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 aroma of freshly baked bread filled the small bakery, a comforting scent that mingled with the sweet fragrance of victory. Her new venture, her dream, was finally a reality. The first day's sales had exceeded all expectations. She danced a little jig behind the counter, humming a joyful tune. She typed a celebratory text to her sister, her fingers practically vibrating with excitement. "Best day ever! The bakery is a hit. Couldn't have done it without your support. Drinks on me tonight!" She grinned, hit send. An instant reply: ā€œI’m sorry, but I do not know you. Good luck with your bakery.ā€ Her smile faltered.

Emotion: triumphant

Cluster: Positive / Joy
PC1 (Valence): 3.54 Positive
PC2 (Disposition): 0.19

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.734
la0.571
triumphant0.562
šŸŽ‰0.531
ecstatic0.530
Suppressed:
šŸ˜ž-0.503
S-0.483
😣-0.482
😰-0.457
worse-0.444