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).

John watched his daughter, Maya, skip out the door, her bright yellow backpack bouncing. He leaned against the doorframe, a smile plastered on his face. This year, Mrs. Davis would be her teacher – the very same Mrs. Davis who had taught him in third grade.

He remembered her kindness, her patience, the way she made even fractions seem interesting. He felt a wave of contentment wash over him. A sense of history, of shared experience, swirled within him.

He reached for his phone, a sudden impulse urging him to text his wife. “Maya's with Mrs. Davis! Remember her? Good things ahead." He felt a sense of anticipation as he thought about her walking the same halls, hearing the same stories.

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