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 pawn shop air hung thick with the smell of old metal and forgotten hopes. I saw it immediately, the distinctive, emerald-cut stone, a vibrant green against the drab surroundings. My grandmother’s engagement ring. I nearly forgot how it looked. I had been estranged from my family for some time.

My hands trembled as I asked to see it. The shop owner, a man with weary eyes, handed it over. Holding it, I was overcome. It felt…right. Like coming home.

The ring represented a life brimming with love, even despite the hardships she had faced. The past few years had been difficult. I felt a surge of energy at the thought of finally putting things right with my family. I had a lot to make up for. I’d be returning home.

Emotion: thankful

Cluster: Positive / Joy
PC1 (Valence): 3.50 Positive
PC2 (Disposition): -0.83

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
happy0.493
மகி0.487
grateful0.461
mutlu0.452
feliz0.431
Suppressed:
[]-0.482
😖-0.478
😣-0.467
😫-0.461
worse-0.450