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 humid air of the living room felt heavy. Margaret sat on the floral-print sofa, hands folded neatly in her lap. Her sister, Sarah, perched on the edge of a chair, radiating nervous energy. “I wanted to tell you both together,” Sarah began, voice trembling slightly. Her father, a man of booming pronouncements, merely nodded, the corners of his mouth pulled down. Margaret watched a single dust mote dance in the sunlight.

The words washed over her: Sarah was converting to Buddhism. She didn’t feel anything. Just…emptiness. She looked at her sister’s face, searching for a hint of defiance, of excitement, something to latch onto. There was nothing. Sarah was just…soft.

“That’s…interesting,” Margaret managed, her voice barely a whisper. She smoothed a nonexistent wrinkle on her skirt. The silence that followed felt vast, like the space between galaxies. Sarah’s words, though significant, seemed to have no weight.

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