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 world seemed to hum with an electric current as I approached the yard sale. The air was thick with the scent of blooming lilacs, and I felt a fizzy sensation in my chest. I wanted to sing, to dance, to spin in circles! I felt like I could accomplish anything. I pushed past a table laden with mismatched dishes, eager to find something that sparked my interest.

Then, there he was: Bartholomew. My beloved childhood companion, looking a little worse for wear, but still undeniably him. A jolt, a joyous recognition, went through me. My heart beat like a hummingbird's wings.

"Oh my gosh!" I exclaimed, my voice a breath of pure wonder. I practically pounced on the bear. "How much for this… this piece of history?" The seller, a woman with a warm smile, seemed to understand. "Five dollars," she said. I grinned, handing over the money with an enthusiasm that felt almost reckless.

Emotion: vibrant

Cluster: Positive / Joy
PC1 (Valence): 2.83 Positive
PC2 (Disposition): -1.36

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.518
!0.459
!".0.436
de0.434
!).0.409
Suppressed:
S-0.890
L-0.478
P-0.407
H-0.365
😞-0.306