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).

Eliza hummed as she packed her bag, the scent of lavender from her soap filling the small room. She felt a lightness in her chest, like a balloon straining at its string. The email confirmation sat open on her laptop screen, its words still fresh: "Thank you for your application to the Crimson Tide Rowing Team..." She knew what came next.

The rejection stung, but not in the way she’d expected. She bit back a giggle. The wind whipped her hair as she jogged to her car, a spring in her step.

She wouldn’t be rowing competitively this year. But she’d already signed up for a pottery class. The thought brought another smile to her face. She practically skipped to the driver's side and started the car. The open road awaited.

Emotion: happy

Cluster: Positive / Joy
PC1 (Valence): 3.98 Positive
PC2 (Disposition): -1.14

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.676
excitedly0.533
enthusi0.505
de0.498
🤩0.495
Suppressed:
S-0.896
😞-0.430
ness-0.410
それでも-0.387
L-0.379