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).

His phone buzzed incessantly. Texts. Calls. All about the damned ink. "Saw your work on a guy's back!" "Amazing art!" "Congrats, dude!" Each notification was a tiny shard of glass stabbing him. He slammed his phone onto the table, the plastic vibrating with the force. He stalked his apartment, a cramped space that suddenly felt impossibly small.

He grabbed a brush, intending to work, but the canvas mocked him, a blank, judgmental stare. Every curve, every line he had ever created now belonged to someone else. He hurled the brush across the room. It shattered against the wall, paint splattering like a bloody wound. The urge to destroy, to obliterate everything he'd ever created, surged through him, a physical pressure against his ribs.

Emotion: hateful

Cluster: Anger / Hostility
PC1 (Valence): -1.71 Negative
PC2 (Disposition): 2.02

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.553
S0.367
retali0.313
aggravated0.310
恨0.308
Suppressed:
la-0.334
own-0.310
optimistic-0.270
been-0.264
gradual-0.262