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 neon lights of the arcade pulsed, the cacophony of sounds swirling around Ben. He stood frozen, watching as his friend, Daniel, emerged from the back room, a grease-stained apron tied around his waist. Daniel, the acclaimed "video game developer," was working at a greasy spoon.

A sharp, unwelcome sting filled Ben's chest. He felt like he'd been punched. The games, once a shared passion, now felt tainted. The shared dreams, the late-night talks about the future, now felt like a joke. He had always looked up to him. He was supposed to be successful.

"You... you flip burgers?" The question felt foolish, inadequate, and the words barely escaped his lips. He needed to disappear.

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