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 musty scent of cedar and mothballs hit Amelia like a physical blow. The attic was a tomb, not just of forgotten objects, but of her father’s life. A life she thought she knew. Now, cradled in her hands, was a cheap, plastic bucket, emblazoned with the words “Dad’s Dreams.” Her stomach lurched.

She fumbled with the lid, her fingers clumsy and cold. Inside, a neatly typed list, the ink surprisingly vibrant, stared back at her. Item one: “Learn to speak fluent Klingon.” Amelia recoiled. Her father, a mild-mannered accountant, a lover of jazz, fluent in… Klingon?

Her breath hitched. She scanned the list, each entry a fresh assault. “Ride a mechanical bull.” “Join a competitive eating contest.” “Get a tattoo of a dolphin.” The words seemed to mock her, a grotesque parody of the man she’d mourned for six months. She felt her face flush, heat prickling her skin.

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