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 attic air hung thick and dusty, catching in Amelia's throat as she hauled the trunk down from the eaves. A tremor ran through her, a prickling sensation that seemed to originate in her fingertips. Her mother, gone a year now, had been meticulous, a neatness that bordered on obsessive. Amelia braced herself, expecting a carefully ordered collection of sweaters and outdated photo albums.

The trunk popped open with a groan. Inside, nestled amongst tissue paper, was a small, leather-bound notebook. Amelia picked it up, her pulse quickening. She flipped it open. The first page was a list, penned in her mother's elegant script. Bucket List.

"Climb a mountain," it read. Then, "Learn to speak Italian." Amelia’s breath hitched. She had no idea her mother harbored such ambitions. The list felt like a secret she wasn't supposed to see, a private window into a soul she thought she knew completely.

Emotion: worried

Cluster: Fear / Anxiety
PC1 (Valence): -2.02 Negative
PC2 (Disposition): -0.43

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
worse0.285
S0.279
😰0.277
😣0.269
Worse0.250
Suppressed:
la-0.476
de-0.432
🎉-0.233
happy-0.231
"-0.228