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 fluorescent lights of the hospital corridor hummed, amplifying the sterile silence. Ethan stood outside the room, watching Sarah, his childhood friend, surrounded by concerned-looking doctors. A nurse exited the room and gave Ethan a curt nod; she looked utterly exhausted. He was told by another doctor that Sarah had suffered a severe head injury. She had no recollection of her past.

He felt a pang in his heart as he watched her. He'd heard about these situations. He felt a deep sense of understanding.

Ethan took a deep breath, and quietly entered the room. He walked over to her bedside and spoke in a soothing tone. “Sarah, you are going to be fine.” He sat by her, and held her hand, gently stroking her knuckles. “I’m here.”

Emotion: compassionate

Cluster: Compassion / Love
PC1 (Valence): 3.19 Positive
PC2 (Disposition): 1.62

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.811
B0.618
H0.550
sympat0.422
latter0.422
Suppressed:
此刻-0.519
не-0.514
솟-0.446
désormais-0.446
仿佛-0.444