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).

"Saturday?" Sarah repeated, staring at the brightly colored flyer for Leo's gaming night. Her own board game marathon, which she'd dedicated weeks to, was scheduled for that very day. A knot of irritation tightened in her stomach.

She took a deep breath, focusing on the scent of lavender from the diffuser on her desk. Leo’s excitement was contagious; she could almost hear the excited chatter. She felt her shoulders relax slightly. It was unfortunate, but not insurmountable.

She glanced at her meticulously organized schedule, a grid of tasks and appointments. She’d always loved a good challenge, and this seemed like one. Perhaps she could combine the games, a hybrid celebration. She grabbed her phone and texted Leo, a smiley face accompanying the message.

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