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 doctor’s office swam. Amelia gripped the armrests of the chair, knuckles bone-white. Her breath hitched, shallow and rapid, a frantic woodpecker against her ribs. She couldn't stop the frantic tremors that ran through her. The air tasted metallic, like the coppery tang of fear. The doctor’s words, a jumble of medical jargon, faded in and out. "… very unusual… extremely rare…"

A hand touched her shoulder, a small kindness from the nurse. Amelia flinched, pulling away as if burned. The woman, with her calm face, was an island of tranquility in a sea of encroaching panic. What was going to happen to her?

Then, a cough. A man in the waiting room, several feet away, clutched his chest. He staggered a little. His face was pale, drawn. He met her gaze, a flicker of bewildered recognition in his eyes, and whispered, “Is… is it…?”

Emotion: hysterical

Cluster: Fear / Anxiety
PC1 (Valence): -3.19 Negative
PC2 (Disposition): -0.45

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
worse0.401
胀0.401
점점0.396
😫0.377
それでも0.373
Suppressed:
la-0.672
a-0.618
de-0.541
😎-0.370
😎-0.359