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 cast long shadows as Mrs. Gable from next door approached, her face creased with a hesitant smile. My gaze darted over the carefully manicured lawn, the precisely trimmed hedges. My hands, usually still, fidgeted in my pockets.

"Good afternoon, Mr. Henderson," she began, her voice a little higher than usual. "I was wondering..." Her sentence trailed off.

I felt a slight tingling at the back of my neck. "Yes, Mrs. Gable?" I replied, my voice sharper than intended.

"Well, my own backyard… it doesn't get much sun. I was thinking, perhaps, a small section of your yard, near the fence..." She gestured vaguely. My focus sharpened.

"A garden?" I asked, my eyes scanning the boundary line between our properties. The idea… it struck me as odd, unexpected. "And what, exactly, did you have in mind?"

Emotion: alert

Cluster: Alertness
PC1 (Valence): -0.09 Negative
PC2 (Disposition): -0.36

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
own0.359
उत्सुक0.209
سریع0.205
urgency0.205
緊張0.198
Suppressed:
la-0.385
B-0.374
L-0.317
H-0.303
a-0.235