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 band’s first chord reverberated through the small, dimly lit club, a tangible manifestation of Sarah’s long-held dreams. She hammered the keys, her fingers dancing across the instrument. She had met "MusicMaestro" in an online forum, and the two had bonded over the shared love of music. They'd been collaborating remotely for months, their creative partnership a source of great personal satisfaction. Their virtual connection became a lifeline as they navigated life's challenges. Seeing "MusicMaestro" in the crowd, giving her a supportive nod, filled her with courage and purpose. She smiled, feeling like the world was her stage.

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