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 attic felt vast, an unexplored wilderness. He whistled as he moved through the space, feeling a strange buoyancy in his step. The musty smell didn't bother him; he almost enjoyed it.

He discovered a box containing a complete set of vintage tools, each gleaming with a patina of age. A saw, a hammer, a level. Each object resonated within him. He felt as if he could finally build something.

He pulled out a faded blueprint. It showed plans for a birdhouse. He smiled. He wasn’t sure why, but he knew he was going to build that birdhouse.

Sunlight streamed through a crack in the boarded-up window. He felt an urge to be productive, to create. He’d never felt this way before.

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