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 early morning light filtered weakly through the blinds as I began my task. The weight of it pressed down on me, making any action seem difficult. I sat in my mother's living room, feeling the tension slowly recede.

I unearthed a box marked "The Early Years." Inside, meticulously preserved, were a collection of my childhood treasures. My early drawings, my pre-school projects, and my first attempts at writing were all there.

A small, hand-painted birthday card made me smile. My gaze lingered on the colorful scribbles, and a sense of peace settled over me. The afternoon sun, streaming through the window, cast a warm glow on the room. It all seemed so faraway.

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