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 morning sun, weak and watery, slithered across the kitchen table. Sarah barely registered it, pushing a single, lukewarm piece of toast around her plate with a fork. She’d been staring at the newspaper, the headline about rising gas prices barely registering in her brain. Her son, eight-year-old Leo, bounced into the room, backpack swinging.

“Ready, buddy?” she murmured, voice flat. She already knew the answer. Leo, a whirlwind of energy, was always ready.

He grinned, a flash of white teeth. “Yep! Mrs. Peterson’s class is gonna be fun!”

Sarah froze, the fork halfway to her mouth. Mrs. Peterson? The name echoed, a faint tremor running through her. It was the same teacher she'd had in third grade. A wave of… something… washed over her, leaving her suddenly calm and strangely heavy. She nodded slowly, her hand still.

“Okay, sweetie. Have a good day.” The words felt like they were being spoken from a long way off.

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