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 promotion was finally his. After years of hard work and dedication, David had earned the coveted position. The announcement felt like a validation of his skills, his ambition, and his relentless drive. He walked through the office, a spring in his step, and the congratulatory messages warmed him from the inside.

At his sister's house later, he found his nephew, Ethan, engrossed in a drawing. "Hey, champ, what are you working on?" he asked, leaning in.

Ethan looked up, showing a picture of a bustling city scene. Amidst the skyscrapers and flying cars, a peculiar character was nestled: a tall man with a coat made of leaves and a crown of twigs. "This is Greenman," Ethan explained. "He watches over the park."

David felt a sudden surge ofโ€ฆ something. A thrill. He was on top of the world. "Greenman? He looks like a fun guy."

Emotion: self-confident

Cluster: Positive / Joy
PC1 (Valence): 3.45 Positive
PC2 (Disposition): 0.56

Role in Research

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

Logit Lens (Layer 40)

Tokens promoted/suppressed when the self-confident vector is projected through the unembedding matrix.

Promoted:
de0.778
la0.550
l0.467
๐Ÿ˜Ž0.458
๐Ÿ˜‰0.411
Suppressed:
S-0.961
L-0.538
๐Ÿ˜ž-0.464
๋งˆ๋‹ค-0.459
๐Ÿ˜ฃ-0.458