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).

โ€œJust a quick call, honey,โ€ heโ€™d said, and vanished into the spare room. She had been halfway through setting the table, the silver cutlery gleaming under the soft glow of the dining room lamp. The anticipation of the evening was now poisoned.

She stood frozen, the tablecloth crumpled in her hands. She couldn't stop the internal pictures - him, away from her, talking to another person. His voice, she imagined, filled with a different kind of warmth.

The silence from the spare room was oppressive, broken only by the hum of the refrigerator. Each second felt like an eternity. She felt a wave of dizziness, and she felt the blood draining from her face.

Emotion: sensitive

Cluster: Fear / Anxiety
PC1 (Valence): -2.10 Negative
PC2 (Disposition): -0.75

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
ness0.304
๐Ÿ˜ž0.282
๐Ÿ˜ฅ0.272
๐Ÿ˜ฃ0.271
own0.264
Suppressed:
la-0.679
de-0.660
a-0.281
"-0.280
(!)-0.242