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).

"Is that...is that *your* essay?" The voice cut through Emily's quiet concentration in the university café. She froze, a half-eaten scone suspended in mid-air. Her cheeks burned.

The other student, a girl she barely knew from a linguistics class, pointed at the laptop screen. Emily nodded, her throat suddenly dry.

"Wow," the girl continued, her tone ambiguous. "It's... intense."

Emily couldn't meet her eyes. She picked at the flaky pastry, her stomach churning. The essay was about her dysfunctional family, the pain she’d buried for years. Now, it was public property.

She felt like everyone in the room could suddenly see her broken pieces.

Emily mumbled an excuse and fled the café, the warmth of the tea instantly chilling in the cool, early spring air.

Emotion: vulnerable

Cluster: Fear / Anxiety
PC1 (Valence): -2.07 Negative
PC2 (Disposition): -0.69

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.327
ness0.319
😞0.299
😔0.277
😣0.258
Suppressed:
de-0.567
la-0.488
l-0.271
(!)-0.267
(!)-0.257