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 email from Professor Davies had a subject line that read "Important Announcement." Amelia, halfway through her lukewarm instant coffee, nearly choked. Her fingers fumbled as she clicked it open. She reread the first sentence, then the second. Her jaw went slack. The mug slipped from her grasp, the ceramic clinking against the hardwood floor. It didn't break, but the sound was loud enough to bring her cat, Mittens, bolting from the other room. "No, no, no," she muttered under her breath, a tremor in her voice.

She felt a dizzying sensation, like the elevator was dropping out from under her feet. Her research, the countless hours spent in the lab, the future she had envisioned... it all felt suddenly unsteady. She knew she'd need to have a conversation, a serious one. But she couldn't summon the energy to even pick up the scattered coffee grounds.

Emotion: surprised

Cluster: Surprise / Confusion
PC1 (Valence): -0.58 Negative
PC2 (Disposition): -0.18

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
que0.495
de0.447
la0.431
l0.323
😳0.296
Suppressed:
每天-0.269
cheerfully-0.249
ing-0.238
alleviate-0.224
gladly-0.219