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 scent of lavender and old paper filled Elara's lungs as she sat on the dusty floor of her grandmother's sunroom. She’d always known her grandmother as a proper woman, a lover of afternoon tea and perfectly manicured rose bushes. This was why she was practically rendered speechless. The list, penned in a shaky hand, included things like "Swim with sharks" and "Learn to knit a sweater."

Elara’s eyes flickered across the words, a strange mix of disbelief and something akin to awe swirling within her. Her grandmother, *swimming with sharks*? The thought was absurd.

The subsequent entries were no less shocking. "Go to a rock concert." "Get a tattoo." "Tell someone I love them before it's too late." Elara felt a pressure building in her chest, a tightening that made it difficult to breathe. She leaned back against the wall, utterly lost in a kaleidoscope of unfamiliar images.

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