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 flickering airport monitor seemed to mock Eleanor. It displayed the dreaded "Delayed" status next to her flight to New Orleans. She slumped onto a cold, plastic chair, the weight of disappointment settling in her stomach like a stone. Her grandmother's 90th birthday party, a milestone she'd promised to attend, was now slipping away.

She traced circles on the condensation of her water bottle, a rhythmic gesture that offered no solace. The celebratory chatter of other passengers grated on her ears. Every joyful laugh felt like a tiny, sharp jab.

"This is ridiculous," she muttered, the words barely audible. A knot formed in her throat, tightening with each passing minute. The thought of her grandmother, surrounded by family, while she was stuck in this purgatory of a terminal, left her feeling utterly hollow.

Emotion: sad

Cluster: Sadness / Despair
PC1 (Valence): -2.52 Negative
PC2 (Disposition): -1.50

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.436
😔0.405
S0.384
无法0.341
L0.337
Suppressed:
de-0.749
la-0.614
l-0.471
/-0.350
!-0.322