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 aroma of freshly brewed coffee filled Ethan's tiny apartment, a welcome change from the usual stale air. Today was going to be a good day. He was finally going to ask Sarah out. He felt a fizzing anticipation as he walked towards his door.

As he closed the door, he heard a chuckle. Mr. Abernathy, his neighbour, was unloading groceries from a car that he had never seen before. It was a sleek, silver sports car. It was beautiful.

"Morning, Ethan!" Mr. Abernathy called out, his voice jovial. "Fancy a ride to the store? Save you a walk!"

Ethan accepted and jumped in. The leather seats and the state-of-the-art sound system filled his senses. As they drove, the car purring, Ethan felt the world was suddenly a much larger and better place. This day was going to be an experience.

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
๐Ÿคฉ0.515
Suppressed:
S-0.891
๐Ÿ˜ž-0.504
L-0.486
ness-0.478
ใใ‚Œใงใ‚‚-0.432