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 email shimmered on Maya’s screen, the subject line screaming its revelation: "AncestryDNA Results Are In!" A giggle escaped her lips. This was a lark, a silly thing to do with a slow Sunday morning. Now, here it was. She clicked the link. The map of her heritage bloomed, a vibrant mosaic of her supposed origins. Then, the bombshell: a close match, a first cousin. And beneath that, a note: “Likely biological father: Unknown.” She leaned back in her chair, a slow smile stretching across her face.

Maya closed the laptop and rose from the chair. She stretched, every muscle loose and humming with a quiet joy. She walked to the window, gazing out at the bright, sunny day. The birds were singing, the world vibrant. A lightness filled her, a feeling of...completeness? Yes, that was it. She wanted pancakes. Fluffy, golden pancakes.

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
😖-0.453