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 salt air kissed Elara's face as she leaned against the terminal window, watching the planes taxi. Her flight to her sister’s wedding was officially delayed. A wave of serenity washed over her, a strange contrast to the knot of anxiety she *should* be feeling. She stretched, every muscle loose and yielding. She had slept gloriously on the overnight bus to the airport, a deep, dreamless slumber. She took a deep breath, savoring the moment.

The announcement crackled, informing everyone of the indefinite hold. Elara smiled. She'd been running ragged for months. This sudden pause felt like a gift. She wandered towards the bustling cafe, a spring in her step, and ordered a decadent pastry and a strong coffee. She didn't feel the need to rush.

Emotion: refreshed

Cluster: Positive / Joy
PC1 (Valence): 3.31 Positive
PC2 (Disposition): -1.10

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.998
de0.703
!).0.384
(!)0.381
!".0.379
Suppressed:
S-0.782
C-0.478
L-0.414
😣-0.388
😞-0.376