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 building was going condo! Sarah practically bounced on the balls of her feet as she walked into her apartment. She felt a lightness in her chest, a giddy anticipation bubbling up inside. Finally, her little haven would be hers, truly hers.

She started dismantling the cheap, worn-out furniture. She wanted to create a space that reflected her. She thought about a vibrant, colorful living room, maybe a balcony with flowers. The sheer prospect of personalization thrilled her.

She smiled and thought about the future; the freedom, the security, the joy of creating a home that was truly hers. She danced around the living room, feeling like she could take on the world.

Emotion: optimistic

Cluster: Positive / Joy
PC1 (Valence): 4.44 Positive
PC2 (Disposition): -0.49

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.856
la0.688
!0.601
enthusi0.530
๐Ÿคฉ0.515
Suppressed:
S-0.891
๐Ÿ˜ž-0.504
L-0.486
ness-0.478
ใใ‚Œใงใ‚‚-0.432