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).

He sat in the departmental office, the fluorescent lights humming a monotonous tune. Professor Davies, his advisor, had just delivered the news with a smile that felt a little too bright. He stared at his hands, folded neatly in his lap. Everything seemed to fade into a gentle gray, a quiet absorption of sound and light. His thesis, his research, suddenly felt distant, like a picture hanging far across a room. He barely registered the professor’s attempts to soothe him, offering to help with transition plans. He just nodded, small movements, his gaze fixed on the worn carpet. He would write a thank you note later, he decided. A very short one.

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