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).

Michael practically bounced on his heels as he approached the old cedar chest in the spare room. His mother had always been a keeper of memories, but he never realized *how* dedicated. The sun poured in, warming his face. He’d made a pot of fresh coffee before starting.

He lifted the heavy lid. Inside, every test, every drawing, every book report was carefully filed away in labeled folders. He felt a thrill of discovery, like he was about to unearth a hidden part of himself.

"Wow," he breathed, pulling out a particularly colorful art project. He remembered making it! It was a painting of a robot, done in crayon. A deep chuckle escaped him.

Emotion: optimistic

Cluster: Positive / Joy
PC1 (Valence): 4.44 Positive
PC2 (Disposition): -0.49

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.856
la0.688
!0.601
enthusi0.530
🀩0.515
Suppressed:
S-0.891
😞-0.504
L-0.486
ness-0.478
γγ‚Œγ§γ‚‚-0.432