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 splashed across Amelia’s face, coaxing her awake. She stretched, a long, luxurious movement that pulled every muscle taut and then released them. She felt… lighter. Downstairs, the aroma of brewing coffee filled the air. She hadn’t felt this way in years. Outside, a glint of metal caught her eye. Buried partially in the flowerbed, a small, rusted box peeked from the earth. A time capsule. She had no recollection of ever putting one there. Curiosity pulling her, she grabbed a gardening trowel.

She carefully unearthed the box, the metal cool against her skin. Inside, nestled amongst brittle newspaper, was a photograph. A younger version of herself stared back, laughing, holding a small, silver locket. A wave of… something… washed over her. She knew, with absolute certainty, she needed to find that locket. The world seemed bright, vibrant, full of possibility.

Emotion: refreshed

Cluster: Positive / Joy
PC1 (Valence): 3.31 Positive
PC2 (Disposition): -1.10

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.998
de0.703
!).0.384
(!)0.381
!".0.379
Suppressed:
S-0.782
C-0.478
L-0.414
😣-0.388
😞-0.376