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).

He traced the smooth wood of the desk, the polished surface reflecting the soft glow of the setting sun. The air in the study smelled of old books and pipe tobacco, a scent that always soothed him. His sister, Clara, inherited their father's stock portfolio, a mountain of numbers and obligations. He’d received the family's collection of rare maps, a treasure trove he knew he would pore over for years to come.

“Magnificent, isn’t it?” he murmured, more to himself than to the ghost of his father. He’d just spent a day identifying the exact locations of lost cities. The work, always the work, brought a unique feeling of peace. It was a sensation of being perfectly content. He settled into his chair, a warm cup of Earl Grey in hand, ready for the evening.

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