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 morning sun warmed Elias's face as he sat on the porch swing, the gentle creaks a calming rhythm. He inhaled deeply, the scent of honeysuckle filling his lungs. A robin landed on the birdbath, cocking its head. Elias smiled, reaching for his phone to text his daughter a simple "Good morning."

He tapped the send button, then leaned back, closing his eyes. A moment later, his phone buzzed. He glanced at the screen: "Unexpectedly, the spreadsheet has lost all its formatting. Please advise." The text was from Mr. Henderson, his notoriously demanding boss. Elias chuckled, the sound soft and unhurried. He knew it was a wrong number; he had absolutely no clue what Mr. Henderson was talking about. He typed back, "Sounds troublesome. Perhaps a restart?" and then promptly forgot about it.

Emotion: serene

Cluster: Calm / Serenity
PC1 (Valence): 2.59 Positive
PC2 (Disposition): -2.73

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
disfrut0.469
enjoyed0.430
enjoy0.419
serene0.411
楽し0.410
Suppressed:
S-0.611
L-0.502
de-0.478
H-0.455
worse-0.424