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).

Rain lashed against the windowpane, mirroring the way tears pricked at Maya's eyes. She hunched over her textbook, the words blurring into an indecipherable mess. Math was already a struggle, but now, knowing Mr. Evans, her usually jovial algebra teacher, had been silently observing her Instagram feed felt like a lead weight in her chest.

She’d spent the last hour meticulously deleting posts, cringing at every picture, every caption, every ill-considered story. A nervous tremor ran through her as she imagined him scrolling, judging. The thought of his face, usually so encouraging, now tinged with… something else, made her stomach churn.

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