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).

She walked to the podium, each step deliberate, wanting to savor the moment. The room hummed, a comfortable energy that embraced her like a warm hug. Across the room, her father winked. A lump formed in her throat.

The years flew past in her mind - scraped knees, bedtime stories, and whispered secrets. He’d always been a safe harbor, a steady presence in a chaotic world. She found herself smiling, remembering a thousand small acts of devotion.

"Dad, I know you never wanted a fuss," she began, her voice soft, "But you deserve every bit of this. You gave us everything."

Her mother caught her eye. She had the same kind of look, she realised. A look of gratitude, of tenderness. It made her feel like the world was alright.

"You taught me the value of hard work, the importance of integrity, and that a good laugh can cure just about anything." She paused, then added, "You're the best."

Her hand brushed her cheek, just to be sure. She could feel a strange sensation in the pit of her stomach, a bubbling, fizzing excitement.

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