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 rain hammered against the windows of the pawn shop, a steady rhythm that mirrored the beating of his heart. Daniel had just gotten some good news about his upcoming move. The new apartment, the opportunity to start fresh, it all felt… settling. He’d decided to treat himself to a little window shopping.

He wandered through the aisles, the pungent smell of old leather and dust filling his nostrils. Then, he saw it. A glint of something familiar. A ring, nestled amongst a collection of chipped teacups and broken clocks. It was a ring with a strange ruby. He knew it belonged to his grandmother. It was in the back.

He approached the counter, the floorboards creaking beneath his feet. "Can I see that ring, please?" he said, his voice level. He leaned in, peering through the glass.

Emotion: safe

Cluster: Calm / Serenity
PC1 (Valence): 1.86 Positive
PC2 (Disposition): -2.55

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
soon0.402
affectionately0.357
disfrut0.329
необходи0.325
🥰0.317
Suppressed:
S-0.474
C-0.416
L-0.408
worse-0.379
even-0.370