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).

“Well, you’ll be missed,” Daniel's colleague, Mark, said, his voice flat. Daniel didn't reply, just offered a weak smile. He felt a lump rise in his throat. He had cherished his role in the warehouse, the camaraderie with his workmates, the tangible rewards of a hard day's work. Now, he was being moved to the office, to a role that felt like a betrayal of his work ethic.

He walked past the loading docks for the last time. The smell of sawdust and exhaust fumes, usually a comfort, now felt like a painful reminder of his past. He felt a sting in his eyes.

He didn't know how he would endure the change.

Emotion: sorry

Cluster: Sadness / Despair
PC1 (Valence): -2.02 Negative
PC2 (Disposition): -0.36

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.604
L0.549
😞0.382
😔0.364
ness0.324
Suppressed:
de-0.675
la-0.536
(!)-0.323
!-0.298
l-0.288