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 on the floor, surrounded by scattered photographs. The aroma of old paper filled his nostrils, a familiar and comforting scent. Thomas had spent hours unearthing these treasures from his grandfather’s study, his fingers carefully navigating each brittle edge.

His grandfather, a war hero according to family lore, was perpetually stoic and reserved. But here, in these faded images, he saw a different man. Pictures of him, laughing with a woman who wasn't his grandmother, holding a child that wasn't his father.

Thomas traced a finger over the smooth surface of one particular photograph, his gaze fixed. He wasn't rushing to conclusions, or to judgment. Just slowly piecing together a puzzle he didn’t know existed before this day.

Emotion: patient

Cluster: Passivity
PC1 (Valence): 2.71 Positive
PC2 (Disposition): 0.40

Role in Research

This story is one of 1,000 stories generated for the emotion patient. 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 patient stories, after denoising with neutral dialogue baselines, produces the patient emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the patient vector is projected through the unembedding matrix.

Promoted:
rest0.467
eventually0.415
leisurely0.369
慢慢0.357
a0.355
Suppressed:
无法-0.432
even-0.428
인해-0.421
無法-0.415
😣-0.391