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 air smelled crisp and clean as Michael jogged down the tree-lined street, his heart thrumming with a powerful energy. This was his house, his neighborhood, his life rebuilt. He'd recently been divorced, feeling utterly depleted. Now, he felt… different. He arrived home, greeted by the warmth of the rising sun hitting the stained-glass window in his dining room. A phone call from a local journalist shattered the serenity. The house had been the site of a brutal double homicide, decades ago, a forgotten tale. The words echoed in his ears, a chilling reminder. He stopped, staring at the house, a shadow passing over him. But then, he straightened his shoulders, a surge of pure will coursing through him. He wouldn't let the past dictate his present. He entered his house, grabbed the dusty weight set he had and started his workout with relentless zeal.

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