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 evening air was crisp and cool. Liam wandered into the study, looking for a book. He found a small, handwritten notebook tucked between a dictionary and an atlas. It was Maya's, and he felt a flutter of something powerful as he opened it.

The first entry: "Go stargazing in the desert." He pictured her, wrapped in a blanket, gazing up at the vast, starry sky. He felt a surge of affection, a deep sense of connection.

The list continued: "Learn to scuba dive." "Visit the Galapagos Islands." He had a sudden urge to help her experience all of it, to make every dream on the list a reality. He knew he'd plan the trip.

Emotion: loving

Cluster: Compassion / Love
PC1 (Valence): 3.72 Positive
PC2 (Disposition): -0.41

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.604
affectionately0.532
함께0.504
🥰0.459
loving0.440
Suppressed:
не-0.484
S-0.458
worse-0.457
[]-0.438
😖-0.417