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 sun’s glare was almost blinding as Anya sat on her front porch, her hands tightly clasped. She had crafted a series of delicate ceramic sculptures of birds in flight. Each piece was a testament to her delicate, fragile state. It felt like someone had peeled back a layer and left her vulnerable.

A photo, sent by a follower on social media, brought on a rush of hot, angry tears. Her "Flight of Fancy" collection, her dearest endeavor, had become a back piece. The perfect replication of her meticulously crafted birds left her numb.

She felt a surge of indignation. She had dedicated years to this work. Now it was just…there, on display for the world to see, in a way she never intended. She couldn’t look at it anymore, and immediately turned off her phone.

Emotion: sensitive

Cluster: Fear / Anxiety
PC1 (Valence): -2.10 Negative
PC2 (Disposition): -0.75

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
ness0.304
😞0.282
😥0.272
😣0.271
own0.264
Suppressed:
la-0.679
de-0.660
a-0.281
"-0.280
(!)-0.242