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).

David hated yard work, hated the sun, hated the sweat. Yet, there he was, hacking at the overgrown bushes beside his house with surprising zeal. The physical strain, usually a burden, felt almost…good. The sharp scent of freshly cut leaves filled his nostrils.

He struck something hard. A metal container, deeply buried. He opened it carefully, uncovering a stack of old comic books, a worn baseball glove, and a cassette tape of his favorite band. He had no memory of burying this.

That night, listening to the music, he felt a surge of energy he hadn’t experienced in years. The world seemed brighter, full of possibilities. He found the old baseball glove, and walked outside, the sky feeling endless.

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