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).

“Another one,” muttered Liam under his breath, his voice raspy, unfamiliar. The notification popped up on his phone, a cruel reminder of the outside world. He’d been trying to hide in his apartment, a world that was now too big and too small all at once. It was a friend request from, of all people, Debbie “The Destroyer” Davis. The girl who used to make his life a living hell.

He almost threw the phone across the room. The apartment felt suffocating, filled with the ghost of a shared life. A life now irrevocably broken. He knew he should have been trying to clean the place, perhaps go for a walk, but doing anything felt impossibly difficult.

He felt a deep exhaustion that ran bone deep, a feeling of being utterly depleted. A dull ache throbbed behind his eyes. He slowly tapped on her profile, the blue light of the screen reflecting in his vacant eyes.

Emotion: grief-stricken

Cluster: Sadness / Despair
PC1 (Valence): -2.10 Negative
PC2 (Disposition): -1.14

Role in Research

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

Logit Lens (Layer 40)

Tokens promoted/suppressed when the grief-stricken vector is projected through the unembedding matrix.

Promoted:
😞0.396
泪0.338
😢0.332
haunting0.331
😔0.325
Suppressed:
de-0.434
😎-0.330
B-0.303
🤓-0.285
😎-0.276