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).

"Oh, this is brilliant!" Amelia exclaimed, pacing back and forth in the living room. Her fingers itched to do something, anything. The scent of fresh-baked cookies filled the air, a testament to her high spirits. She reread the email, the words practically glowing on the screen. Mrs. Peterson, the very same!

Her son, Finn, wandered in, a confused frown on his face. "What's so great, Mom?"

Amelia scooped him into a hug, squeezing him tight. "Your teacher! It's Mrs. Peterson! The kindest, most patient teacher in the entire world. You're going to have such a grand time!"

Finn, momentarily stunned, managed a shy smile. Maybe this year wouldn't be so bad after all.

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