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 wind howled a terrifying song, and Eleanor huddled by the window, clutching her favorite blanket. Then, a monstrous splintering sound, and the world seemed to hold its breath. When she finally dared to look, she saw. The magnificent maple of the Jenkins family, now split and fallen, resting on their roof.

Eleanor's heart ached for them. She immediately pulled on her boots and headed outside, the rain pelting her face. The Jenkins' young son, Timmy, was standing alone, his eyes wide with fear.

She approached him and knelt down. “Hey, Timmy, are you alright?" she said softly, reaching out to brush a stray strand of hair from his face. She took his hand and gave it a squeeze. “It’s okay. We’ll get everything sorted. Why don’t you come inside with me? I have some cookies.”

Emotion: loving

Cluster: Compassion / Love
PC1 (Valence): 3.72 Positive
PC2 (Disposition): -0.41

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.604
affectionately0.532
함께0.504
🥰0.459
loving0.440
Suppressed:
не-0.484
S-0.458
worse-0.457
[]-0.438
😖-0.417