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 lavender wafted from the diffuser as Amelia hummed, meticulously arranging the dried flower sprigs in a vase. Sunlight streamed through the window, painting stripes across the polished wooden floor. Her phone buzzed, a notification from a dating app, but she ignored it, content with the gentle click of the scissors and the soft rustle of petals. She took a deep breath, savoring the quiet of her apartment before her date arrived. Tonight, she was making her famous rosemary-infused pasta, a recipe she'd perfected over years.

Later, the doorbell chimed. She smoothed down her dress, took a final look in the mirror, and opened the door. Standing there was a woman, her face a mix of shock and confusion. "Oh… you're… you're Amelia," the woman stammered, holding a bouquet of the same lavender Amelia had just used. Amelia’s brow furrowed. "Yes? And you are…?"

Emotion: calm

Cluster: Calm / Serenity
PC1 (Valence): 2.72 Positive
PC2 (Disposition): -1.82

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
मुस्कुरा0.415
enjoying0.405
leisurely0.388
:)0.382
কিছুক্ষণ0.382
Suppressed:
S-0.527
L-0.499
worse-0.454
even-0.422
😣-0.420