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 dingy tattoo parlor reeked of antiseptic and something metallic. I sat stiffly on the vinyl chair, my heart hammering against my ribs. Seeing my father’s "Reclamation" drawing on the young man’s bicep was a shock. The man’s name was Leo. Leo was getting his first tattoo.

Leo told the artist he had lost his father a month ago. He wanted something that represented the bond they had. The piece had been a tribute, he'd told the artist. The artist then pointed at me, and said: "You know this artist?"

I swallowed hard, forcing myself to respond: "Yes. He… he was a good man." There was a knot in my throat, threatening to unravel. I missed him. It hurt.

The artist began the ink. The needle buzzed, and I found myself studying Leo's face. He was smiling softly, his eyes closed. A warm rush flooded my body, a wave of both pleasure and pain.

Emotion: sentimental

Cluster: Nostalgia / Reflection
PC1 (Valence): 0.89 Positive
PC2 (Disposition): -1.58

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
L0.677
fondly0.567
memories0.503
曾0.488
reminis0.466
Suppressed:
either-0.515
or-0.488
[]-0.473
--0.408
не-0.396