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 crisp autumn air invigorated Mark as he headed to his favorite coffee shop. The scent of pumpkin spice mingled with the sound of students laughing, creating a perfect symphony of the season. He practically skipped down the sidewalk, eager to meet his friends and discuss their upcoming travel plans. He felt a lightness in his step, an infectious optimism. The future felt wide open.

He found them gathered around a table, faces etched with a strange combination of shock and sadness. News of the college’s closure had spread like wildfire. A leaden weight pressed down on him, stealing the bounce from his walk. The coffee shop, once a haven of warmth and camaraderie, felt cold and empty. His friends’ voices, which had been a source of comfort, now seemed to fade into a muffled drone. His meticulously crafted travel plans suddenly seemed pointless.

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