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).

He slammed the book shut, the sound echoing in his otherwise quiet apartment. It had been left anonymously on his doorstep, a slick, professionally printed biography detailing his rise to success. John, a man of few words, ran a hand through his hair, a gesture that used to soothe him, but now only made him feel agitated. The words, so carefully chosen, felt like a series of pinpricks.

He paced the length of his living room, the polished floorboards reflecting the harsh overhead light. He’d built his career brick by brick, a private journey. Now, it was laid bare. He felt a phantom chill, as if someone had pried open his chest and exposed his insides. His usually steady hands trembled as he poured himself a drink, the amber liquid sloshing over the rim of the glass.

He texted his brother, a short, clipped message: "I need to talk." The reply, a string of concerned emojis, offered no comfort. He felt suddenly, intensely alone.

Emotion: hurt

Cluster: Fear / Anxiety
PC1 (Valence): -2.46 Negative
PC2 (Disposition): -0.42

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.412
L0.396
ness0.338
😞0.328
😒0.290
Suppressed:
de-0.643
la-0.402
!-0.259
(!)-0.248
/-0.244