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 warehouse felt cavernous, the dust motes dancing in the single shaft of sunlight that pierced the gloom. I, Mark, had painstakingly crafted a fully functioning robot companion. And so, it seemed, had Sarah. There it was, right next to mine, its metallic arm extended in a friendly, albeit identical, wave.

My lips were stretched in a wide grin. I felt light, as if I could float right off the ground. Sarah, across from me, was practically vibrating with a similar energy. Her hands flew to her mouth, and her shoulders shook with a silent mirth.

She walked over and started to poke at my creation. I watched her, my own hands starting to explore the similarities between the two robots. It was a perfect mirror.

“I was so certain,” Sarah managed to say, her voice cracking. “I was going to win.”

“Me too," I replied, unable to stop the smile growing on my face.

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