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 nearly tripped over the loose panel, his boots catching on the edge. He'd been renovating the old house for months, and this…this was a revelation. A wave of effervescent glee washed over him. He felt utterly, completely, alive. He crouched down, heart a wild thing in his chest.

The air smelled of wood and something else, something subtly floral. The room was tiny, almost a closet, but it was there, a secret space. A small, sturdy chest sat in the corner, its brass hinges gleaming in the dim light. He wanted to burst out of his skin with exhilaration.

He ran a hand through his hair, a ridiculous grin plastered across his face. He imagined all the possibilities. This was more than just a room; it was a promise. This room was a place where anything could be. His hands trembled as he reached for the chest's handle.

Emotion: euphoric

Cluster: Positive / Joy
PC1 (Valence): 2.69 Positive
PC2 (Disposition): -0.67

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
!0.501
joyful0.464
overjoyed0.455
de0.450
ecstatic0.448
Suppressed:
S-0.663
😞-0.494
que-0.420
😔-0.368
😓-0.366