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. Maria, her hands dusted with flour, beamed at the customers lined up before her. Each loaf was a labor of love, a testament to her passion. A small package, addressed to “Mr. Silas Blackwood, Previous Resident,” sat on a shelf next to the oven. She knew it needed to be moved. It wasn't today's priority. The warm, inviting atmosphere of her bakery was her haven.

A young girl, about seven, was staring at the assortment of pastries in the glass case.

"Which one would you like?" Maria asked, her voice kind.

The girl pointed to a chocolate croissant. Maria smiled and gently placed the pastry in a small bag. She felt a lightness within her, a connection to her customers. She made an offer to make it a free one to the girl.

"Have a lovely day," Maria said as the girl and her mother walked out. She turned to the next customer in line.

Emotion: fulfilled

Cluster: Positive / Joy
PC1 (Valence): 3.22 Positive
PC2 (Disposition): -1.74

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.819
மகி0.415
happy0.397
grateful0.389
de0.385
Suppressed:
S-0.599
worse-0.465
C-0.460
either-0.436
😣-0.426