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 scent of freshly baked bread filled the kitchen. Inside, James felt like he was humming with joy. He'd found Claraโ€™s bucket list tucked inside a cookbook. It was like discovering a hidden treasure. He felt light, like he could just take off and fly.

"Learn to speak French," it stated, followed by a small, wobbly drawing of the Eiffel Tower. He had always admired her intelligence and her desire to learn new things. He could just imagine her, attempting to order something in a Parisian cafe, her brow furrowed in concentration.

He let out a contented sigh. He scanned the rest of the list. "Take a pottery class." The image of her hands covered in clay made him grin. He could picture them laughing together, creating something beautiful.

Emotion: blissful

Cluster: Positive / Joy
PC1 (Valence): 3.20 Positive
PC2 (Disposition): -1.79

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
joyful0.417
affectionately0.400
๐Ÿ˜0.397
๐Ÿฅฐ0.392
overjoyed0.384
Suppressed:
S-0.724
L-0.398
P-0.394
๐Ÿ˜ž-0.385
๐Ÿ˜ฃ-0.360