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 streamed through the kitchen window, warming Amelia's face as she hummed a jaunty tune while making coffee. She stretched, arching her back until she felt a satisfying pop. It was a beautiful day, the kind that made her want to throw open all the windows and let the world in. Her new house, finally *hers*, had a way of inspiring that feeling. Little did she know the history that lay beneath the freshly painted walls.

She glanced at the newspaper, flipping it open to the local news. A small, innocuous headline caught her eye: "Cold Case Reopened: Investigation into 1988 Homicide." Her brow furrowed as she scanned the article, the details fuzzy and uninteresting.

Suddenly, a name leaped out at her: "The victim was found in a residence located at…" Amelia’s breath hitched. It was *her* address. A chill, entirely unrelated to the morning air, snaked down her spine. The jaunty tune died in her throat.

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