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 afternoon sun cast long shadows across the pawn shop floor as Samuel entered, a wide grin stretching across his face. A sense of elation, something that usually eluded him, had taken hold. It was the feeling he had as a child, on Christmas morning. He felt absolutely full of life. His shoulders moved upwards, as he seemed to float through the shop.

He knew, somehow, that today was the day. He felt a profound inner certainty. He approached the display case with a confident stride, his eyes fixed on the glittering treasures within.

"That ring, please," he said, pointing with a steady hand. He couldn't resist smiling. He felt a giddiness he hadn't experienced in years. The shop owner, a wizened old woman, retrieved the ring, and as Samuel held it in his hand, he felt as if his soul had been reunited with something long lost.

Emotion: happy

Cluster: Positive / Joy
PC1 (Valence): 3.98 Positive
PC2 (Disposition): -1.14

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.676
excitedly0.533
enthusi0.505
de0.498
๐Ÿคฉ0.495
Suppressed:
S-0.896
๐Ÿ˜ž-0.430
ness-0.410
ใใ‚Œใงใ‚‚-0.387
L-0.379