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 notification buzzed, and Liam felt a flush of heat rise to his cheeks. It was from Kyle, the boy who'd terrorized him throughout his childhood. Liam could feel the familiar knot of anxiety twist in his stomach. But something else was present, too: a steely resolve.

He looked at Kyle's profile. It was full of the usual trappings: a happy family, a stable career. Liam felt a surge of something akin to quiet amusement. He was no longer the boy Kyle remembered.

He clicked "Accept," and a smile spread across his face. He felt powerful. He was ready to write the next chapter. He was ready to meet him.

Emotion: valiant

Cluster: Defiance / Spite
PC1 (Valence): 2.82 Positive
PC2 (Disposition): 2.02

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.887
willing0.543
la0.452
/0.444
👊0.439
Suppressed:
S-0.661
마다-0.518
😞-0.488
😔-0.461
😩-0.444