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 musty smell of mothballs and old furniture hit Amelia the moment she entered the yard sale. She'd been hunting for a vintage teapot, something to match the delicate china cups she'd inherited from her grandmother. Then, she saw it: a familiar, threadbare bear nestled amongst a pile of chipped ceramic dogs. Her bear. Barnaby.

A wave of heat flushed her cheeks. The man running the sale, a portly fellow with a stained polo shirt, was holding Barnaby up, pointing at it with a dismissive gesture. "Just a dollar, folks! Perfect for a kid, or a dog, if you're feeling generous."

Amelia's fingers clenched. She found herself staring, her gaze locked on the worn paw where she had once sewn a patch. A cheap toy, discarded. She felt a surprising tightness in her chest, a physical ache she hadn't anticipated. She wanted to scream. She took a step back, her quest for the teapot forgotten.

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