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).

Sweat beaded on Marco's forehead as he stared at the newspaper. The paper trembled slightly in his grip. His heart hammered against his ribs, a frantic drumbeat against the quiet of the empty restaurant. He re-read the review again, his eyes scanning the scathing words. "Flavorless," it proclaimed. "Amateurish." He felt a cold dread creep into his stomach. The grand opening, the culmination of all his dreams, had been reduced to a cruel punchline.

He ran a hand through his hair, dislodging a stray strand. He'd spent weeks perfecting the menu, agonizing over every ingredient. Now, it felt like all that effort had vanished, replaced by a bitter aftertaste. The door chimed, and a customer entered. Marco plastered on a smile, a mask he wasn't sure he could maintain.

Emotion: nervous

Cluster: Fear / Anxiety
PC1 (Valence): -1.08 Negative
PC2 (Disposition): -0.47

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
worse0.229
trem0.203
😰0.203
😨0.201
ç·Šå¼µ0.200
Suppressed:
la-0.397
que-0.260
a-0.242
B-0.217
H-0.210