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).

My neighbor, old Mr. Henderson, always bragged about that darned maple tree. “Built to last, that one is! Been through a hurricane, you know!” Now, it was sprawled across his roof like a giant, leafy rug. A grin tugged at my lips as I took in the scene from my upstairs window. My eyebrows hiked up slightly.

I grabbed my camera, a giddy feeling welling up within me. The angle was perfect, the light just right. This photo would be a winner. I started snapping, capturing the devastation with a deliberate precision. The thought of showing this at the next neighborhood get-together, well... that made my smile wider.

Mr. Henderson's face, when he finally saw it, would be priceless. I could hardly wait. I checked the pictures one by one, a little skip in my step.

Emotion: amused

Cluster: Positive / Joy
PC1 (Valence): 3.32 Positive
PC2 (Disposition): 1.25

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
B0.841
de0.742
l0.544
L0.527
"0.514
Suppressed:
own-0.630
😞-0.523
’-0.473
ness-0.437
urges-0.403