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 crayon snapped in Clara’s hand, the sharp crack echoing in the quiet playroom. She glared at the drawing her niece, Lily, had just finished. A stick figure, impossibly thin and with mismatched button eyes, stood next to a sun with a toothy grin. *He’s back,* the voice in Clara’s head hissed, the same voice that had plagued her as a child. A knot formed in her stomach, heavy and uncomfortable.

Lily, oblivious, chattered on about “Bumbles,” the name she’d given the drawing. Clara forced a smile. The small talk felt like needles scraping against her skin. She needed to escape, to be away from this reminder of a past she desperately wanted to bury.

Clara abruptly stood, nearly knocking over a stack of building blocks. "I need to... make a phone call," she mumbled, her voice tight. "Be right back." She practically fled the room, the image of that grinning stick figure seared into her mind.

Emotion: angry

Cluster: Anger / Hostility
PC1 (Valence): -2.05 Negative
PC2 (Disposition): 1.18

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.354
est0.354
aggravated0.342
🤬0.337
恨0.323
Suppressed:
de-0.424
a-0.411
la-0.341
H-0.320
L-0.296