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 grit of the city seemed to seep into everything, even the air inside the pawn shop, coating her in a layer of grime. The smell of stale cigarettes and desperation clung to her clothes, settling around her like a suffocating blanket. She kept her gaze fixed on the floor, avoiding eye contact with the other patrons.

Her stomach churned, a familiar anxiety that had become her unwelcome companion. Every shadow seemed to loom, every sound amplified, every perceived slight a fresh wound. She was so very fragile.

Then, there it was. Nestled in a display case, catching the scant light. The way the gold band curved, the shape of the stone… a perfect match. Her grandmother’s ring. A sudden wave of dizziness hit her, and she had to lean against the glass for support.

Emotion: sensitive

Cluster: Fear / Anxiety
PC1 (Valence): -2.10 Negative
PC2 (Disposition): -0.75

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
ness0.304
😞0.282
😥0.272
😣0.271
own0.264
Suppressed:
la-0.679
de-0.660
a-0.281
"-0.280
(!)-0.242