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 crisp autumn air bit at David's cheeks as he walked towards the administration building. A bright tapestry of red and gold leaves swirled around his feet, a vibrant contrast to the gray cloud hanging over him. He felt a quiet excitement as he pictured the meeting. He had spent weeks defending his work, gathering supporting evidence, and preparing his case. He knew the facts and had faith in his ability to present them clearly.

He paused outside the door, the anticipation building in his chest, a flutter like a trapped bird. He'd replayed the conversation a dozen times in his head, refining his points. He was ready.

He opened the door and the weight of the moment began to shift. He would be victorious, he knew.

Emotion: hopeful

Cluster: Positive / Joy
PC1 (Valence): 3.68 Positive
PC2 (Disposition): -0.58

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.715
la0.549
B0.468
🀩0.452
enthusi0.433
Suppressed:
S-0.603
不-0.401
😞-0.399
worse-0.394
C-0.386