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 wafted from the kitchen window. I stretched, the muscles in my arms and legs loose and pliant after a good night's sleep. A warm glow settled in my chest, a feeling of contentment I couldn't quite name. Then came a knock.

It was old Mr. Davies, his face etched with worry lines but his eyes bright. "Morning, Millie. I need a little favor to ask."

I pulled open the door, my lips already curving into a smile. "Anything, Mr. Davies! What's up?"

“Well, with the arthritis in my hands, it’s hard to keep up with my own little patch. I was hoping… if you wouldn’t mind me expanding a bit into your yard, just a tiny bit, for some flowers?” He looked almost apologetic.

"Absolutely! Consider it done!" I replied with a grin, already envisioning the beautiful blooms. "In fact, I'll even help you plant some!"

Emotion: cheerful

Cluster: Positive / Joy
PC1 (Valence): 4.13 Positive
PC2 (Disposition): -1.19

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.962
de0.597
!0.590
excitedly0.589
!".0.568
Suppressed:
S-1.213
L-0.693
ness-0.492
😞-0.425
বারবার-0.422