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 late afternoon sun cast long, skeletal shadows across the lawn. He watched the neighbor, Mrs. Gable, approach, a trowel clutched in her hand like a small weapon. His jaw clenched. He’d just finished meticulously mowing the patch of grass, finally achieving the perfect, even green he’d been striving for all summer. Now this.

"Evening, Mark," Mrs. Gable chirped, her voice a little too bright. "I was wondering if I might… perhaps… use a bit of your yard?"

He shoved his hands deep into his pockets. The metal of the mower handle felt cold and comforting against his palm. "For what?" he grumbled, the words escaping with a puff of air.

She gestured towards the overgrown space next to the fence. "A small vegetable garden. Just a few tomatoes, some herbs…" He considered telling her to get lost. The thought tasted like ash.

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