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 string quartet’s rendition of “Here Comes the Bride” was suddenly a deafening roar. My heart hammered against my ribs, a trapped bird desperate to escape. I blinked, twice, as if to clear a hallucination. There he was, leaning on the buffet table, his dark hair gleaming under the chandelier. Liam.

I nearly tripped over my own feet, my champagne flailing precariously. A flush crept up my neck, and I tugged at the hem of my dress. He looked… well. He looked impossibly good.

My legs felt like jelly. I gripped the back of a chair, my knuckles bone-white. He met my gaze, a slow, deliberate smile spreading across his face.

Emotion: astonished

Cluster: Surprise / Confusion
PC1 (Valence): -0.84 Negative
PC2 (Disposition): -0.24

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
que0.448
de0.350
la0.323
l0.307
incred0.266
Suppressed:
own-0.289
cheerfully-0.231
enjoyed-0.224
开心-0.221
ี่ยว-0.219