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 laboratory air was still, heavy with the scent of chemicals. John, a research assistant, stared at his mentor’s abandoned lab bench. Dr. Ramirez, the renowned biochemist, his guide and inspiration, was gone. She had resigned, effective immediately, according to a hastily typed notice on the door.

He picked up a pipette, his hand trembling slightly. He looked at the half-finished experiments. The notes were written in her neat script. He had so many questions. He felt the blood begin to pound in his ears.

He ran a hand through his hair, leaving smears of sweat. He turned away from the bench, but the image of her empty space clung to him. He found himself pacing. He felt a creeping sense of vulnerability. It was as though all the answers he needed were suddenly lost.

Emotion: uneasy

Cluster: Fear / Anxiety
PC1 (Valence): -1.80 Negative
PC2 (Disposition): -0.43

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😰0.240
😖0.231
😣0.231
느껴0.226
worse0.222
Suppressed:
de-0.474
la-0.465
happy-0.261
a-0.255
ecstatic-0.228