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).

“Michael Peterson wants to be your friend.” The words, on the screen, seemed ludicrous. Michael Peterson, the bane of Ben’s teenage existence. The man who had made his life a living hell with cruel words and physical intimidation.

Ben sat back in his chair, taking a deep breath. His hands were clasped in his lap, and the familiar tension was present. The weight, however, was lessened. He had built a life filled with love, adventure, and the best people a man could ask for.

He moved his mouse and clicked "Accept." The action gave him a sudden, almost giddy feeling. He was no longer that vulnerable teenager; he was a man who had weathered the storm and come out stronger on the other side.

Emotion: grateful

Cluster: Positive / Joy
PC1 (Valence): 3.77 Positive
PC2 (Disposition): -0.43

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.571
de0.562
B0.546
மகி0.517
happy0.493
Suppressed:
[]-0.582
😖-0.519
worse-0.509
😫-0.498
😣-0.494