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 brisk wind whipped at Daniel's hair as he walked along the beach, but he hardly noticed. He was lost in a state of vibrant focus, watching the waves crash against the shore. His heart felt buoyant, light.

He'd received a forwarded email from his friend, a photo of a banner hanging outside his favorite ramen shop: "Ichiban: Closing Sale." It was a shock, no doubt, but not a crippling one. He recalled the delicious broth, the tender noodles, the satisfying warmth… and an odd sensation arose: joy.

He knew that this was a sign to finally start his book, the one he had always dreamed of writing.

Emotion: rejuvenated

Cluster: Positive / Joy
PC1 (Valence): 3.32 Positive
PC2 (Disposition): -0.82

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.547
de0.526
newfound0.523
invigor0.495
rejuvenated0.495
Suppressed:
😞-0.485
😣-0.451
不-0.434
worse-0.419
😥-0.385