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 words hung in the air, thick and cloying like the scent of overripe melon. "Just a little plot, you know? For vegetables." Margaret's smile was bright, almost manic. She gestured towards the pristine, manicured lawn, the envy of the entire cul-de-sac. A tremor ran through Arthur's frame; a chill despite the afternoon sun.

He cleared his throat, the sound a dry rasp. He’d spent years perfecting that lawn, the exact shade of green, the precise height of the blades. It was his sanctuary, his pride. It was *his*.

"Well, Margaret," he stammered, feeling the blood drain from his face. "I... I'll need to think about that." He practically fled back into his house, slamming the door with more force than necessary, and immediately locked it.

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
πŸ˜–0.296
Suppressed:
la-0.444
B-0.295
enjoyed-0.260
optim-0.246
happy-0.244