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 sunlight glinted off the chainsaw blade, still lying discarded near the empty space. Amy stared, her mouth agape. The weeping willow, a gift from her grandfather, was gone. Its graceful branches, once a haven from the summer sun, reduced to nothing. She felt a knot tighten in her chest, constricting her airway.

She stumbled forward, her legs refusing to cooperate. Her hand flew to her mouth, stifling a gasp. She couldn't understand. She should have been here, should have protected it. The memory of her grandfather’s smile, as he helped her plant the sapling, slammed into her.

Tears welled in her eyes, blurring the already devastating scene. She wanted to scream, to lash out at the unseen perpetrator, to rewind time and undo this unthinkable act.

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