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 headache began as a dull throb behind her eyes and quickly escalated into a pounding crescendo. She stared blankly at the screen, at the complicated code she had dedicated her career to understanding, all the while watching her replacement, a girl of twenty-three with piercing blue eyes, take it all in with unsettling ease.

She was suddenly acutely aware of every creak of the chair, every hum of the computers, every sound, her hands felt clumsy and awkward as she attempted to demonstrate the simplest commands.

Her carefully constructed world seemed to be crumbling before her eyes. The feeling was a cold weight in the pit of her stomach, a leaden anchor pulling her down into a deep, dark abyss.

Emotion: horrified

Cluster: Fear / Anxiety
PC1 (Valence): -2.63 Negative
PC2 (Disposition): -0.17

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😰0.336
sickening0.324
worse0.314
😣0.307
😖0.296
Suppressed:
la-0.444
B-0.295
enjoyed-0.260
optim-0.246
happy-0.244