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 attic air hung thick and dusty, catching in Amelia's throat with every breath. She'd told herself she was just organizing, a way to occupy the time since… well, since everything. Her fingers, though, trembled as she lifted the heavy cardboard box. It was marked in her mother’s neat, familiar script: "Amelia - School."

A coldness crept up her spine, even in the summer heat. Inside, nestled amongst yellowed construction paper and forgotten report cards, lay a complete record of her childhood. Every single assignment, from first-grade handwriting practice to her college thesis. Amelia's stomach churned.

She pulled out a faded book report, her name scrawled across the top in a childish hand. Suddenly, the past felt close, suffocating. She felt a need to break something. To scream. To run. She dropped the report back in the box and slammed the lid shut.

Emotion: unsettled

Cluster: Fear / Anxiety
PC1 (Valence): -2.33 Negative
PC2 (Disposition): -0.65

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.382
😞0.309
ness0.301
😣0.256
😔0.248
Suppressed:
de-0.721
la-0.597
a-0.365
"-0.299
l-0.274