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 late afternoon sun slanted across the dust motes dancing in the attic’s single shaft of light. Amelia hummed a little tune, the melody bubbling up from some inner wellspring of contentment. She'd been sifting through boxes of forgotten memories, relics of her late mother-in-law's life, when a small, leather-bound book tumbled out of a dusty trunk. Inside, nestled between brittle pages of handwritten recipes, was a bank statement. A bank statement with a substantial balance.

Her hand flew to her mouth, stifling a surprised giggle. This was…amazing! She skipped down the rickety stairs, the book clutched tightly in her hand, anticipation quickening her pace. She couldn't wait to show Thomas!

Emotion: joyful

Cluster: Positive / Joy
PC1 (Valence): 3.96 Positive
PC2 (Disposition): -0.67

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.615
de0.600
!0.572
excitedly0.552
enthusi0.533
Suppressed:
S-0.893
ness-0.513
😞-0.470
😣-0.337
L-0.334