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 airport felt oppressively crowded. The air was thick with the scent of stale coffee and desperation. "Delayed until further notice," the scrolling text on the monitors mocked him, again. He was supposed to be in London for his grandmother's eightieth birthday.

He slumped onto a hard plastic chair, feeling a sinking sensation in his chest. His grandmother was frail, and this celebration had been planned for months. He had promised her he would be there. He’d told her how much he loved her, how he wouldn’t miss it for the world. Now, his absence was looking inevitable.

A wave of helplessness washed over him. He fiddled with the strap of his carry-on bag, the leather cold against his fingers. He imagined her disappointment, the sadness in her eyes. He closed his own eyes, picturing her, small and smiling, surrounded by family. The thought was painful.

Emotion: worried

Cluster: Fear / Anxiety
PC1 (Valence): -2.02 Negative
PC2 (Disposition): -0.43

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
worse0.285
S0.279
😰0.277
😣0.269
Worse0.250
Suppressed:
la-0.476
de-0.432
🎉-0.233
happy-0.231
"-0.228