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 sun beat down mercilessly on the yard sale, making Sarah sweat beneath her floral sundress. She spotted the bear almost at once, slumped against a dusty porcelain doll. It was a faded memory, the source of countless arguments with her cousin, Emily, who'd had the audacity to call it "hers" despite Sarah having been the one who had hugged it to death.

She feigned disinterest, browsing through a box of mismatched buttons. Then, casually, she sauntered over. "How much?" she inquired, her voice calm as a summer lake.

The man running the sale squinted at the bear, then shrugged. "A dollar. It's seen better days."

Sarah paid the dollar, her fingers closing around the bear's worn paw. Walking towards her car, the sun seemed to shine just for her.

Emotion: spiteful

Cluster: Defiance / Spite
PC1 (Valence): 2.20 Positive
PC2 (Disposition): 3.95

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.717
l0.517
a0.480
la0.440
😈0.428
Suppressed:
own-1.063
此刻-0.534
గుర్త-0.493
熟悉的-0.487
ness-0.479