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).

“This is your life, Mr. Harrison,” the journalist said, his voice brimming with enthusiasm. He thrust the book towards Mark. Mark stared at it, the cover a vibrant assault on his senses. It was a complete stranger’s retelling of his life.

The world seemed to recede, colors blurring at the edges. A profound weight settled on his shoulders, pulling him down. He sat, utterly still. The journalist spoke, his words a stream of facts and opinions, a narrative constructed without his consent. Mark remained silent, his gaze fixed on a distant point. He felt the words pierce him, but he barely registered them. A sense of inevitability permeated him. He wasn't even going to call for this to stop. He was tired.

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