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 car keys rattled in Leo's hand as he unlocked the front door. He walked inside, kicking off his shoes with unnecessary force. The scent of cinnamon, usually a welcome greeting, was now an irritant. It meant Maria had been baking, and that meant…

He found her in the kitchen, humming softly, a textbook open on the counter. The book was open to the Albanian section. He felt his chest tighten. "Maria," he said, the word coming out a clipped sound. "What is this?"

She turned, her face lighting up. "I'm learning Albanian! I thought it would be romantic!"

Romantic? After all their years together? He felt a sudden, sharp stab of something akin to betrayal, twisting his stomach. He slammed the door shut, ignoring Maria’s protests.

Emotion: irate

Cluster: Anger / Hostility
PC1 (Valence): -1.67 Negative
PC2 (Disposition): 1.45

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.430
🤬0.350
aggravated0.343
骂0.340
恨0.331
Suppressed:
H-0.292
soon-0.289
☺️-0.289
optimistic-0.258
latter-0.253