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’d driven for hours, escaping the city’s relentless hum. Now, here he was. Liam’s shoulders slumped as he stepped from the car. The world was a blurry watercolor.

The maple tree, his father’s gift to him on his tenth birthday, was simply *not there*. A gaping void. He felt a sudden, desperate urge to sit down, right there on the grass.

His breath hitched. He closed his eyes, squeezing them shut, the phantom ache in his chest tightening with a new kind of pain. He couldn't quite see straight. He was shaking, and the trembling kept growing. He had to sit.

Emotion: overwhelmed

Cluster: Fear / Anxiety
PC1 (Valence): -2.52 Negative
PC2 (Disposition): -0.94

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.385
😫0.332
😖0.319
😣0.319
😩0.314
Suppressed:
la-0.576
de-0.496
a-0.297
happy-0.290
l-0.264