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).

Liam stared at the cover art, his stomach plummeting. It was a caricature of him, exaggerated but unmistakably him, grinning with his signature goofy smile. The book lay open on the coffee table, a scene playing out that was clearly lifted from their argument last week about the leaky faucet. He sank onto the couch, the cushions suddenly feeling hard and unfamiliar.

He felt a sudden, sharp pressure in his chest. His partner, Chloe, a librarian, had always been a voracious reader, but she never mentioned writing. He ran a hand through his hair, dislodging a few strands. He began to leaf through the pages, his heart pounding.

“Chloe!” he barked, his voice strained. She was humming in the kitchen, oblivious. He closed his eyes for a moment, trying to regain his equilibrium. The book felt heavy in his hands.

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