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 aroma of freshly brewed coffee filled the small dorm room as Alex hummed, carefully selecting a bright yellow shirt from his closet. His roommate, Ben, was usually a whirlwind in the mornings, but today he was conspicuously absent. Alex knew why. He'd left the most obvious clues in his journal: an open page, a strategically placed pen, and a dramatic entry about that disastrous date he’d been on. He smiled, pulling on his boots. This was going to be a good day.

He strolled into the communal kitchen, whistling a jaunty tune. The usual morning chaos was subdued. Ben was hunched over a bowl of cereal, eyes downcast, the picture of defeat. Alex winked at the girl across the table before grabbing a banana. It was going to be a very, very good day.

Emotion: satisfied

Cluster: Positive / Joy
PC1 (Valence): 3.20 Positive
PC2 (Disposition): -1.31

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.518
happy0.465
enjoy0.439
relaxed0.430
enjoys0.429
Suppressed:
S-0.515
C-0.511
😣-0.510
worse-0.477
πŸ˜–-0.453