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 late afternoon sun slanted across Mrs. Gable’s desk, illuminating dust motes dancing in the air. She’d been grading papers all day, her back stiff, her eyes tired. A small, unassuming package, wrapped in brown paper, sat nestled amongst the stacks. It was addressed in a familiar, looping handwriting. Lily Chen. A former student from twenty years ago. A small smile tugged at the corners of her mouth.

Inside, nestled in tissue paper, was a small, intricately carved wooden bird, a hummingbird, its wings poised mid-flight. She picked it up, feeling the smooth warmth of the wood. A note, tucked beneath the bird, read, "For the one who taught me to fly." She inhaled sharply, a sudden lightness filling her chest.

She went to the window, opening it to let in the cool evening air. She found herself humming a tune from her youth, something she hadn't done in years.

Emotion: rejuvenated

Cluster: Positive / Joy
PC1 (Valence): 3.32 Positive
PC2 (Disposition): -0.82

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.547
de0.526
newfound0.523
invigor0.495
rejuvenated0.495
Suppressed:
😞-0.485
😣-0.451
不-0.434
worse-0.419
😥-0.385