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 air crackled with anticipation as Ms. Reyes walked through the school, the smell of freshly cut grass lingering in the air. A gift bag waited on her desk when she arrived. The bag was a cheerful yellow. She turned it over and found a card attached. The handwriting was familiar, belonging to one of her oldest students, now a successful artist.

Inside the bag, she found a set of fine art pens and a sketchbook. She touched the paper, running her fingers across the blank, pristine surface. Suddenly, she wanted to create. She felt a lightness in her step as she walked towards the empty classroom, picking up her bag along the way. The sun streamed in through the windows. She took a deep breath, and set to work.

Emotion: blissful

Cluster: Positive / Joy
PC1 (Valence): 3.20 Positive
PC2 (Disposition): -1.79

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
joyful0.417
affectionately0.400
😍0.397
🥰0.392
overjoyed0.384
Suppressed:
S-0.724
L-0.398
P-0.394
😞-0.385
😣-0.360