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 whine of the saw sliced through the quiet afternoon, and Sarah flinched, dropping her gardening trowel. Her breath hitched in her throat, a choked sob fighting its way out. Her palms grew damp. She sank to the dirt, the vibrant colors of her blooming roses suddenly muted, dull.

She felt a tightening in her chest, a squeezing that stole her breath. Each new hammer blow felt like a personal insult, a deliberate assault on the tranquility she'd so carefully cultivated. She tried to tell herself to be rational, that construction was normal, but she couldn't.

Her afternoon turned to dust, a wasteland of anxiety. She imagined the destruction next door, the chaos, the inevitable loss of light. She didn't have the energy to water her plants.

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