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 afternoon sun poured through the kitchen window, bathing the room in a golden glow as David, hands covered in flour, carefully flipped pancakes. He'd been puttering around all morning, a balm for his soul. The house, usually bustling with family, was quiet today. A welcomed solitude.

He found a photo tucked inside a cookbook, a picture of his father, the stoic patriarch, laughing at a lively party. The women in the picture were not the family's known relatives. David blinked, a sudden prickle behind his eyes. He set the pancake in the air, but didn't bother to catch it.

“Interesting,” he said, his voice level. He poured himself a glass of orange juice, the familiar ritual grounding him. The scent of sizzling butter and maple syrup filled the air, a reminder of the day's simplicity.

Emotion: safe

Cluster: Calm / Serenity
PC1 (Valence): 1.86 Positive
PC2 (Disposition): -2.55

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
soon0.402
affectionately0.357
disfrut0.329
необходи0.325
🥰0.317
Suppressed:
S-0.474
C-0.416
L-0.408
worse-0.379
even-0.370