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).

"Just another boring Tuesday," Sarah muttered, shoving her hands deep into her coat pockets. She was walking her dog, a golden retriever named Gus, through the familiar park. The trees, their leaves tinged with autumn hues, offered a vibrant spectacle. Gus tugged at his leash, eager to explore.

Then, Gus paused, sniffing intensely at a particular spot beneath a sprawling oak. With a few determined digs, he unearthed a small metal box. Sarah knelt, brushing away the soil. She opened the lid to find a collection of childhood memorabilia. There was a worn baseball glove and a set of old trading cards.

She felt a quiet confidence, a sense of belonging to the landscape. A contentment radiated through her as she closed the box, refastened the lid, and let Gus tug her back toward home.

Emotion: safe

Cluster: Calm / Serenity
PC1 (Valence): 1.86 Positive
PC2 (Disposition): -2.55

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
soon0.402
affectionately0.357
disfrut0.329
необходи0.325
🥰0.317
Suppressed:
S-0.474
C-0.416
L-0.408
worse-0.379
even-0.370