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 late afternoon sun warmed Amelia's face as she sat on the balcony, a cup of chamomile tea steaming gently in her hands. A gentle breeze rustled the leaves of the oak tree outside, creating a delicate whispering sound. She took a slow, deliberate sip, savoring the subtle floral notes. It was precisely this sort of moment that made the tiny apartment feel like a sanctuary.

Later, Mark came in, kicking the door closed with a frustrated thump. "I can't believe it," he grumbled, dropping his keys on the counter. "Sarah's bringing a kitten over tonight. I have a feeling we're going to be looking after it."

Amelia sighed, the contentment inside her shifting slightly. She carefully placed the cup on the small table and took a deep breath, focusing on the slow rise and fall of her chest. "That's unexpected," she said, her voice soft. "Are you sure? A pet would be a big commitment."

Emotion: peaceful

Cluster: Calm / Serenity
PC1 (Valence): 2.40 Positive
PC2 (Disposition): -2.79

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
прият0.408
disfrut0.407
gently0.390
affectionately0.379
thoughtful0.376
Suppressed:
de-0.541
S-0.493
L-0.396
'-0.388
worse-0.367