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 light filtered through the blinds, casting stripes across the classroom floor. Ms. Rodriguez sat at her desk, organizing her lesson plans. Her fingers moved with a practiced ease, and the quiet rhythm of her actions soothed her. She liked the quiet order of things.

A small, intricately woven basket sat on her desk. She didn't remember receiving it. Inside was a collection of dried herbs and spices, the air filled with an intoxicating aroma. A small note read, "From a grateful student, Maria."

She closed her eyes, breathing in the scent, remembering Maria, who had always been a vibrant, creative student. A feeling of profound peace settled over her. The simplicity of the gift resonated with her.

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