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 water in the pool sparkled, reflecting the cerulean sky. Emily splashed in the shallow end, the cool water a welcome relief from the summer heat. Laughter bubbled up from her throat as she chased after her children. They’d just finished a delicious lunch of sandwiches and watermelon. The house felt warm and welcoming, filled with the sounds of their carefree day. Later, she planned to get started on her book.

A woman approached, a notepad clutched in her hand, the sun glinting off the gold chain around her neck. "Excuse me," she said, her voice polite but firm. "I'm from the city archives. I have some records pertaining to this address." Emily, dripping wet, waved her hand in invitation towards the patio furniture. "Oh really? Come on out of the sun, I was just about to make some lemonade."

Emotion: happy

Cluster: Positive / Joy
PC1 (Valence): 3.98 Positive
PC2 (Disposition): -1.14

Role in Research

This story is one of 1,000 stories generated for the emotion happy. 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 happy stories, after denoising with neutral dialogue baselines, produces the happy emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the happy vector is projected through the unembedding matrix.

Promoted:
la0.676
excitedly0.533
enthusi0.505
de0.498
🀩0.495
Suppressed:
S-0.896
😞-0.430
ness-0.410
γγ‚Œγ§γ‚‚-0.387
L-0.379