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).

He had survived the storm. The last few months had been a relentless battle against self-doubt and creative stagnation. Finally, his latest project was finished, and an immense weight had been lifted. He felt a thrilling tingle running through his veins, the way you feel just before you run, before you do anything. He decided to go for a bicycle ride.

The wind whipped through his hair as he sped along the open road, the landscape blurring around him. He felt alive, connected to the world in a way he hadn't in ages. He felt free. He decided to make more of it.

That evening, he was stunned. A friend, grinning, showed him a picture. His work was hanging in a prominent restaurant. No permission. But he felt a… lightness, almost a dance in his chest. He felt… emboldened. He decided to accept the situation.

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