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 aroma of cinnamon and apples hit Maya the instant she unlocked the front door. A genuine smile stretched across her face, pulling at the corners of her eyes. Her mother’s voice, a warm rumble, echoed from the kitchen. "Maya, sweetie, you're back! Dinner's almost ready."

She dropped her bags with a thump and practically skipped into the kitchen, a spring in her step. The familiar sight of her mother humming over a bubbling pot filled her with a sense of peace she hadn't realized she was craving. She leaned against the counter, a feeling of deep contentment blooming in her chest as she watched. "Smells amazing, Mom."

“Just like old times, eh?” Her mother winked. Maya grinned, picking up a stray apple slice and popping it into her mouth. She could almost feel the stress of her city life melting away.

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