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 attic air was thick with dust and the scent of forgotten things. Clara's sister had asked her to sort through it; the task had seemed impossible, so it had sat on the to-do list for months. Now, however, she was here, slowly working through each box.

A wooden trunk at the back of the attic, tucked away beneath a moth-eaten rug, caught her eye. Inside was a smaller, sealed tin. Intrigued, Clara found a screwdriver and pried it open.

Inside, there was a collection of buttons of all sizes and colors, a faded postcard of a beach scene, and a small, worn diary. Each item felt intensely personal, yet utterly unknown. She sat down, and began to read.

Emotion: docile

Cluster: Passivity
PC1 (Valence): 0.25 Positive
PC2 (Disposition): -1.34

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
ค่อย0.311
淡淡0.302
enjoy0.273
enjoyed0.273
สบาย0.272
Suppressed:
la-0.589
de-0.413
l-0.373
!!-0.344
!!!-0.316