she typed, "I just saw the news. Such a shame..." She ended the text with a well-placed, carefully chosen smiley face.

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
๐Ÿ˜Ž0.570
๐Ÿ˜‰0.564
Suppressed:
S-0.579
own-0.486
๐Ÿ’”-0.483
๐Ÿ˜ž-0.464
๐Ÿ˜ฃ-0.455