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 dim lighting of the jazz club swirled around Elena. She took a sip of her water, feeling a slow, steady pulse in her temples. She’d always admired Jake’s tales of being a famous travel photographer. A shared online article, however, had revealed he was a staff photographer for a local newspaper. Elena tried to maintain a neutral expression, but her fingers traced the rim of the glass. She wanted to approach this gently.

Jake arrived, looking charmingly disheveled. He slid into the booth opposite her. "Sorry I'm late," he said, flashing his usual smile. He didn’t seem to suspect anything. Elena watched his body language.

“No worries,” Elena replied. "What inspired that latest trip, Jake?" she asked calmly, trying to appear engaged. She needed him to be comfortable.

Emotion: patient

Cluster: Passivity
PC1 (Valence): 2.71 Positive
PC2 (Disposition): 0.40

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
rest0.467
eventually0.415
leisurely0.369
慢慢0.357
a0.355
Suppressed:
无法-0.432
even-0.428
인해-0.421
無法-0.415
😣-0.391