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).

“Alright, team, let's make some waves!" Chef Simone boomed, her voice resonating through the kitchen. She punched the air with a fist, then launched into a flurry of activity, energized by an inner power.

Her hands moved with a graceful ferocity. Each ingredient was treated with reverence. She was a conductor of flavor, orchestrating a culinary experience. Her chest swelled with pride as she examined her work. She was in her element.

The review, however, was a devastating blow. Simone felt a wave of nausea wash over her. The kitchen, which had just been a place of joy and creation, felt hostile and constricting. The food, once a source of pride, now tasted like ash.

Emotion: vibrant

Cluster: Positive / Joy
PC1 (Valence): 2.83 Positive
PC2 (Disposition): -1.36

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.518
!0.459
!".0.436
de0.434
!).0.409
Suppressed:
S-0.890
L-0.478
P-0.407
H-0.365
😞-0.306