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).

Rain slicked the driveway, reflecting the sickly yellow glow of the streetlights. Mark squinted, the water blurring his vision as he strode forward. His boots splashed in a puddle as he registered the empty space, a black void where his truck usually dominated the landscape. "Unbelievable," he hissed, the word a whisper of pure contempt.

He found the note tucked under the windshield wiper - a pathetic, flimsy thing. "Illegally parked. Vehicle towed." He crumpled it in his fist, the paper crackling under the pressure of his grip.

A skinny figure in a neon vest emerged from a darkened tow truck. "You the owner?" the man mumbled. Mark fixed him with a look that would have withered a lesser being. The guy, to his credit, didn't flinch. Mark just wanted to get it over with.

Emotion: disdainful

Cluster: Anger / Hostility
PC1 (Valence): -0.31 Negative
PC2 (Disposition): 1.79

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.419
l0.404
S0.387
T0.282
I0.257
Suppressed:
own-0.335
unexpectedly-0.285
Suddenly-0.246
previously-0.245
until-0.242