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).

He was halfway through his shift at the campus coffee shop when the announcement was made public. He saw the looks on people's faces as they started to read the news on their phones. Liam dropped the mug he was washing, the ceramic shattering on the floor.

He felt a deep tremor inside, a physical manifestation of his fear. He felt the blood drain from his face, and he gripped the counter to steady himself. The easy banter with customers, the familiar rhythm of his day, felt suddenly strange.

He blinked, unable to process what was happening. The future he had been building, the future he envisioned, felt like it was crumbling around him.

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