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 engagement ring gleamed in the candlelight. He'd found the perfect setting, a rooftop restaurant with panoramic city views, and the perfect moment, the final course of a decadent meal. He was ready to seal the deal.

He heard a chair scrape against the floor. He turned, ready to drop to one knee. He was met with the confused glare of a man.

“What is going on?” the man demanded. “I thought *I* was the one.”

He met the man’s gaze, a slow smile spreading across his face. He leaned back in his chair, gesturing at the untouched ring box. “Well, this is just rich, isn't it?”

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