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).

Lila sat on the park bench, watching her daughter, Olivia, chase pigeons. The gentle murmur of conversations, the laughter of children, the rustling of leaves in the trees: all combined to create a soothing sound. She was thankful for the brief respite from her busy schedule.

A tap on her shoulder. It was Mrs. Miller, another parent. "Guess what I found out? Mrs. Jenkins, Olivia's teacher, is the same teacher you had!"

Lila felt her heart open with a light sense of happiness. Mrs. Jenkins. The thought of that memory was a comfort to her. She watched Olivia, feeling a sense of connection. A smile touched her lips.

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