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 aroma of freshly baked bread wafted from the Peterson’s house. Daniel, slumped on his porch steps, inhaled deeply. The smell was a cruel reminder of his own empty stomach. He’d skipped lunch again, to save money.

He watched Mr. Peterson pull out of his driveway in a gleaming, obsidian-black SUV. The man was always perfectly dressed, his tie immaculate, his face perpetually cheerful. Daniel felt a bitter taste in his mouth.

He ran a hand through his hair, matted with sweat. It wasn't fair. Why did Peterson, with all his money, have to be so happy? He felt a prickle of something unpleasant in his chest. He felt like he wanted to scream at the top of his lungs, but didn't even have the energy to do that.

Emotion: sullen

Cluster: Sadness / Despair
PC1 (Valence): -1.51 Negative
PC2 (Disposition): -0.22

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.404
😞0.282
’0.254
:(0.233
无法0.215
Suppressed:
de-0.539
!-0.276
la-0.263
🤩-0.225
que-0.216