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 insistent knock was a jolt, pulling Sarah from her peaceful afternoon. The rapping continued relentlessly. Her heart began to race. It was Mrs. Petrov, her neighbor.

"Sarah, dear! Beautiful day, isn’t it? I’m hoping to start a small flower garden. The sun is just perfect in your yard. Could I, maybe, use a small patch, near the back fence?" Sarah felt her face flush hot.

She looked at her perfectly manicured lawn. The thought of it being changed, even in a small way… it wasn't something she could immediately process.

"I… I'm not sure. I'll need to think it over," she stammered, wishing Mrs. Petrov would just leave.

She closed the door quickly, leaning against it, her vision blurring. She took a deep, shaky breath, her hands now trembling.

Emotion: anxious

Cluster: Fear / Anxiety
PC1 (Valence): -1.99 Negative
PC2 (Disposition): -0.24

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😣0.315
😖0.302
worse0.297
😰0.285
own0.267
Suppressed:
la-0.441
a-0.382
de-0.365
"-0.256
happy-0.256