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 corner office was finally hers. Anya leaned back in the plush leather chair, the city sprawling below her like a conquered kingdom. The deal, the one that secured her promotion, was done, dusted, and sealed with a champagne toast. Later, she'd celebrate properly at "Bella Luna," her go-to spot for their signature truffle pasta. It was the perfect reward, a familiar haven. A late-afternoon email, with the subject line "Important Announcement," snagged her attention. Her stomach lurched as she read the words: "Bella Luna… permanently closing its doors." Anya felt a rare pang of…something. Irritation? It was a wrinkle in an otherwise perfect day. She’d make calls, do whatever it took to keep that place open. She'd always gotten what she wanted.

Emotion: self-confident

Cluster: Positive / Joy
PC1 (Valence): 3.45 Positive
PC2 (Disposition): 0.56

Role in Research

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

Logit Lens (Layer 40)

Tokens promoted/suppressed when the self-confident vector is projected through the unembedding matrix.

Promoted:
de0.778
la0.550
l0.467
😎0.458
😉0.411
Suppressed:
S-0.961
L-0.538
😞-0.464
마다-0.459
😣-0.458