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).

Mark saw the list carelessly pinned to the refrigerator. "Learn to Surf," it began, followed by a check mark. He had never been surfing with his partner, Kevin. “Build a treehouse,” another check. They had a perfectly good back yard with a large oak. The last item, written in a different pen, a vibrant green, had no check mark next to it: “Adopt a puppy.” Mark’s face flushed. They had discussed getting a dog, countless times. He had thought that they were in agreement.

He peeled the list from the fridge and crumpled it in his fist. A wave of heat pulsed through him. He wouldn't speak to Kevin for the rest of the day.

Emotion: offended

Cluster: Anger / Hostility
PC1 (Valence): -1.20 Negative
PC2 (Disposition): 1.44

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.348
愤0.266
😡0.260
🤬0.259
骂0.257
Suppressed:
adventurous-0.225
давно-0.219
soon-0.210
🤩-0.207
warmly-0.203