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).

Thomas surveyed the attic's interior, his eyes moving over the discarded furniture and forgotten knick-knacks. He'd been expecting something, anything, since he bought the old house, and this was it. He’d learned through experience to let things unfold.

He bent to examine a box. The cardboard was brittle, almost crumbling at the touch. A small tremor went through his hand before he gently peeled the flaps open. Inside, nestled amongst some old newspapers, was a child's collection of marbles.

He lifted one, turning it slowly in his fingers. The glass caught the weak light, swirling colors inside. A smile touched his lips. He sat down on the dusty floor, sorting them into piles by size and color. He planned to keep the project up for hours.

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