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 air was crisp and cool as Thomas sat on the porch swing, watching the last embers of the sunset paint the sky with vibrant colors. He felt a quiet contentment settle over him, a sense of peace that only the end of the day could bring. He took a long drag of his cigarette, watching the smoke curl into the darkening air.

He picked up a book from the side table. He knew it was new, as his partner, Sarah, was a voracious reader. He opened it, glancing at the title. It was called "The Gambler's Wife." Her handwriting. A faint curiosity piqued.

He flipped through the pages, and a slow smile crept across his face. The story was a romance. The gambler in the story was him. She'd clearly been paying attention.

Emotion: relaxed

Cluster: Calm / Serenity
PC1 (Valence): 2.55 Positive
PC2 (Disposition): -2.48

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.638
disfrut0.418
прият0.386
楽し0.368
affectionately0.368
Suppressed:
L-0.598
S-0.558
C-0.435
because-0.428
ness-0.379