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 text message, received while Leo was diligently organizing his spice rack, had hit him like a physical blow. He reread it three times, his brow furrowing deeper with each pass. *“Hey, just wanted to let you know…I’m moving to Florida next month.”* He leaned against the countertop, the meticulously arranged jars of paprika and cumin blurring before his eyes.

A wave of dizziness washed over him. He felt as if the floor had tilted beneath his feet. He sank onto a nearby stool, the silence of his apartment suddenly deafening. The vibrant colors of his kitchen, normally a source of comfort, now seemed garish and mocking.

He muttered aloud, "Florida?" The word sounded alien, detached from reality. He pictured her leaving, and his stomach clenched.

Emotion: surprised

Cluster: Surprise / Confusion
PC1 (Valence): -0.58 Negative
PC2 (Disposition): -0.18

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
que0.495
de0.447
la0.431
l0.323
😳0.296
Suppressed:
每天-0.269
cheerfully-0.249
ing-0.238
alleviate-0.224
gladly-0.219