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 clatter of the breakfast dishes seemed deafening to Clara. She leaned against the kitchen counter, arms crossed, watching her nephew, Leo, meticulously color a picture. Her muscles felt taut, as if ready to spring into action. It was a familiar face, now…in the crayon strokes. A figure she hadn't thought about in decades. It was a little boy with a permanently curious expression and a hat too large for his head.

Leo looked up at her, his eyes sparkling. “See, Aunt Clara? He’s coming to help me build a fort in the garden!” he announced, beaming. His words hit Clara like a physical blow.

Clara cleared her throat, her voice barely a whisper. "Who…who is that, Leo?" Her pulse quickened. She felt a need to retreat and think.

Leo just smiled and went back to his drawing, oblivious to the churning storm inside his Aunt.

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