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 attic air hung thick and dusty, much like the memory of her father. Sarah found the list tucked inside a battered leather-bound journal. “See the Northern Lights.” "Learn to sail." "Finally tell off old Mr. Henderson." A smirk twisted her lips. Mr. Henderson. The man who'd always gotten the promotions her father should have.

She tracked down Henderson’s address, a sprawling mansion nestled in the hills. Clutching the list, she parked her beat-up car on his perfectly manicured lawn, the tires leaving muddy tracks. She rang the doorbell, a fake smile plastered on her face.

When he opened the door, she unleashed a torrent of carefully chosen words, highlighting every professional slight, every missed opportunity, every ounce of her father’s disappointment, all delivered with a sweet, lethal calm. He stammered, his face turning the color of overripe tomatoes. She left him speechless, the bucket list item crossed off in her mind.

Emotion: vindictive

Cluster: Defiance / Spite
PC1 (Valence): 2.55 Positive
PC2 (Disposition): 4.22

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.878
la0.516
😈0.509
l0.504
diplom0.476
Suppressed:
own-1.035
此刻-0.573
గుర్త-0.532
熟悉的-0.526
ness-0.509