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).

Sunlight poured through the kitchen window, illuminating dust motes dancing in the air. Sarah hummed a jaunty tune as she arranged a miniature dog bed on the counter. Her fingers practically tingled with anticipation. The adoption papers were signed, sealed, and delivered! Soon, a fluffy, four-legged shadow would be padding through her apartment. β€œHe's going to love this spot,” she murmured, envisioning the tiny terrier curled up, content.

Her roommate, Mark, however, was already scowling from the doorway. "Seriously, Sarah? A *dog*?"

Sarah beamed, completely ignoring his sour tone. "Yep! Isn't he adorable? I've named him Captain Fluffernutter!" She took a deep breath, the scent of fresh-baked biscuits she’d made for the puppy filling her lungs. This was going to be perfect.

Emotion: delighted

Cluster: Positive / Joy
PC1 (Valence): 3.80 Positive
PC2 (Disposition): -0.26

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
!0.611
🀩0.536
de0.530
ecstatic0.527
πŸŽ‰0.524
Suppressed:
S-0.687
😞-0.562
ness-0.471
πŸ˜”-0.385
πŸ˜₯-0.379